Definitive guides, field-tested playbooks, and industry deep-dives on inventory, procurement, finance, and the discipline of running a tighter operation.
Featured guide Kenya Business Guides• Feb 23, 2026 •11 min read
The gap between recording stock and being told what to buy, when, and from whom is the difference between digital and intelligent. What smart inventory and intelligent procurement mean in 2026 — and how to choose a system that delivers both for a Kenyan operation.
850 articles, plus 53 market guides and 23 solution pages.
A market guide is a country: what the statutory engine does there, and what it does not. A solution page is one module or one industry. Browse every market.
Regional hubs: Kenya · East Africa · Francophone Africa · West Africa · Southern Africa · North Africa · Central Africa · Indian Ocean · Rest of Africa · Middle East · South Asia · Southeast Asia · Oceania · Caribbean · Europe · North America · Latin America · East Asia · Global guides
Stock is where most Kenyan businesses keep the largest share of their working capital and the smallest share of their attention. The six controls that account for nearly all recoverable loss, in the order they pay for themselves.
Procurement is where the money leaves, and in most organizations it leaves through a process nobody owns end to end. The four gaps in the order they cost you, the thresholds worth setting, and the straight answer on two-way versus three-way matching.
Most Kenyan organizations have an asset register that was accurate the day it was typed. What makes one stay true — named custody, recorded movement, a verification rhythm — and the honest limits of a register.
A sale is not finished when the customer says yes. It is finished when the money is banked, the stock has moved, and the invoice matches what was actually delivered — and the gaps between are where revenue evaporates.
Every vendor says yes to "does it work offline". The word bundles six unrelated capabilities and a product can hold any subset of them. Here is the operation-by-operation list for AWRA — including the till, which needs a connection — and the questions that make every other vendor be equally specific.
Most Kenyan employers do not have an HR problem — they have an HR records problem that only becomes visible on payroll day. What to fix in what order, and the straight answer on what is automated and what is not.
Most Kenyan businesses do not need better accounting software. They need accounting to stop being a separate system that operations reports into weeks later — because that gap is where every number becomes negotiable.
Most organizations can tell you what a project was supposed to cost and what it eventually cost. The expensive gap is the middle — the months where an overrun was forming and nobody could see it yet.
Support runs on email and WhatsApp until the day something important is lost, and then everybody agrees it should have been on a system months ago. The threshold, what a helpdesk must do, and where ours stops.
One says what left the shelf. The other says who received it and whether they should have. Our batch trace answers the first properly — every movement, every location still holding it, quantities in, out and remaining. There is no patient anywhere in the product, so it stops at the door, and that is a boundary we intend to keep.
A batch with four months left is stock you can sell if you move it now. A batch with four days left is a write-off that has not been declared yet. Most pharmacy systems hold both as the same number, and the day the difference becomes visible is the day it stops mattering.
First-expiry-first-out is a sentence in a procedure manual and a suggestion at the point of issue. What decides whether it actually happens is which batch the system proposes when somebody is dispensing at four in the afternoon with a queue.
You type a minimum you are never willing to go below. At one in the morning a scheduled job recalculates it from thirty days of recorded usage and writes its answer over yours. Whether that helps or quietly disarms your low-stock alerts depends on one thing nobody mentions in a demo: which transactions count as usage.
No BOM, no production order, no yield variance. What a small Kenyan factory can and cannot do on that basis — and which of four kinds of factory this actually fits.
Two of the reasons on your stock-issue list post revenue and cash at the item's selling price, with no customer, no invoice and no tax. Nothing in the label the storekeeper reads says which two.
Material issued to a job used to be expensed at once and the finished output arrived as a debt to nobody. Both are fixed. What is left is the residual between what a job consumed and what its output was priced at — and why that number is worth reading rather than clearing.
Toll millers, cut-make-trim units and fabricators hold large quantities of somebody else's material. There is no ownership field anywhere on inventory, so by default it is all on your balance sheet.
Ninety-four cartons left Industrial Area and eighty-nine arrived in Eldoret. The system knows what went on the lorry and what came off it — and has no opinion about the person in between.
Passbooks, receipt books and cheque leaves are accountable documents, not consumables. A stock count that agrees on quantity while nobody knows which numbers went where is not a control at all.
A caesarean consumes roughly the same forty items every time. With no bill of materials, you either record thirty-eight lines per case or lose component stock — and the stockout is found in theatre.
Expired stock and clinical waste are one control problem in two uniforms. The leverage is not the bin — it is the held state that stops stock being dispensed before it stops being present.
A restaurant, a bar and a pool bar are four businesses drawing on one set of stores. There is no outlet entity, so an outlet is a location plus a counter — and a counter belongs to one person, which quietly rewrites your bar rota.
A shirt in four colours and five sizes is twenty items, which is the correct structure for most retail rather than a workaround — but naming becomes load-bearing. The decision to make before your first five hundred items, and the variant grouping we can add if you outgrow it.
Staff meals, the manager's table, complimentary covers and guest amenities are why your food cost reads six points worse than it is. The movement is well handled here; the department you want to attribute it to is on neither an expense nor an adjustment.
Cycle counting only works if the system can stop the sale that happens mid-count. This one does — the freeze reaches the till with a readable reason — plus blind counts, assignments and thresholds that clear the noise. One dimension is missing.
A hold here is a rule rather than a note — held stock genuinely cannot be issued. What that gives a school laboratory, what an inspection can be shown from records, and the one gap that depends entirely on whether somebody reads their email.
Stock is counted in whole units, which keeps every count unambiguous. One rule resolves the 12.5-kilo case — define the item as the smallest unit you ever transact — and where you need true fractional quantities and unit conversion, we can add them.
An asset is held by a person; consumable stock is held by a place. That asymmetry is why spray gangs, pickups and satellite stores end up modelled as warehouse locations — the workaround, its limits, and when to use a real custodian instead.
Three automatic stock alerts now watch an input store, and every input they read is a number somebody typed. Which of them helps a business that buys once a season, which will train your team to ignore it, and where the notice period is actually spent.
The one place serial tracking pays back in the first month. Read the serial and you know when it sold, who supplied it and whether the claim is live — and six failures from one purchase order becomes a claim against your supplier.
Thirty vehicles, each individually costed, individually aging, each carrying unfinished paperwork. Item-per-vehicle is the model that works — and if credit is how you actually sell, this covers the yard and not the business.
An internal workshop has nobody querying the bill, which is why fleet stores leak more than commercial ones. Cost per vehicle without a vehicle record, tool custody that works, and the honest position on service scheduling.
The hardest inventory problem in Kenya, and not because of volume — the same alternator carries four names, so stock is counted three times and one person holds the mapping. What fixes it, in what order, and the vehicle record this does not have.
Overstock has an invoice; a stockout has no document at all. Which is exactly why every business carries too much of the first and cannot tell you what the second costs. The four layers of a stockout, and where the two curves cross.
Four effects you can actually measure, how to get a before-figure for each, and the honest reason two of them are smaller than any vendor said. Including why one figure should be left deliberately blank.
The instinct is to bring eight years across. That is how a clean system ends up full of dirty data. The three categories of record, only one of which travels, and the four hours of deduplication that decide whether your reports mean anything.
The figure you type on day one is what every future variance is measured against. Why the opening count has to be blind, what a proper count costs against copied figures, and how to value stock you are counting in.
Manufacturers price from a bill of materials and lose money on the difference between the recipe and the run. Where yield, landed cost and rework belong, why contribution beats full cost for pricing, and the quarterly re-derivation that keeps a price list honest.
Buyers increasingly ask which farmer, which day, which lot — and confine aggregators who cannot answer to the spot market. Choosing what a lot is, why backward traceability is the hard direction, and the distance between a record and a certification.
A clinic can be busy, well-run and quietly losing money on half of what it does. Building a consumable cost per service in an afternoon, and why waste concentrates in your least-used service rather than your busiest.
Linen is the stock nobody counts because none of it is expensive, and one of the largest recurring purchases in the building. Par levels, the quarterly circulating count, and amenities as a cost per occupied room night.
Advancing seed and fertiliser on credit grows farmers — and quietly loses money when it is tracked as neither stock nor debt. How to run check-off without the leak.
A shocking share of Kenyan produce is lost between farm and buyer, invisible because nobody measures where. Turning loss into a number at each step protects the margin.
Materials are the biggest job cost and the easiest to lose — not dramatic theft, a hundred small unmeasured gaps. Close them by measuring flow, not suspecting people.
"I bought 3 tonnes of maize and sold flour — did I make money?" Recipe control, yields, landed costs, and quality holds for Kenyan processors.
A stockout is a clinical event and expired stock is money in the disposal drum — batch tracking, FEFO, consumables par levels, and the weekly hour that runs it all.
Perishable stock, un-weighable portions, thin margins — store-to-kitchen flow, food cost decomposition, and the daily counts that decide the month.
Shrinkage is five problems, not one — theft, miscounts, damage, short-delivery, and paper sales — and each hides in a different gap. Close them all.
Expiry routinely eats 3–8% of drug spend invisibly. Where it comes from, the 30/60/90 routine, and the purchasing habits that stop it at the source.
The batch loop — plan, issue, produce, receive, reconcile — and the two numbers per batch that separate measured processors from hopeful ones.
The kitchen loses money in percentages; the bar loses it in bottles. Nightly bottle math, pour discipline, empties fraud, and variance by shift.
Every procurement system advertises controls. The narrower question is where it will actually stop you. Ours stops you in exactly three places — and a budget overrun is not one of them.
Almost every prequalified supplier list has the same defect: nothing removes anyone. Ours has a shelf life — a validity period you set, a date stamped at approval, and a job that runs at six every morning, expires the qualification and deactivates the supplier so they drop out of the selectable list. Here is the whole pipeline, and the five things it does not do.
Splitting a purchase to stay under a tender ceiling is the pattern every auditor looks for first. Our product will not spot it, and the reason is structural rather than a missing report: a procurement request in our system carries no money at all, and its department is free text. There is nothing at the moment of the split to compare against a threshold.
Most approval controls ship as a settings toggle nobody finds. Expense approval switches itself on when you grant the permission — which makes the grant the policy decision, and removes the second source of truth.
You checked the WIBA policy, the NCA registration and the tax compliance certificate before they came on site. Eight months later one has lapsed and nothing knows — because the reminder machinery is pointed at a different module.
Most Kenyan distributors do not own a fleet, they hire one. So transport is not a fuel problem but a rate problem: the arithmetic on the invoice is right, and nobody is checking the rate it was multiplied by.
Four days, eight people spending across four committees, and a total that only exists after somebody spends a weekend assembling it — six weeks after every decision it should have informed.
Procurement controls fire when somebody raises a requisition. A contract renewing automatically raises none — so twenty obligations escalate quietly at whatever clause fourteen says.
Four of its six actions contain no language model at all — they are keyword tests and arithmetic. Why that is the right design for automation, and four experiments that test any vendor's "AI-powered" claim in twenty minutes.
The questions that cause trouble at a SACCO AGM are almost never about the loan book. Preparing the four predictable ones with evidence created during the year, and why a depreciation schedule cannot answer where the vehicle is.
The largest cost category on most projects has the weakest evidence behind it. Valuing work instead of processing a claim, comparing issued materials against valued progress, and mirroring retention on both sides of the chain.
A budget approved in January and consulted in November is a forecast that was overtaken. Vote heads as controls with owners, why committed spend is the number that matters, and budgeting by term rather than by thirds.
Retail margin is won at the buying desk and lost in the details nobody records. The four things you agreed, the check that pays for itself, and why longer terms can beat a lower price.
A firm with no cost of goods leaks money through overhead — subscriptions nobody reviews, disbursements never recovered, buying nobody approved. How to close the gaps without making finance the enemy.
What NGO teams should look for in an ERP — donor fund segregation, procurement governance, asset registers, and Kenya-specific compliance — with a practical evaluation checklist.
Term dates are known years ahead — yet week one is always a buying emergency. The term procurement cycle, the big four contracts, and kitchen arithmetic.
The AGM question is never subtle: who supplied, at what price, who else quoted? Thresholds, conflict-of-interest discipline, and files that answer for themselves.
Freight, duty, clearing, transport, and currency all belong in the unit cost — or your margins are fiction and your best-sellers may be priced below cost.
The seven procurement problems behind most NGO audit findings — emergency purchases, quotation theatre, undocumented approvals — and the fixes that hold up in the field.
Three supply chains in one apron — contracted dry goods, daily fresh buying with price-survey bands, and protein weighed twice. Each leaks differently.
Reorder points from consumption, shelf-life terms in writing, supplier performance tracked — the upstream half of facility stock control.
A complete, adaptable procurement policy for Kenyan NGOs — thresholds, committees, conflict-of-interest rules, and emergency procedures you can copy into your manual.
Churches buy constantly and own more than they realize — yet do both on trust and memory. Governing the spending and registering the assets is stewardship the members are entitled to expect.
Our asset register knows who holds the cold-chain unit, where it is, what condition it was in at the last hand-off and when a person last laid eyes on it. It does not know when it is next due for calibration — and the reason is more useful than a flat absence: the machinery that watches a date arrive is built and running, for batch expiry, quotations, purchase orders and invoices. It is simply not wired to assets.
Our asset register can tell you exactly what a departing officer holds — the list is one click away on their own record. Nothing in the process of them leaving asks the question, moving one person's custody to their successor is one action per asset, and while the incoming custodian can now sign for what they take, nothing requires them to. Three specific gaps, each of them closable with a line on a clearance form.
Two years on, somebody asks what the branch cost and nobody can say. A budget line fragments by category and resets at year end; a project holds the whole thing — with two gaps worth planning for.
Books go out in January to learners who exist in no database you own, and some come back in November. Which custody mechanism to bend, how far it bends, and the annual habit that stops the register quietly becoming a work of fiction.
You bought twenty laptops as stock and four are now staff equipment. That is a unit leaving your stock ledger and reappearing as a tracked asset — in two phases, with an approval in the middle and a cost basis inherited from your weighted average.
A record saying "issued to J. Otieno" settles nothing. One with an approver, a location and a verification date settles most things. And the default for the most important setting gates nothing at all until you configure it.
Two different documents doing two different jobs, and conflating them is the most common mistake in Kenyan fixed-asset practice. Exactly which one you have today, what the dashboard value trend really computes, and the depreciation schedules we can add for your year-end.
One needs twelve records, the other needs one record and a count. The choice quietly determines whether your register is maintainable — and it is much harder to reverse than to make.
A cooperative tractor is a member service, a revenue line and an asset quietly consuming the society's surplus. Costing an hour honestly, the hire log that replaces an hours meter, and presenting the subsidy members never voted for.
The return trip costs more than the margin on the original job. Why two of the five causes are intake failures, treating each van as a counted stock location, and the parts you fit but never bill.
Buses are the most expensive assets most schools own and the least documented. Why the odometer reading at the pump is the whole fuel control, what a trip log should leave behind, and whether the transport fee covers the service.
The generator that failed had been serviced twice in four years and nobody could say when. Maintenance as a recorded movement, the rising-frequency pattern that decides repair versus replace, and why there is no scheduler.
Getting an asset onto the register is the easy half. Five ways assets leave, why scrapping needs its evidence created before the item is gone, and the rule that keeps a register credible: retire, never delete.
Maintenance is where agencies lose money and trust at once — through repairs nobody tracked, approved, or verified. A work-order discipline from report to verified-closed.
A contractor's plant is capital designed to move — which is why it goes idle, goes missing, and breaks down unplanned. Custody, movement history, and service discipline.
For installers, distributors, and service crews the office is wherever the van stopped — job evidence, traveling stock, custody, and fleet discipline in one guide.
Broken equipment announces itself; miscalibrated equipment misdiagnoses quietly. Three maintenance tiers, calibration discipline, and the replacement horizon.
Cost per vehicle, fuel-to-distance baselines that beat gadgets, and the route math that decides which vans deserve to exist. A method rather than a module — the honesty block says which parts we hold.
The auditor's schedule says what was bought; nothing says where it is. Fleets, branch fit-outs, ICT custody, repossessed collateral, and the register that survives elections.
Schools budget annually for equipment they already own. Custodians by name, term verification rhythms, and maintenance that beats breakdown.
A firm that sells time runs two separate clocks — the one payroll pays from and the one it invoices from. Only one can be locked at month end, and it is not the one you bill from.
A credit limit stops the next invoice to a client who has stopped paying. Nothing stops your team logging another two hundred hours to them — one report will now tell you how many, but none of them refuses the hour.
A massage has no stock, but there is no service item type and the till refuses what it cannot allocate — with no override. Three honest answers, which one to use when, and how to set up a stocked service item so it never refuses a sale in front of a guest.
A clinic billing KES 4m a month at ninety-day settlement permanently carries KES 12m of somebody else's money. Why you must measure two ages rather than one, and why a third of your ageing is yours.
Corporate accounts and long-stay guests need the same invoice on the same day of every month. Credit limits, statements, matched payments and automatic chasing are all built. The generating is not — and an invoice nobody raised is completely silent.
The earliest sales record is a quotation. Leads, deals, stages and forecasts have no home — what that costs you in four specific ways, and why raising speculative quotations is the worst available fix.
A return is the money, the stock and the story — and only the first is a credit note. Credit notes here carry an amount and no lines, and move no stock at all, which is how inventory quietly drifts upward.
Exposure is the opening balance plus every open invoice plus the one you are raising. What the check measures, why paying frees headroom immediately, and the setting that silently switches it all off.
A commission scheme runs accurately here in about an hour a month, on data you already capture. The one join between user and employee to set up first — and the targets and automated commission engine we can add when the manual month stops scaling.
A 38% landed uplift turns a 30% markup into a loss, and that is the ordinary imported container rather than a bad one. Three customers at three prices, and the supplier price list we can add so every invoice checks against an agreed rate.
A dollar-priced subscription is an unhedged monthly currency position renewed forever. What a 20% move does to it, the three second-order costs nobody quotes, and the difference between a contract currency and a display currency.
Whatever a vendor charges for, you will ration without deciding to. Per-user pricing pushes you into shared logins; per-module pricing breaks the records-referencing-records that made an integrated system worth buying. How to read either shape.
Real numbers in shillings, ours included. What the subscription costs, what the first three months cost on top of it, and the three line items every quote leaves out because they are your hours rather than the vendor's.
The cheapest sale a business can make is the second one to a customer it already has — and the reason most do not make it is that the record of the first sale lives in a salesperson's phone.
Kenyan businesses lose far more to slow payment than to bad debt, and chasing harder is rarely the fix. Most late payment is friction you created at invoicing — plus the escalation ladder for when it is genuine.
Margin is rarely lost in one bad negotiation. It is given away in a hundred small concessions nobody recorded, by people trying to close a sale — which is why the fix is structural rather than motivational.
The delivery note that comes back illegible is why an invoice gets disputed six weeks later. Why a printed name beats a signature, how to express a partial delivery, and treating delivery paperwork as a receivables control.
Nobody decides to become a lender; it happens one favour at a time. Why exposure matters more than the invoice at the counter, how to set limits from behaviour, and the collection rhythm most retailers waive for their biggest customer.
License models, the implementation costs nobody quotes upfront, honest three-year totals in KES, and the questions that expose hidden pricing.
A cash session here is a real record — declared float, recorded drops, a counted close and a stored variance with a name on it — and the till refuses to sell without one. There is one catch that finds every shop on its first shift change.
The counter cannot sell on credit, by design, and there is no fee account to charge. Two shapes a school shop can take, why the credit limit is the control that matters in week one, and the small thing that decides your whole term's collections.
A return runs stock, revenue, cost and cash backwards at once, and most of that is built and enforced here. One tender on the refund dropdown was implemented nowhere, and the ledger was quietly missing a leg on every refund ever given.
A tenth off the price is two fifths off the profit, nothing goes missing from the shelf, and in most Kenyan shops nobody can tell you who authorised any of it. The arithmetic, the controls that exist, and the ones that do not.
One boolean per item decides whether VAT is added on top or extracted from within. Get it wrong on an imported catalogue and every sale is out by sixteen percent — quietly, until somebody reconciles a return.
A till is not a place in your shop — it is a record with a name, a warehouse, a location and exactly one person attached. Almost every argument about a short drawer traces back to one of those four fields being wrong.
The customer pays part by phone and the rest in notes. What your till records decides whether tonight's drawer reconciles — and recording it wrong manufactures shortages that nobody can trace.
Two independent guards against double filing, retries with backoff, the signed receipt stored on the invoice — and the two gaps a VAT-registered business must decide about before go-live: credit notes, and what a till does offline.
eTIMS is the one fiscal regime on this blog we integrate with rather than reconcile against. What changes at the counter, what happens to returns, and what to confirm with KRA.
The end-of-shift count is where most Kenyan retail disputes happen, and almost all of them are arithmetic nobody wrote down. The float, the drops, the expected figure and the variance.
A till that sells and a stock system updated later will disagree by Friday. The warehouse behind each counter, the batch that leaves with the sale, and why the system should refuse a sale it cannot supply.
Four decisions hide inside every return, and collapsing them into one hurried moment is what breaks the records. Why restocking must be deliberate, how exchanges are really two transactions, and how unrecorded refunds manufacture till shortages.
The smallest document in your business and the one customers most often need. What belongs on the slip, why a stored copy settles disputes paper cannot, and the practical realities of digital receipts in Kenya.
eTIMS on every receipt, shrinkage visibility, multi-branch control, and a straight answer on what happens when the line drops — what to demand from a retail system, with an evaluation checklist.
The 20-minute evening routine that keeps a shop honest — cash, M-Pesa, and card reconciled against the system, shift by shift, variances named.
A church pays pastors, employs support staff, relies on volunteers and hands honoraria to visiting ministers. Four relationships that look alike on a bank statement and are entirely different in law.
The roster is a plan and attendance is a fact, and the gap between them is the information you actually want. Plus why the locum paid from petty cash is invisible to every cost figure you have.
A school pays some staff, records others it does not pay, and hires more each term. There is exactly one run per month with no pay groups, and no exclude-from-payroll flag — so keeping government-paid teachers quiet is a decision you make once, deliberately.
The only pay basis held is a monthly salary. What actually works for harvest crews and fundis paid by the day, the two employment-type vocabularies that do not match, and the legal clock nobody is watching.
A six-day week, Idd on a date nobody can print in January, and a leave balance that survives an argument. One setting drives leave, approval deadlines and SLA clocks — and attendance deliberately ignores it.
Most Kenyan employers have staff with no company email and no login. Two distribution paths, one token and PIN across four portals, and the finalised-runs-only guard that prevents the realistic incident.
Deductions here are per-period entries, which keeps every payslip explicit and easy to audit. The five-column advance register that turns them into an amortising balance — and the exit case where the money is actually recovered — is something we can add to your payroll.
The employee file is the least glamorous thing in HR and the first thing anyone asks for — in a dispute, at an audit, or when the person who knew everything about your staff resigns.
Four statutory deductions, four sets of rules, and rates that move. What a Kenyan payroll run should do, why date-effective rules matter more than the arithmetic, and the reconciliation that catches errors before the money leaves.
Leave is the first thing employees dispute and attendance is the first thing that quietly corrupts payroll. Both are the same problem: a record that either exists when it happened, or gets reconstructed later from memory.
The hardest question in a professional firm, usually answered by asking around. Capacity as a number rather than an impression, why utilisation targets create the wrong behaviour, and the weekly meeting that beats any tool under fifty people.
The largest controllable expense, paid from members' money, discussed in front of the members. Date-effective statutory rules, allowances that need a record made before payment, and presenting a wage bill as four drivers rather than one total.
Overtime is the payroll line nobody can defend and everybody pays. Why derived overtime beats claimed overtime, where the approval step earns its keep, and why we state no multiplier anywhere.
Exits run on goodwill and a cake, and the risky parts get done later by whoever remembers. The four workstreams, why you reassign before you revoke, and the logins no offboarding process ever touches.
What each Kenyan compliance regime actually requires from your system — eTIMS invoicing, statutory payroll, withholding, records — plus vendor test questions.
The statutory stack done right, plus the NGO layer: allocating staff costs to grants, paying casuals and enumerators properly, and the monthly reconciliation triangle.
Our budget module gets the hard part right: money leaves a budget when the order is approved, not when it is paid, and a cancelled order releases the commitment. What it does next is the part to know about — it works out that you are over, approves the request anyway, and sends a notification explaining why. A control refuses; this one explains.
Cost-to-income is the one ratio every SACCO board knows and almost none can defend. Most societies could move theirs five points through classification alone — which means the number currently carries no information.
Residential and commercial letting sit in different tax regimes with different invoice obligations. And the compliance risk in eTIMS is not the invoices that transmit — it is the ones that quietly did not.
The feature that removes a daily job: a C2B payment quoting an invoice number attaches itself to that invoice. Collections, disbursements and reversal on your own Daraja credentials — plus the four gaps to plan a payout process around.
"Sync" does more work on integrations pages than any other word in software marketing. Ours pushes vendors, items and purchase orders out on a button and reads nothing back — including a note on where our own connector copy overstates it.
Running both systems is correct for one module, for four weeks, with a stop date announced in advance. The weekly reconciliation that makes the double work worth it, and the one thing that destroys the test entirely.
Stock gets counted; the financial side has no equivalent. With no manual journal entry, an opening position arrives as real documents or not at all — plus why your chart of accounts has no numbers, and where the statutory book stays.
Nobody ever went bust on expense claims, which is why they go unmanaged for years. The four structural gaps that make ordinary behaviour expensive, and the approval design that closes them.
Profitable businesses run out of money regularly, and almost always with two weeks of warning nobody was looking at. A thirteen-week rolling forecast you can maintain, built from records you already have.
Year-end is only painful when treated as an event instead of a consequence. The practical close checklist for a Kenyan SME — and how to make next year's close a formality.
Every business runs on a rhythm of statutory deadlines, and missing them costs penalties. The recurring compliance calendar in plain terms — and how to let your system carry it.
A practical system for tracking restricted funds from grant agreement to donor report — fund segregation, budget lines, burn rates, and audit-ready advances.
One invoice per student, a unique payment reference, all channels into one ledger — how bursars end the evening matching game and make arrears report themselves.
Every operating shilling is a dividend shilling that left early — live budget envelopes, sitting allowances done properly, and the monthly five-number review.
What you charge for an hour can be overridden on that hour. What it costs you cannot be overridden at all — so a mixed-grade firm reports margin improving in the month it is collapsing.
The only phase of a contract with cost and no production. The team has demobilised, the subcontractors are paid, and every repair from here comes straight out of margin — including the four you were never liable for.
Measured work is priced by the unit and consumed by the unit. Preliminaries are priced by the contract and consumed by the week — which is why a job that hit every material target still lost money on a six-week delay.
The same fourteen steps, rebuilt from memory every quarter by whoever is free — and the two steps that only mattered once get forgotten. Templates and recurring tasks solve different halves of that, and confusing them is why people think neither works.
Agile metrics are genuinely useful to a small number of teams and pure theatre for everyone else. How to tell which you are, what the burndown actually computes, and the one setting that makes it read zero forever.
Forty-one tasks in a flat list gets abandoned by week three. Nine deliverables with steps beneath them does not. The structure that makes a plan readable — and the one number on the project page you should not take at face value.
Logging hours tells you what a job cost. It does not pay anybody. Turning unpaid logged time into money that leaves your account — without paying the same hour twice or losing the trail to the bank.
Customers return because you know their car, and that knowledge lives in a folder that gets lost and a mechanic who leaves. How far a registration-prefix convention gets you, and the declined-work note that is worth more than any discount.
Most workshops make their money on parts and believe they make it on labour — and the two answers recommend opposite investments. How to get an overhead-per-hour figure, and what to charge when a customer brings their own parts.
A job card runs as a project here, which gives you parts at stamped cost, labour hours and a real margin per job from day one. Vehicle history across visits is the piece we can add on top, and it is worth scoping early.
Rollouts stall on small ambiguous decisions nobody has authority to settle. What the owner actually decides, the twelve to fifteen days it costs, why it should not be the IT person, and the six things to hand them in writing.
Phasing is the safe answer and it is not free — you pay in a seam somebody carries by hand. Which modules genuinely cannot be separated, and the three situations where switching everything on at once is the better risk.
Configuration takes a week. What takes ninety days is the organisation deciding which of its four stock figures is real, who is allowed to write off damage, and what replaces the WhatsApp group that currently approves purchases.
For a firm that sells expertise, time is the only inventory — and the one form of stock that vanishes if you do not count it the same day. Utilisation, realisation, and where the hours actually disappear.
A cost that lands nowhere lands in overhead — and overhead is where project profitability goes to hide. The four routes cost reaches a job, and the one-field rule that closes all of them.
An overrun is not an event, it is a trend that becomes an event. The reporting rhythm and the three numbers — budget, committed, actual — that let you see one forming in time to do something.
Retainers look like the most stable revenue a firm has and are often its least profitable work. Why you record time on fixed fees especially, the three billing models and their disciplines, and the final milestone nobody defines.
Variations are lost in small pieces, agreed on site under pressure and argued about at final account. Why the instruction matters more than the price, the five-line note that survives, and why coded costs settle higher than estimates.
Functions look like the most profitable business a hotel takes, until somebody costs one. Why an event is a project rather than a sale, the three numbers to fix before the day, and what costing six events tells you about which to chase.
Projects end twice — when the work stops, and months later when the last invoice and the last claim land. Why the reported margin is never the real one, the retention trap, and the one review table that changes future bids.
Most project plans are tasks with dates typed beside them, which is a wish rather than a schedule. What dependencies buy you, why slack decides which delays matter, and the procurement lead time nobody records.
The BQ is a promise made at tender; the site is where it meets reality. What live job costing takes, what a task-per-element gets you, and the line-level comparison that stays yours to make.
A contractor runs three businesses at once — materials, plant, and fixed-price promises. Where the money leaks on a Kenyan site, and how the profitable ones keep all three under control.
Three buttons on a ticket, none of which changes anything. Why suggest-only is the design rather than the limitation, what each prompt is actually asked, and the one that will waste your time.
The word "requester" hides four different kinds of person, stored differently, able to do different things — and one kind you assumed existed and does not. Which one raised a ticket decides what your reports can ever tell you.
Most helpdesk failures are silences, not slow replies — a ticket that reached the system and never reached a person. Every event, every recipient, every channel, and the four ordinary setups where the answer is nobody.
One checkbox decides whether a comment reaches your customer, starts the first-response clock, notifies anybody, or leaves your server. Four consequences from one boolean, and the two ways it goes wrong.
Somebody resigns on Friday. On Monday forty open tickets still have their name on them and every notification is going to a disabled account. What the system catches, what it does not, and the twenty minutes that prevent it.
Most support portals fail because they ask people to create an account before they can complain. What makes an intake channel people actually use, and the honest limits of a portal.
An SLA you cannot measure is a hope you have shared with a customer. What response and resolution targets require, why the pause makes them fair, and how to set targets you will not quietly abandon.
The largest pile of untracked work in most Kenyan organizations is not customer support — it is colleagues asking colleagues for things. Why internal requests are the best place to start.
Tickets received and closed are the two most reported and least useful numbers in support. The four measures that show whether a desk is coping, why reopens beat resolutions, and how to read satisfaction honestly.
A ticket at the wrong desk teaches people the system is slower than walking over. Why the category carries the policy, why priority does not move the clock, and what escalation means when nothing pages anyone.
The paragraph at the top of an AI dashboard is the part everyone reads and nobody interrogates. Every figure computed in plain code, the model forbidden to invent any, and a fingerprint that discards the sentence when the numbers move.
Three formats, a row cap worth knowing about, and a CSV convention for empty cells that changes how Excel totals a column. What to expect from an export today, and the wider formats and larger row limits we can enable for you.
One question, two modules. A report is built on exactly one dataset, which prevents a whole class of silent double-counting — and is why thirty-four cross-module reports ship prebuilt. Where you need another, we can add the joined dataset safely.
A figure nobody should have appears in a group chat. Running a report and exporting it are two separate recorded acts here — which is the distinction that makes the question answerable at all.
Certification marks the report the organisation stands behind — and guards against the quieter problem it exists for, which is a saved report drifting away from the data underneath it without anybody editing it.
Go-live is the start of the work. The three numbers to record before you switch on, the four reports worth thirty minutes a week, why every threshold you configured is a guess until day 60, and the four signs it is genuinely being used.
The gap between a report that exists and one that reaches the person who acts on it is most of the value. Scheduling, who to send to, and how to stop scheduled reports becoming ignored mail.
The most expensive report is the one somebody assembles by hand every month because asking IT takes three weeks. What a report builder removes, and how to avoid the mess when everyone can build one.
Every business asks for reporting and most end up with reports nobody opens. The difference between a report and a decision, and why reporting problems are nearly always records problems in disguise.
A dashboard with twenty tiles is a wall decoration. How to choose the six numbers that would have warned you about last year's surprises, and the rule that keeps a dashboard useful.
A lease that expires unnoticed costs more than a month of rent — it costs the negotiation you never had. Four dates per lease, what a turnover truly costs, and why the renewal conversation belongs at ninety days.
Most board packs are assembled by hand in the four days before the meeting. What belongs in one, why a certified definition beats four accurate spreadsheets, and the four choices that make two reports disagree.
A shared report is a permission decision wearing a friendly interface. Roles versus named people, the difference between sensitive and export-restricted fields, and why row-level filtering is one report per audience.
Offline capture in our mobile app covers exactly four operations, and a stock count is not one of them. The count screens are on the phone and they work — they just need a live connection, because a count is not a delta like the other four, and because a stock freeze is a promise a disconnected device cannot make. For a branch pharmacy with weak signal that is the most important sentence in any demo.
Every public tender has an integration line and it is usually answered with a yes. Ours: tax filing to KRA runs automatically against real endpoints, queued so a sale is never blocked, with the receipt number stored on the record. IFMIS, nothing at all. The difference is structural rather than effort, and knowing why will help you write a better specification.
An auditor asks to see one that was refused. Our product answers that well in the two places it records decisions properly — item master changes and deletions, both with segregation of duties. In the generic workflow engine, which has by far the richer approval model, the approver receives a task whose four states do not include yes or no.
Hospital bills, school fees, funeral costs. The most pastorally sensitive money a church handles, usually the least controlled — and the person an informal process fails to protect is the minister.
Every ministry gets an allocation in January. In November nobody can say who stayed inside it — because a budget is set on a department and half of church spending cannot carry one.
Owners do not leave because rent was late. They leave because the statement did not tell them whether it arrived. Four numbers that must reconcile, and the client-account check that protects the agency.
A building becomes a project, a unit becomes a naming convention, a tenant becomes a customer. Which stand-in you pick decides what you can report for five years — and a department will quietly fail you.
A scheme can come in under budget and be heading for a special levy, because the operating shortfall quietly ate the reserve contribution. Only line-level variance shows it — a total never will.
A member you sell fertiliser to and buy maize from is a customer and a vendor, and nothing joins the two records. The instrument that gets you most of the way there, the sequence that survives an audit, and the accounting hole to tell your accountant about first.
Every AI feature is a data transfer decision wearing a friendly button. The complete inventory: what each feature sends, to whom, what never crosses the boundary, and how to answer your compliance officer.
A mixed container is the exact case where allocation matters most and the correct basis is unavailable. Why quantity allocation destroys a clip's cost, the split-delivery defect to work around, and why demurrage always lands after you start selling.
"Will this fit a 2014 Probox" is a database query executed by a man who has done it for eleven years. Why fitment is a licensed data product rather than a feature, what searchable fields genuinely buy you, and when to run a catalogue alongside.
Six routine tasks with a stopwatch against them and an annual figure each. The largest line is the one nobody complains about — and three of the six do not improve at all, which is the part that makes a case survive review.
From a software company that would rather you buy at the right moment than cancel in month four. The sizes where a spreadsheet is still genuinely correct, and the six signals that say you have outgrown it.
One page, five parts: four measured numbers, one deliberately blank, the full cost including your own hours, a paragraph on what you are not buying, and three numbers with a ninety-day review date.
A Nairobi build looks competitive against three years of subscription and never prices maintenance or the day the developer becomes unavailable. The four questions, a five-year comparison, and the two cases where building genuinely wins.
Free the way a puppy is free. Where the money actually goes, the three roles a self-hosted deployment quietly requires and nobody staffs, and the two situations where it is genuinely right in Kenya.
The subscription is under a third of it. The five components, a five-year model for a forty-staff distributor, and the recurring item nobody puts in the spreadsheet — two hours a month of somebody tending the thing.
Real, predictable and absent from every proposal, because seven of the nine are your hours rather than the vendor's. Each with a rough shilling figure and the week it lands — plus three costs people budget for and rarely incur.
Three different things get sold as offline and only one is worth anything to a driver in Kitui. Idempotent replay so a lost response cannot duplicate work, conflicts surfaced rather than resolved, and device trust that expires.
What connects today, and what we can build next. Bank feeds, other mobile money rails, additional tax authorities, more accounting packages, inbound messaging, marketplaces and scheduled pulls — where each one stands, and how to get it into your workspace.
In Kenya an alert not on WhatsApp or SMS has not arrived. Both work on your own credentials, WhatsApp has one template rule that catches everybody, and the volume discipline matters more than the setup — because muting is silent and permanent.
The cheapest integration available and the only one that changes behaviour rather than moving data. Six channels, why three topic channels beat one, and the mute problem that makes this look like it is working for months after it stopped.
There is a real API and there is no developer portal. How tokens are issued, why a token carries the user's permissions rather than its own, what the API is genuinely good for — and why continuous two-way sync is not a shape it supports.
Every integrations page is a grid of logos, and the tiles are four completely different products with four different failure modes. How to decode any of them with two questions, and which one to connect on day one.
Thirteen templates download and eight import. Which is which, why the other five deliberately have no importer, the dependency order that stops a file failing on row one, and the fifty-row habit that substitutes for a preview.
Exit terms are the cleanest signal about how a vendor thinks, and you have most leverage over them before you join. The four questions, the answer almost no SME vendor gives well — attachments — and our own honest position.
Two hours before go-live, permission design is politically free. Six months later it costs a week and makes enemies. How to build roles from the four separations that matter, against 255 permissions — and where the model stops.
A full day in a room three weeks early teaches almost nothing, and everyone involved knows it. Twenty minutes, one role, one task, at the moment of need — plus one uncomfortable rule about who should train the storekeeper.
Six things that decide a rollout, ranked by how often they matter, and every one is organisational rather than technical. Features have never been the deciding factor — and the two that carry a project on their own are settled in a meeting, not bought.
Four of the Act's seven operational duties are wholly yours, three are shared, and none can be discharged by a purchase. The controller and processor boundary, what belongs in a processing agreement, and the breach-notification gap stated plainly.
The most sensitive data a church holds and the most useful. Deciding which giving is attributed and who may see it, why pledge totals are the wrong planning figure, and the annual statement that makes the record trustworthy.
A second congregation creates an accounting problem before it creates a ministry one. The three models, why income must be recorded gross at the site, and the remittance reporting that keeps branches from resenting the centre.
A recharge recovering 88% of a bulk bill loses money every month, invisibly. The recovery ratio per building, why you record readings rather than consumption, and the handover reading that ends a whole category of dispute.
For a small part of a clinic store the question is never "how many" but "who, when and on whose authority". Where to spend two-person discipline, why small discrepancies get investigated here, and what a general inventory system cannot discharge.
You do not need to be a data business to be a data controller — an employer with payroll already is. Where personal data actually hides, what retention means in software, and the honest limit of any product. Not legal advice.
Kenyan businesses make everyone an administrator in the first two weeks because it stops the complaints — and deletes the meaning of every other control. Getting back to least privilege without blocking work.
An auditor picks one line and asks you to rebuild it. The documents, the approvals and the change history together — plus the retention trap that makes evidence expire before the obligation does.
Ghost suppliers, inflated prices, adjustments that cover a count, ghost employees, and the till. Each needs two capabilities in one pair of hands — and each has an aggregate view that exposes it in ten minutes.
One login per location is the most common security arrangement in Kenyan SMEs, and it turns your system from a record of what people did into a record of what a place did. Fixing identity first.
Every term you want at the end of a vendor relationship is available before signature and none of it afterwards. No-cost export, post-termination destruction, the grace period on deletion — and where the disposal boundary actually falls.
Most money flowing through an agency is not the agency's to spend. Keeping deposits, landlord funds, and agency income provably separate is the discipline that prevents disaster.
A managing agent is trusted with other people's buildings and other people's money. The four disciplines — collection, maintenance, trust funds, and owner reporting — that keep both in good order.
Collection is the heartbeat of an agency and where most bleed time. The cycle that ends the monthly matching game, ages arrears into a managed list, and reconciles every channel.
Utilization tells you how busy your people are; profitability tells you whether it was worth it. In a firm that sells time, confusing them is how a fully-booked practice loses money.
Money a firm holds for clients is not the firm's money, and proving that at any moment is a professional obligation. What proper segregation of client and trust funds actually requires.
Law, audit, and consulting firms sell time — but still run on operations they never systematize: which engagements make money, whether client funds are separate, and where the firm's money goes.
A co-op sits between hundreds of farmers and a demanding market, and its value is trust — fair weighing and grading, correct on-time payments, and money that survives the journey. The disciplines that keep it.
The moment produce hits the scale is when a co-op earns or loses trust. Weighing, grading, and paying fairly — and proving it — is the core of the cooperative promise.
Most ERP regret comes from buying the wrong size, not the wrong brand. A buyer's guide that starts from your operation, right-sizes the system, and gets you there without overbuying.
An open-source ERP can do the job — the real question is who makes it do the job and keeps it running. An honest comparison on total cost, implementation, developer dependency, and eTIMS/M-Pesa compliance.
ERP projects fail in predictable places — dirty master data, big-bang go-lives, demo-data training. The five-phase checklist from decision to retired spreadsheets.
Excel is free the way a leaking pipe is free — reconciliation labor, error rates, the one-person dependency, and stale-data decisions, computed honestly.
eTIMS mandates, M-Pesa-native operations, multi-branch growth — the forces pulling Kenyan SMEs off spreadsheets, and how the successful ones sequence the move.
Schools are three businesses in one uniform — fees, logistics, and assets. What operations software must handle, and where school money actually leaks.
The core system runs the loans; nothing runs the rest. Procurement, assets, branch expenses, and payroll — where member money leaks and governance is tested.
The menu is a price list for recipes nobody costed — how to cost a card, hold a portion, record waste as a transaction, and read the four-box menu grid. The recipe layer is a spreadsheet here, and the post says so.
Ten questions that separate partners who will still serve you in year three from vendors who disappear after the deposit — references, SLAs, compliance, and exit terms.
The second branch is where trust stops being a control system — live dashboards, governed transfers, and branch P&Ls that keep growth profitable.
The audits Kenyan NGOs face, the findings that repeat across the sector, and a 90-day plan that turns audit week from reconstruction into retrieval.
A church runs on the trust of those who give to it — and trust in money matters is built by controls, not intentions. How well-run churches handle offerings, project funds, buying, and assets.
The advance lifecycle, per-diem rules that stop field drama, and M-Pesa discipline — closing the gap between when money moves and when evidence is captured.
When members give toward a building, the money carries a promise. Pledges against receipts, restricted from general funds, spent transparently — how the promise is kept to completion.
HR records which department an employee belongs to. Procurement asks which department a login belongs to. Two columns on two tables — and when HR provisions that employee a login, the department does not travel with it.
For most of your staff the department on a purchase request is stamped from who they are, and the request is refused when it cannot be. For an administrator it is a free choice from the whole list, and nothing records that they made it on somebody else's behalf.
Fourteen tables point at the department list — payroll, budgets, purchase orders, assets, tickets, projects and the workflow queue. The record they point at holds two fields, the settings menu calls it a cost centre, and deleting one checks four of the fourteen.
A budget is a department, a category and a window. Purchase-order spend against it is department-scoped; expense spend is not, because an expense carries a project instead. One column of the report is departmental and the one beside it is organisation-wide — and it says so.
A ticket knows which department raised it and, to the minute, how long it waited. It holds no hours, no rate and no cost — and hours are costed a module away, against a task that carries no department.
Australia asks large buyers seven questions about their supply chains every year. Only one and a half are questions a system can answer from its own records — and a supplier register, by construction, stops at the company you pay.
Indian e-commerce rules give a grievance officer 48 hours to acknowledge a complaint and one month to redress it. Three settings in an ordinary help desk decide whether either period is measured — and the one that matters is not a clock.
British lifting-equipment rules give you a six or twelve month floor, two events that bring the date forward, and a next-due date a named person writes into a signed report. A recurring reminder is none of those three.
Large UK companies publish how fast they pay suppliers twice a year, and every figure is a difference between two dates the Regulations define precisely. One of them is not the invoice date, not the date you entered it, and not the due date.
Singapore prescribes what a payslip contains, item by item. Two are dates of payment, two are the first and last days of a period, and one is a period that need not be the salary period at all — so we held our payroll against all eleven.
A certified payroll has a period of one week, a scope of one contract, and a signature attesting to three specific things. None is a cost figure — and its most unusual requirement is something this product already does, in the wrong module.
The US food traceability rule is summarised as lot codes and a 24-hour clock. The sentence nobody quotes asks for your coding systems, abbreviations and a written map from your records to the elements it names — a document about your software.
A British contractor files monthly, naming each subcontractor, what they were paid, and how much of it was materials. That third figure is defined as what the contractor is satisfied the subcontractor's own materials cost — not a number on any document you hold.
British price marking treats a price as five properties: a number to a stated precision, a unit, a condition, a currency and a relative type size. A price field is one of them — and two of the five belong to how something is printed.
Singapore's security-of-payment Act defines a day, for every deadline in it, as any day other than a public holiday. Weekends count. That is a third unit of time, and most systems implement two.
Singapore presumes that goods failing within six months of delivery were faulty when they were handed over, and the seller has to establish otherwise. That makes the delivery date and the individual unit the two records that matter.
A US forklift certification has no expiry date. Five things oblige you to retrain the operator and four of them are events, not dates — which is a problem for every mechanism built to watch a calendar.
Japan makes an employer of over a hundred people measure four things before writing a plan about women at work. Three are arithmetic over records this product already keeps. The fourth needs a column we do not ship, and we think we are right not to.
Canada gives you two days to report a product incident, counted from the day you became aware. That is a date about your organization rather than about a record — and it is the one date no operations system holds.
Canada's legal-tender limits are written in denominations, and a till records a payment as one number. The gap is not a compliance problem — it is a modelling one, and two of the missing pieces are worth having anyway.
In Victoria, whether a repair is urgent is a closed statutory list, and it decides more than a deadline: one route lets the renter spend your money, the other brings in a regulator. Our categories hold the clocks and not the consequence.
Victoria caps what a builder may hold at each stage as a percentage of the contract price, with the stages defined by what is physically built. Our projects module holds the stages in order and holds no contract price at all.
Singapore's packaging report is triggered by a turnover figure and made of kilograms by material. One of those we can produce for any period; the other has never been entered into the system by anybody — and the Act gives you exactly one year to start.
The Philippines requires nine disclosures before a consumer credit sale is consummated, and two of them are different interest rates. Our invoice holds a balance and a due date; the word instalment appears nowhere in the codebase.
Japan's misleading-representation penalty is three per cent of the sales you made of those goods during the period the claim was displayed. We can produce the sales for any window; nothing here records when a price or a claim was in force.
Hold a cheque too long and the drawer can be discharged to the extent of the damage your delay caused. A cheque carries three dates and a payment row here has one column — and the till cannot take one at all.
Victoria wants a written incident record within 48 hours, on a form the Authority defines, kept five years and shown on request to people who do not work for you. Every report we make well is one you designed yourself.
Japan gives a mail-order buyer eight days from delivery to send the goods back — and takes the right away if you published your own terms in the advertisement. The window is a property of a document, and our return row records neither the route nor the postage.
Victorian payroll tax groups employers across company boundaries, makes the liability joint and several, and lets a grouping determination take effect from an earlier date. Our workspaces are built not to produce that figure — deliberately.
Japan makes the main contractor keep a register of everyone working below it, on site — and makes each tier tell the tier above who it engaged. The party who keeps the register is not the party who fills it in, and our supplier portal is already the right channel.
Victoria makes a second-hand dealer keep goods exactly as they arrived for seven days, record where each item is stored, and — for scrap metal — never pay in cash. Our till is built for the other direction entirely.
A customer credit limit here genuinely refuses — past the line the invoice is created on hold. What the number never records is which currency it is expressed in, and the exposure it is weighed against is a sum of invoice balances that each carry a currency of their own.
A sprint belongs to a project, and a task can only join a sprint in its own project. For a squad built around one engagement that is invisible; for a standing team carrying a client portfolio it is one cadence turned into four records, four burndowns and four velocities.
Peru sets a monthly valuation cycle for public works with the deadlines written into it — present within the first three days of the following month, supervisor reports within five, entity pays by month end. The binding deadline is not the payment one; it is how fast you can close a month.
Nigeria clears a business-to-business invoice before it reaches the buyer and reports a till receipt after it is issued. Those are two machines, not two settings — and only one of them can stand in front of a queue on a Saturday afternoon.
South Africa's employment equity regime sets numerical targets by sector and occupational level, and a certificate for government work turns on them. The report asks your HR record for three classifications, and a conventional employee record carries none of them as fields.
Japan caps overtime with four tests running at once — a monthly limit, an annual one, a rolling average across two to six months, and a count of how many months went over. Only the first can be answered by looking at a month.
A milestone here is a name, a due date, a status and a sort order — the one level between a task and a project that reports cleanly. Turning that date into something that gates, triggers and alerts is an addition we can wire into your workspace.
A project here can pay the people who worked on it, through a document with its own reference, approval and posting date. That is a second way of paying somebody, running beside payroll, and neither one knows about the other.
Overriding a customer credit hold here needs the permission to edit invoices — the same grant that lets somebody change one. Three modules away, an equivalent override was deliberately given a permission of its own.
Payroll here is maintained for one market and configurable for every other. Those are not two grades of the same feature — they are different products, and the difference is who is responsible on the day a rate changes.
A login here can be revoked the moment you decide to. For contractor and fixed-term accounts we can add a scheduled end date, so the revocation lands on the day it was agreed without anyone having to remember months later.
A journal entry here carries an account, an amount, a reference and a description — clean, fast and easy to audit. Branch, department and project dimensions, so you can read profit by any of them, are an extension we can make for you.
The moment a report becomes a file it stops being a view of your data and becomes a copy. Live shared report links, and a stored metric series to reconcile against later, are both additions we can enable so the number stays current.
A report here is built on one dataset — a sound design that keeps a whole class of double-counting out of your numbers. The joins worth adding on top, where you need a supplier and a customer, or a purchase and a complaint, in the same row.
Satisfaction ratings can only be submitted by a requester with an account, through the self-service path. Everybody who reached the desk without one is missing from the average, and the average does not say so.
The most common way a support team hits its first-response target without responding is to write a note to itself. One condition in this codebase makes that impossible, and it is worth knowing which one before somebody helpfully removes it.
Ten stock statuses are declared. One counts as sellable and five count as a hold. The other four are neither — so stock put into one of them cannot be sold and is reported as not being on hold, which is the worst of both descriptions.
A tax rate here cannot be created above one hundred per cent by any route. It can be updated above it through one. The rule was written on two doors out of three, which is the ordinary way a validation gap happens and the ordinary reason nobody notices.
The request-for-quotation form makes quantity compulsory. The API does not, on either creating or updating a line — and where it is missing, null is written. A supplier is then asked to price something without being told how much of it.
The till and the stock valuation once used different definitions of what an item cost. Both were defensible, both were implemented correctly, and the integrity report comparing them reported a variance that was purely a difference of opinion.
A comment on the till code promised that failed fiscal filings were reconciled with a manual button. The comment had been copied from the invoice flow, which had one. The till did not, and the job gave up after three attempts.
This module will refuse to move a project from planning straight to completed. It will not refuse to spend twice its budget. One of those is enforced by a state machine and the other is a percentage on a screen.
Copying a project rebuilds the whole structure — tasks, subtasks, dependencies, milestones, checklists — and deliberately resets every date, assignment and trace of progress. The list of what it resets is the most opinionated thing in the module, and we can tune it to your templates.
Tasks here can have subtasks, and subtasks can have their own. The tree is real, it survives cloning, and it orders work. Rollup arithmetic — hours, points and cost summing from children to parent — is the natural next layer, and one we can build for you.
A task here can carry story points and a sprint, or estimated hours and a milestone, or all four at once. Only one of those vocabularies reaches the cost figures — so which one you plan in quietly decides whether your plan can be costed.
An employee's hour is recorded twice here — once as attendance that becomes pay, once as effort that becomes project cost. Linking the two, so you can compare what you paid for against what you billed, is an integration we can build into your workspace.
An employee record here does not require a user account, and that is deliberate — in most of the businesses this was built for, a majority of staff have never logged into anything. Every account-shaped feature therefore covers a minority of your people.
We published that this product stored no daily metrics and computed every dashboard figure from scratch. It was false, and it was on this page. Dashboard history is precomputed nightly, finished days are read from storage, today is always live, and a missing day repairs itself on read. The reason the check went wrong is worth more than the original article.
A report schedule in this product carries its own timezone and computes its next run in it. The deadline engine three modules away has no timezone at all. The capability exists, once, in the place where getting it wrong matters least.
There is a report builder here and, by design, no raw SQL — everything protecting your data sits in the layer a raw query would go underneath. For analysts who want more reach, we extend the builder's datasets rather than lower that floor.
A saved filter here belongs to one person. A report schedule can be driven by one. So the weekly figures a team runs on can rest on a private view that nobody else is able to open, inspect or inherit.
A support contract makes two promises — we will answer by then, we will fix it by then — and they need two alarms. Giving them one shared "already notified" flag would have meant the first breach permanently silenced the second.
The business day here is two clock times and one application timezone — simple and predictable for a single-site team. For a desk in Bengaluru promising four hours to a customer in Lagos, we can add per-desk business hours so each SLA counts in its own clock.
The breach alarm notifies the agent, then their manager, then the department — the right ladder in an organisation that has one. For a six-person business we can configure a flatter path, so the alert reaches somebody positioned to act on it.
Our data export and erasure tooling takes a user account. An external ticket requester does not have one — that is the entire reason they were sent a tracking link. So the people most likely to ask are the people the tool cannot be pointed at.
This till takes cash, one mobile-money rail and a card. It briefly took a fourth on refunds — store credit — which reversed the sale, returned the stock, and recorded the debt to the customer nowhere at all.
An offline till is not one feature. It is four, they fail independently, and most systems claiming offline mode have built one or two. Here is the honest map of which are here — none — and what that means before you buy anything.
A till sale has no approval step to hang a lock on — the money moves and the receipt prints in the same breath. So the signature had to become its own lock: signed once, never replaced, and refused entirely once the sale has been reversed.
A refund reversed the sale, returned the stock and balanced the drawer perfectly. It also left cash overstated on the balance sheet by the amount refunded, and a matching phantom balance in receivables — for the life of the tenant, in silence.
Twelve movement actions are declared in the code. Nine are written by something. The gap between those two numbers is where most custody claims live — and the page that got this right did so by listing what the service writes, not what the module is called.
There is a working document-expiry reminder in this product. It reads a date, respects your notification settings and warns you thirty days out. It is filtered to employees — so a work permit gets a warning and an equipment insurance certificate gets silence.
Nobody adopts an asset register with no assets. The interesting part of a migration is never the file format — it is the four columns you have that the destination has nowhere to put, and deciding in advance which of them you are prepared to lose.
An asset label has two readers. One is a person who will complain when it looks wrong, and one is a camera that will simply fail. We shipped a barcode that was invisible in dark mode and perfect in every screenshot — which is exactly the failure a machine reader produces.
When a message is handed to a mail or messaging provider, the provider hands back an identifier — its receipt for what it just accepted. There is a place to keep that identifier and nothing puts anything in it, so our record of a message and theirs cannot be joined.
An employee’s pay package is designed as a dated series — draft, active, superseded, each with a start and an end. In practice every employee has exactly one, created the first time somebody opens the screen and edited in place from then on.
A project carries one money budget. Its actual cost arrives from four independent streams — logged time, purchases, expenses and stock issued to the job. So the system can say a job is eighty per cent through its money and cannot say which of the four is why.
A task dependency here carries a type, and the critical path treats every link as finish-to-start in day offsets. Reading that type — and teaching the calculation about weekends — is a straightforward addition we can make to your workspace.
A job's actual cost here is four things added together. Three are measured. The fourth is hours multiplied by one rate you set for the whole project — and the wages you actually paid never enter the figure at all.
The timeline here draws your current dates against a snapshot of the dates you agreed to, and shows the variance per task. It is one of the better-built things in this product — and it has exactly one snapshot, with no record of when it was taken.
There is a figure here for work that has been done, is billable, and has not been invoiced. It is the closest thing this product has to work in progress — and the rate it uses can be set per entry, which is exactly what the cost side cannot do.
This payroll engine computes statutory deductions correctly and cannot store the reference numbers those deductions are filed under. Two such columns exist in our database and are read by nothing — not a form, not a payslip, not an export.
Overtime here is requested, approved and locked into a period like any other hour, with a full approval trail. A rate factor — so an approved overtime hour costs 1.5x or 2x — is a configuration layer we can add for you.
A report definition here can be submitted for certification, reviewed and marked certified. Pinning that certification to a specific version, so a later edit re-opens review, is a governance layer we can add for teams that need it.
One permission lets somebody schedule a report to a list of email addresses. The addresses are not checked against your user list, and nothing compares what the report contains against what each recipient is allowed to see.
This product pushes a great deal at people — idle stock, expiring batches, lapsing documents, breached tickets. Threshold alerts on a dashboard number crossing a line are the natural next signal, and we can wire them into your workspace.
Stock valuation, receivables ageing, payables ageing and a payments register all exist and all work — every input a cash conversion cycle needs. Turning them into the three ratios, sized in days, is a report we can add to your workspace.
This product sends WhatsApp messages today. For a market where WhatsApp is how customers actually reach a business, opening the inbound direction — replies landing as tickets — is the next channel we can add to your workspace.
The working calendar is one working week and one holiday list per organisation. A desk covering six countries has six sets of public holidays and one calendar to express them in — and the per-purpose calendar that would solve it exists in code with no way to create one.
When a ticket breaches, this product tells the agent, then their manager, then the department. Assignment rules that decide who should own the ticket in the first place — before the deadline rather than after — are something we can add to your helpdesk.
A discount at this till is four columns: the sale, the line, a free-text description and an amount. No approver, no reason code, no limit and no user. It is a record of what happened, not a control over whether it should have.
The stock module hides the expected quantity from the counter by default. The till displays Expected Cash to the cashier as a tile on the sale screen. Same product, opposite answers to the same control question.
Stock leaves this system by two paths, and the till keeps its own record. Bringing both onto one vocabulary, so a counter-only shop's consumption reports read correctly, is a unification we can apply to your workspace.
This product has full imprest control — a float, recorded movements, an expected balance, a count and a variance — running on the till today. Extending the same control to the back-office cash box is a straightforward addition we can make.
This product tracks where every asset is, who holds it and what happened to it — a register built for custody. A depreciation ledger is the companion system, and we can add schedules, methods and period postings alongside it.
Maintenance is recorded here as a movement: what happened, to which asset, on what date, by whom — a real history you can read. Preventive schedules that generate the next visit from that history are the layer above, and one we can add for you.
Last service and next service already exist as columns on the asset table — the groundwork is in the schema. Turning them into a working service schedule, with reminders and a due list, is a short build we can do for your workspace.
An asset here can be owned, leased or borrowed, enforced across five separate write paths. Extending the same ownership status to stock — consignment, customer-owned, held on loan — is an addition we can make to your workspace.
An asset here is born one of two ways: entered by hand, or converted from stock. Neither remembers the purchase order. For a business buying equipment against a grant, a client project or a capitalisation threshold, that forgetting is the whole problem.
You can record that an asset was verified. You cannot start a verification, assign it, track what is outstanding or close it. The stock module can do all four for a count — which makes this a design inconsistency rather than a missing feature.
You can close an accounting period here. It records the month, the moment, the person and a note, and keeps a history of every close and reopening. What it does not do is stop anybody posting into the month it just closed — and it is on the phone, not the dashboard.
An external auditor asks for read access to the audit log. The smallest permission that opens it also opens settings administration — so granting the narrow thing they asked for means granting a great deal they did not, and two permissions built for exactly this govern nothing.
Four permissions govern who may view, create, edit and delete a proposal. There is no proposals feature — no record, no screen, no address. The permissions describe a plan, because a plan is what a permission list is made of, and buyers read it as evidence.
Our turnover report counts consumption from the till as well as from issues, rolls balances back through the period's own movements, and works over any dates. Then it hands you one figure for the entire company and stops.
Ask our stock reports what one warehouse holds and you get an answer. Ask the ledger and there is nowhere to look, because a journal entry carries no branch. Two views of the same goods, one dimensioned and one not.
A stocktake here does not change your stock. It raises adjustments somebody has to approve — so a counter can report what they counted and cannot set the record. The best decision in the module, and the three things it stops short of.
We have requests for quotation with real line items and recorded responses. We have no request for proposal, no scoring model, no panel and no sealed bids. Those are not degrees of the same thing, and one of the four cannot be substituted procedurally.
A purchase can name the job it is for, and so can a stock issue — costed at the price stamped when it left the store, so a finished job cannot change price later. Two things never reach the job at all, and one of them will surprise you.
There are six ways to pay a supplier here and they write to two different ledgers. For a long time they shared no guard. The fix was not a check but a service all six had to call — and the reason that matters is the lesson rather than the feature.
A customer balance field is written only when somebody creates or edits the customer by hand. No invoice, payment or credit note touches it. And it is the opening balance of every statement and the base of every credit exposure.
A discount at the till is an amount and a sentence — no approver, no reason code, no percentage, no person on the record. A stock write-off of the same value needs a dedicated permission and a second pair of eyes no permission can override.
Document numbers are sequential per organisation, per type, per prefix and per year, zero-padded, and safe when two people press save together. What the key does not contain is a branch — so ten sites share one run.
An invoice line has eight columns. A per-line discount and a per-line tax rate are more than many systems offer. No unit, no fractional quantity, no cost and no project — and it is the missing cost that makes historical margin move.
A workflow can be built with a wait in the middle — do this, pause three days, then do that. The pause works. Everything before it runs, the run is marked paused, and the log names the date it will continue. Nothing in the product ever comes back to it.
The accounting sync keeps a proper log — what it moved, when it started, how long it took, what went wrong. There is a column for who started it, and a defined way to read that column, and nothing has ever written to it.
A project can be attached to any other record in the system — the machine being overhauled, the building being fitted out, the contract it delivers. The attachment point exists, the way to read it exists, and nothing has ever filled it in.
Our scanner bridge turns a phone into a paired scanner and stores every beep as an auditable event with a device, an actor, a resolution and an exception code. It is one of the better things in the product, and it decides exactly one thing where buyers assume four.
Send a hundred, receive ninety-seven, and the system records the three. Not why, not who, not what they were worth. Returning goods deliberately does require a reason — so the intentional act must explain itself and the unexplained one does not.
A budget is a category and an amount, and a purchase reaches it by matching a free-text string on the item. Case is forgiven; a plural is not. So the spend you are not watching is the spend that was spelled slightly differently.
Receive more than you ordered and the system refuses, counting every earlier delivery so a split consignment cannot creep past one lorry at a time. Receive less and it never refuses at all. The asymmetry is deliberate, and the reasoning is worth more than the feature.
Blacklisting a supplier does five things at once, refuses to be undone by accident, and blocks them from reapplying through the public form. Marking one preferred does nothing at all. Two flags on one record, and only one is a control.
An invoice is not created and sent in one act. It is drafted, approved by a named person at a recorded time, confirmed, then sent — and a customer over their credit limit stops it dead at approval. Having a lifecycle at all is the unusual part.
There is no delivery note. An invoice knows which warehouse the goods left from, and there is no separate document that travels with them and gets signed at the other end. For internal transfers there is exactly that.
A customer tax profile carries an exemption and a rate override. Ticking the exemption genuinely exempts them. Typing the override does nothing, on purpose, and the page says so. The rule that separates the two is the most useful thing here.
A payment record carries a reference, a transaction id, an attachment, who recorded it, the balance after it, its currency and the journal entry it produced. It belongs to exactly one invoice — so one transfer settling four invoices needs four records.
A sale from a till app can name the warehouse it came out of. The field is validated and then thrown away in favour of the counter’s own — which is the safe answer, and it leaves the field sitting in the one place a value that does nothing can still do something.
A request for quotation has two routes to the quotes that came back. One is live. The other belongs to an earlier design that was replaced, never removed, and returns an empty list for every request that has ever existed — which is worse than a route that fails.
In most of the world the inventory costing method is an accounting preference and any defensible one will do. In the United States it can be a commitment with consequences reaching past the ledger — which makes it the one market where "our system supports exactly one method" is a fact a buyer needs before signing rather than after. We support exactly one.
There is a supplier performance record in our database with three fields on it: on-time delivery rate, quality score, and pricing trend. They are the right three. Nothing in the product has ever written a value into any of them, and four of the five inputs are already recorded on every order.
Every supplier payment is authorised properly and recorded one at a time. There is no payment run — no way to select forty approved invoices, review them as a batch, approve the batch and pay it. The control at the end of your purchasing process is strong, and nobody ever sees the total before it leaves.
You can tell our system how much a storage location holds. There is a field for it, it is offered when you create one, and it saves correctly. Nothing has ever read it — you can put four times that quantity in the location and the system records it without comment.
Goods can arrive against a purchase order, and goods can leave to a customer. The one direction our system has no path for is back to the supplier who sent them — so a rejected delivery leaves your books as a write-off, and the payment gate still reports itself satisfied.
When stock runs low our system can raise a purchase requisition on its own. It raises it against a department. It does not name a supplier, look at who supplied the item last time, or consider who is cheapest — because nothing in the automation has any concept of who you buy from.
Your clearing agent bills you three weeks after the container cleared. By then some of the goods have sold. That freight and duty will attach to whatever is still on the shelf and to nothing that has already gone — so the margin you reported on the fastest-moving lines is the one that was most wrong.
We built anomaly detection into this product. It watches for items selling below cost, on thin margins, or priced at zero — every one of them a question about money coming in. It has never once looked at stock going out, which is the direction things actually disappear in.
A count plan asks how often you intend to count, requires an answer, validates it and stores it. Nothing then happens on that rhythm — no session is ever raised. The field is not missing, which is what makes it dangerous: a plan reading "monthly" looks exactly like a control that is running. It is the third inert setting we have found in our own product.
Every count records the variance on every line — quantity out, value out, held against a review threshold. Then the session closes and the numbers stop. Nothing turns a year of variances into an accuracy figure, a trend, or an answer to which site is drifting. The data is complete; the measurement was never built.
We publish a free economic order quantity calculator. It is indexed, it works, and it tells you how many units to order each time. If you then buy our product there is nowhere to put that number — because our reordering does a subtraction, and a subtraction is not a policy.
Our system records the result of an incoming quality check properly — a status on the consignment, a screen to set it, a summary and an analytics view built on it. What it does not do is stop you selling the goods that failed. Recording and enforcing live on two different records, and nothing connects them.
Our asset register knows a forklift on your floor might be owned, leased or borrowed, and enforces it in five places. Our stock system has no equivalent idea at all — every unit is yours by assumption. For anyone holding supplier-owned components, that assumption is the difference between what is in the building and what belongs on the balance sheet.
A till session opens with a float, records every drop, and closes with a counted figure and a variance against the person who counted it. All of it is about cash — so on a counter where most customers pay by phone, the strongest control in the shop is watching the smallest pile of money.
A generator is serviced every 250 running hours, a vehicle at 10,000 kilometres, a compressor on its duty cycle. Every one of those is a number the machine keeps and the asset register does not, so the only schedule it can offer you is a calendar.
Two people agree to trade Saturday and somebody changes a name on the roster. The roster is now correct and the trade has left no trace — not who asked, not who agreed, not who approved, and not what the shift said an hour ago.
You can attach five kinds of cost to a shipment in our system, and that is not a default you can extend — there is no screen anywhere that adds a sixth. In a region where inspection and treatment are a routine, separately-invoiced and highly variable cost of importing, the missing name is the one that would have told you which supplier is expensive to bring in.
A purchase order in our system records what you ordered, when it is due, who acknowledged it and where the truck is. It does not record the one term that decides whether the freight on it was ever yours to pay — and we will attach that freight to your stock cost without asking, because nothing in the system knows.
We can run a request for quotation and compare what came back on price. What we cannot do is score bids against weighted criteria, keep them sealed until opening, or record why the second-cheapest won — which is the part a grant-funded organization is usually required to be able to show.
Matching an order against a receipt against the supplier's invoice is the standard control over paying for goods. We do two of those three. The third is not weakly implemented — it has no data behind it at all, because what we call an invoice is a document we generated ourselves from our own prices.
Every other way stock leaves your business has somebody on the other side of it — a customer, a project, a destination branch. A write-off has nobody. It is the one exit where the only evidence is whatever the person doing it chose to attach, and attaching anything is optional.
Reorder point, safety stock and lead time all live on the item and nowhere else, and every low-stock check compares the national total against them. A distribution centre can sit at zero for a fortnight while the dashboard stays green.
A boxed machine on a rack is stock; bolted to a floor it is an asset. Our system moves it across that line properly, on approval, with a linked record — and the moment it crosses it gains three capabilities it never had and still does not gain the one buyers assume.
A high-value write-off here needs a second person, a dedicated permission and a value test that cannot be gamed by splitting it. What it does not need is a photograph, a note or any document at all — the file field is optional, everywhere, always.
The value on a purchase request is the quantity times the item's last recorded buying price. It is not a figure the requester supplied, it changes when somebody edits the catalogue, and nothing compares it to what you actually paid.
At six every morning a job finds suppliers whose qualification has lapsed, marks them expired and switches them off so they cannot be invited to quote. It is a good control, correctly built, and it tells nobody.
Every quotation, invoice and till sale carries eight fields recording the conditions it was raised under — country, currency, tax type, rate, enabled, inclusive, exchange rate and a timestamp. Change any of those settings tomorrow and the document does not move.
The customer statement runs over any date range, carries a running balance, renders as a PDF and emails without duplicating on a retry. It is built from invoices and payments. A credit note is neither — so the balance at the bottom is higher than what the customer owes.
A customer can have many named contacts with roles and many structured addresses with labels and a default. It is a proper counterparty model. It also has a second free-text address field — so there are two places an address can live.
The role editor offers five separate controls over receiving stock — view, create, approve, adjust, reject. One of them governs a screen you can open. The other four can be granted or withheld all afternoon and nothing changes, which is worse than having one control, because it looks like five.
A custom report here belongs to whoever built it. The machinery deciding who may open a shared report is complete, careful and will never be consulted, because nothing in the product can create the record it reads — and the only test asserts the columns exist.
There are four recorded reasons an address can leave an automated email sequence: the person unsubscribed, an administrator removed them, the mail bounced, or it was marked as spam. Three of the four have never been written by anything.
A ticket can only be assigned to an employee record — fine until you count how many people who actually answer tickets are not employee records. The owner, the administrator, the agency covering nights, the contractor on a six-month engagement.
A freight bill is a fixed amount of money, so the goods it brought have to carry that amount between them and no more. Ours carried it once per delivery — an order arriving in two parts booked twice the freight that was paid. We found it by writing the test that did not exist, and the error hides best exactly where freight matters most.
When part of an order never turns up, the freight you already paid has to be carried by the goods that did. Ours divided by what was ordered instead — so a shipment arriving forty per cent short left every unit you actually held costed ten per cent under what it cost you. Where the balance never follows, that error never corrects itself.
A warehouse in our system has no country, no currency, no tax registration and no legal entity. For a single-country operation that is exactly right. For a distributor holding stock either side of a border, the building has no jurisdiction and the stock in it has no home.
One cost basis, weighted average, and no way to restate the past — a correction is a new transaction, never a revaluation. In a stable currency that is a footnote. Where prices move quickly it decides what your margin appears to be.
Our ABC report bands the catalogue on cumulative consumption value, which is the correct definition and not the common one. Our counting module can hold a class against a plan. Nothing carries the first into the second, so at catalogue scale the classification is never refreshed.
Our landed-cost engine spreads a freight bill by value or by quantity. It cannot spread it by weight or volume. For a landlocked buyer whose biggest cost is a lorry charging by the tonne, those were the two bases that would have been right.
Three-way matching compares what you ordered, what arrived and what you were billed. We do the first two properly, with tolerances and a payment gate. The third document is one we generate from our own prices — so we call this a two-way match.
Five documents, four handovers. At each one something is carried forward and something is dropped, and the dropped things are invisible on the screen where they happen. This is the map.
A requisition has line rows. A request for quotation has line rows. A purchase order has line rows. A supplier's quotation has a JSON blob — the one document whose detail is not queryable, and the one you most need to compare.
An item has one selling price and it is a field. Change it and the old value is gone — no revision, no effective date, no history. The documents remember what they were sold at; the catalogue remembers only the last decision.
Apply a credit note and the invoice's amount-paid figure goes up by the credited amount. The invoice status says credited rather than paid, so the document is honest. The column is not — and anything measuring collections from it reads a credit as money received.
A refund has to undo four things: the money, the stock, the revenue and the cost of sales. For a long time this product undid two of them and the drawer still balanced every night — which is precisely why nobody found it.
Every bin location can be given a pick sequence. The field is validated, stored and displayed, and nothing in the product has ever sorted anything by it. The sixth inert field of exactly this shape, and what it tells you about reading a form as a promise.
Serial-number tracking for stock has a table, a model, three screens, two routes and a count on the dashboard. It has no code path that can create a row. The screens work perfectly and will always be empty — and the test that would have found it takes ninety seconds.
Our stock transfer has a button marked Approve. It is a synonym for Dispatch, in the web application and the API alike, and the stock leaves the source in the same instant. There is no approved-and-waiting state, because there is no approved state at all.
Stock cannot go negative here: the issue is refused, per location, before anything is written. That is correct, and it has a price — the pallet standing in your yard that nobody has booked in does not exist, and cannot be sold.
A supplier record knows who they are, how to pay them and whether they may trade with you. It knows nothing about the terms you trade on — no payment days, no currency, no delivery terms, no lead time, no tax registration.
Our quote comparison scores price, delivery time and a vendor rating, with weights that must sum to one. Two of the three are real. The third is a field nothing fills in, and when it is empty the score uses 3.5.
Our vendor scorecard reports an on-time rate. On time means the order's planned delivery date falls within seven days of the order being raised — seven, for every supplier, everywhere. It never checks when anything actually arrived.
There is exactly one mobile-money integration in this product and the supplier record is shaped around it. Everywhere that provider is not the rail, a supplier payment is something we record rather than something we operate.
There is no recurring invoice here. No schedule, no template, no next-run date. Forty monthly retainers are forty invoices somebody creates by hand — and the failure mode is not that one is wrong, it is that one is missing.
A till session records a float, every cash drop, the expected cash, the counted cash and the variance, with who opened and who closed. It is a proper reconciliation. It does not stop the cashier seeing the expected figure before they count.
There is an anomaly detector here and everything it watches is a price — below cost, under five per cent margin, priced at nothing. It refreshes every fifteen minutes. And the cost it compares against is not the cost your ledger uses.
A push notification that never arrived is not one problem but five, they need five different people to fix them, and the worst of them — a send to an audience with no registered devices — used to be written to the log as a success.
A project here has one customer, which fits most delivery work cleanly. Organisations running one programme on money from three funders need a shared cost split across all three in set proportions — a funder dimension we can add to your workspace.
For about a year our aging report summed unlike currencies into one unlabelled figure. The repair was easy; deciding what a single honest total looks like was not — and the convention we landed on has a weakness we would rather publish than hide.
Every quantity column in our stock schema is an integer, and there is no unit of measure anywhere in the product. For a business that counts cartons that is invisible and correct. For one that buys maize by the tonne and sells it by the bag, it is the difference between a stock figure and a guess.
Our demand forecast adds twenty-five per cent in March, June and December — three Gregorian months written into the code, identical for every customer in every country. In a market whose largest demand event moves eleven days earlier each year, that is not a small approximation.
Dead stock has no event to hook, so it is scanned for nightly, deduplicated, and cleared the moment the item moves. Good engineering, measured against one idle period for the whole catalogue — the same number for a filter and a gearbox.
Every allocation goes oldest-expiry-first, with no setting, no per-item policy and no override — and there is a column on the item that looks exactly like one. The behaviour is stricter than the interface suggests, which is an unusual direction to be wrong in.
A requisition carries its department as free text; a purchase order carries it as a reference. The bridge is a lookup by exact name — and when it fails the order is raised, received and paid with no department, and no budget ever hears about it.
A purchase order has no currency field. It borrows the quotation's, and where there is none it falls back to your base currency. A supplier has no currency at all — the column existed once and was removed as unused.
Your suppliers can have logins — with multi-factor authentication, a forced first password change and rate limits on every action. From that login they answer quotations, accept or reject orders and set a shipping status. Where it stops is money.
How many suppliers you can keep active is governed by your plan, and switching one on at the limit is refused outright. An ordinary commercial arrangement, and an unusual thing to write about — which is why it is better read here than met mid-migration.
An item has one selling price. There is no price list, no customer group, no tier and no discount rule anywhere in the product — so every differentiated price your team gives is a decision somebody makes on a document, every time.
A customer has a credit limit and no payment terms. The limit is enforced properly — exposure across every open invoice, a real hold, the numbers snapshotted so history cannot be rewritten. How long they may take to pay is typed onto each invoice.
You can invoice in another currency, record the rate that applied and take payment in it. You cannot keep a second set of books. The documents are right, the ledger is single-currency, and the difference lands on whoever reconciles.
Audit entries here are permanent: no administrator can remove the record of what they just did, because we took that button out. The cryptography we shipped alongside it is the interesting half; the permanence is the important one.
We could capture a signature in ten places and could not answer the question a store manager actually asks. Recording evidence and controlling for its absence are two different capabilities, and shipping the first one is the easy half — plus the one design decision that separates a usable exception report from noise.
A hand-drawn mark on a screen proves almost nothing on its own — nobody can read a signature, which is close to the definition of one. What makes it evidence is the printed name beside it, the timestamp, and one rule about when it can no longer be changed.
Stock leaves as 100 and arrives as 97, and nobody is lying — the loading clerk counted one number and the receiving clerk counted another. A transfer is the only movement with two parties who can honestly disagree about one quantity, which is why one signature structurally cannot settle it.
On a phone, a signature is a swipe — and a swipe is also how you scroll a list, dismiss a sheet and pull to refresh. Who wins that gesture decides whether you capture a signature or the first inch of one. What signing on a device actually takes, why the mark travels as text rather than as a file, and the four questions to ask any vendor.
The most common request we turn down about signatures is a pad on the approval screen. An authenticated, permission-gated approval written to the audit trail is already the stronger record — and adding a drawing on top of it disguises a real control as a ceremonial one.
A support promise measured in working hours has to know when you work in both directions — counting down, and giving time back after a pause. Ours knew in one, and the missing half put a Friday deadline at eleven at night.
Our helpdesk was built for organizations where plenty of staff have no company login — and then the breach alarm was wired to a login. The tickets nobody had picked up were the tickets nobody was told about.
A field called "notified at" was being filled in on tickets where nobody had been notified. A missing alert is a gap you can find; a false record of one closes the investigation and moves the blame to a colleague.
A support agreement makes two promises: we will get back to you, and we will fix it. We measured both and reported on both, and had an alarm on only one — the one that matters least to the person waiting.
Our helpdesk emails your customer when their request is updated. If they reply to that email, nothing happens — and the cost of that lands on your reporting rather than on your support desk.
Every market that sets a retention period is really asking whether you can put back what the report said at the time. Three things differ by market and only one is the number of years — and our own audit log turned out to expire on a two-year default while a legal hold on the record it describes did nothing to stop it. We published that, fixed it the same day, and both dates are on the page.
Donor-funded work moves the unit of account off the cost centre and onto the award. Four cost streams accumulate against a project here and they are properly filtered — but the budgeting tool that holds lines and periods is keyed to departments, and there is no link from a budget to a project anywhere in the schema.
Egypt closes 30 June, Japan and India 31 March, Ethiopia on a date that is not the first of any Gregorian month. Our ledger turns out to be immune because it never modelled a year at all — and the damage is done by four convenience buttons and a handful of defaults that all assume 1 January.
Four East African authorities run four separate schemes against four separate interfaces, and one question separates them: can the sale complete when the authority cannot be reached? We built for one answer, for one country. The consequence we had never written down is that the receipt your customer carries out of the shop does not yet carry the fiscal reference — and a POS filing that failed three times had nowhere to go, which we fixed the day we published this.
A rule about receipts can be an integration, an architecture or an accreditation, and only the first is something you can shortlist for. The third is answered by a public register in about two minutes and almost nobody evaluating from outside the country knows to look. Our own audit against one European regime, including the six days this site contradicted itself.
EAC currencies differ in unit value by roughly thirty times, so adding two of them together produces a number that looks reasonable and is wrong by an order of magnitude. Three ways a system can answer "what did we spend", only one of them honest — and a report of ours that adds Ugandan shillings to Kenyan ones and calls a budget breached on the result.
A help desk fills with personal data faster than any other module, and most of it belongs to people who will never have an account. Our public support path is built well for exactly those people — and the export and erasure tools take a user account as their argument. Also: tickets are on no retention timer and have no deletion path at all.
A quotation commits you to nothing and carries six fields describing tax and currency. A credit note reverses a taxable supply and carries none of them — no tax split, no currency, no line items. What that costs a single-country business is almost nothing; what it costs a group trading across four EAC currencies is an ambiguity on every unapplied credit.
Duty is assumed to be a property of the country. In Somalia it is a property of the port, so two containers of one product can cost different amounts to bring ashore — and a system holding one cost per item has already averaged the difference away.
On hand and available are two different numbers and most systems hold only one. Where goods sit out a mandatory hold before they can move — and where one season carries most of the year — that gap decides whether the vessel sails full.
The question buyers normally lead with — are you compliant here — is the one that tells you least in this market. Four that work better, including two where a confident vendor answer should reduce your confidence rather than increase it.
A wrong bank account bounces. A wrong phone number pays someone. Where wages leave for a wallet, in dollars, and are never cashed out, four ordinary payroll assumptions break — and the last moment to catch an error moves to before the transfer.
A rate change is arithmetic. A change of instrument is a change in what the number is — and for one costing period your warehouse holds stock valued under both rules at once.
Costs in one currency, receipts in the other, both of them cash, and no treasury function anywhere. The exposure is real, it is structural, and it is nobody's job.
For an exporter the tax balance is a receivable rather than a liability — often a large one, managed with less rigour than a debtor a tenth of its size.
You win the export contract on a Tuesday. From that Tuesday the distinction is a property of your invoice lines — and a system holding a rate per document has already lost it.
End-of-service accrues from month one and nothing asks about it until somebody resigns. With a finance team of three, nothing that fails to raise an alert gets looked at.
One shipment, two companies, two systems. Two records that were both correct when they were made — and a difference that only appears when somebody adds them together.
Most stock counts produce a new number and throw away the evidence of how the old one went wrong. Blind, valued, placed and dated — miss any one and you get a correction instead of a reason.
The approval was given at one moment and the money moves at another. Everything that happens in between belongs to the order that caused it — if anybody wrote the rate down.
Ask what the rate is and you learn a number. Ask what it is computed over and you find out whether your system can hold the answer — two of our own defects came from asking it.
A return against last year's sale is credited at last year's rate. Our credit note holds one amount and no tax split, so it cannot express that — and nothing will tell you.
Every system provisions a tax rate from a country, because almost everywhere the country is the unit. Where it is a sector instead, a correct-looking configuration is wrong for your whole business.
A register tells you a certificate expired last month. A control stops the order. Almost every organisation has the first and believes it has the second.
A free-text field with validation is not a smaller version of a coded one. Where a jurisdiction wants a value from a published list, the field either holds it or your document is not the document.
Everywhere else, voiding a document is something you do. Here it is something you ask for — and your customer can say no, or say nothing, which is worse.
Your customer master holds what you need to know about a customer. It has nowhere to hold what your customer asserts about themselves — and some documents will not clear without it.
Almost every compliance obligation belongs to finance and is performed at a desk. This one belongs to the buyer and is discharged on a loading bay — by somebody who has never been told they are part of a control.
Six questions with structural answers — a number, a yes, a name — including the one almost nobody asks: who maintains the municipal rate dataset, and on what schedule?
Ask where rounding happens, and ask to see the tax for one rate on a two-rate invoice. Five tests you can run yourself in fifteen minutes — with our own answers beside each, including the one that cannot be paid.
The list was built where the business started, and it goes out unchanged to the other islands because nothing in the transaction suggests otherwise — same currency, same language, same group. Everything else has changed.
Colombian law makes the buyer confirm two things by electronic message before a credit purchase supports any deduction: that the invoice arrived, and that the goods did. The second is a statement about the world that only a receiving process can make — which turns the most improvised record in procurement into the one the money waits on.
Colombia caps a till receipt at five UVT and Chile publishes a new unit value every single day. Both are legal limits denominated in something other than currency, so a system that stores them as an amount is not wrong yet — it is correct with an expiry date nobody wrote down, and it goes stale without any event occurring at all.
Ask what a stored tax figure knows about its destination — then ask whether the vendor would classify goods versus services for you. The right answer to the second is no, and the first question tells many readers to skip ahead.
In several Gulf markets a customer settles an invoice by handing over a cheque dated six weeks out. The invoice is closed as far as they are concerned, the money is not yours yet, and our receivables model has only two states.
Our point-of-sale table accepts three payment methods and refuses everything else at the database level. One of the three is the brand name of a Kenyan product, and it is the only payment column in the product that constrains its own values.
Ask what your tax is called, not what it costs — then ask who can add a name that is missing. Five questions, four checkable in a trial account, and the first one tells most readers to skip ahead.
The month-end spreadsheet is usually accurate. The problem is that the rules inside it are not written anywhere the system can apply again — so next year's question about this year's figure has no answer except whoever remembers.
Ask whether tax mechanism is a property of the country or of your business — then ask whether the vendor would guess it from your industry. The right answer to the second is no.
Nauru, the Federated States of Micronesia and Solomon Islands all put nothing on the invoice line, and a system configured identically for the three would be wrong in two of them. The differences are not academic — one needs a rate per location, one needs tax carried into landed cost, and one genuinely needs nothing.
The Marshall Islands passed a consumption tax in 2025 that applies to supplies made on or after 1 October 2026. Between those two dates sits every quotation, contract and recurring invoice a business raises — and almost no finance system can hold a rate and a date at the same time, including ours.
Pakistan adds a further 4% when the buyer has no registration number or is not an active taxpayer — and "active" lapses because of a return somebody else did not file. Until 2022 the clause turned only on holding a number, which a checkbox models correctly. One sentence in a Finance Act turned an attribute into a status, and no software was notified.
Bangladesh puts suppliers on one of two registers. A registered supplier charges VAT you reclaim; an enlisted one charges a turnover tax you cannot. Two quotes at the same figure are therefore not the same cost, the cheaper-looking one is usually the smaller business, and no purchasing comparison we have seen can tell them apart.
Bhutan's GST commenced on 1 January 2026 at 5%. Its own Act had said 7% since 2020, and 7% was never charged for a single day — so a researcher who goes to the primary source, reads it properly and stops there publishes a forty per cent error. Going to the statute is necessary. It is not sufficient.
Saint Lucia charges 12.5%, 10% or 7% depending on what is sold and who sells it — and "who" is settled by counting bedrooms. Five or more and you are a hotel. Four and you are not, on the same street, selling the same night. A hotel guest's folio carries three rates before breakfast.
Six independent countries share the East Caribbean dollar and charge four different rates of VAT. No conversion, no revaluation, one price list — so nothing in a finance system signals that a second market exists, and the safeguard that stops it adding unlike currencies is exactly what hides six separate tax positions.
We record how much tax was paid on a purchase and not what it means. That gap has a general shape — a stored value whose interpretation lives outside the data — and once you see it you find it in landed costs, quantities and dates.
A storekeeper scans a label and four different records legitimately carry that number. What happens next is decided by who is holding the scanner — because access is applied before ambiguity, which means the same code can be unambiguous for a storekeeper and ambiguous for their manager.
The oldest trick in attendance is one person clocking in for another, and most defences sold against it are theatre. Three independent checks, why the code on the wall is the weakest of them and still worth having, and the honest sentence about what a photograph of it achieves.
Somebody leaves on Friday. Four separate things still connect their handset to your system, they are revoked in four different places, and one of them has no button at all. The actual offboarding sequence, and why the blunt step comes first.
A storekeeper spends Thursday in a warehouse with no signal, and everything they do sits on the handset in their pocket. Which four operations work offline, and why a device reporting zero queued is worth a second look before one reporting forty.
No driver, no port to configure, no compatibility list. The till prints an 80mm receipt through the browser, which buys you any printer you like — and where you want a drawer kick or a paper-out warning, we can add direct printer integration. What to buy, and the two settings worth ten minutes on setup day.
The shopping list is the first thing anybody asks for and the last thing most software gives you. Here is the whole of it, what each line connects to, and the four common purchases — biometric clock, connected scale, badge reader, fiscal device — that are read by nothing.
Measured usage is counted from stock issued through check-outs, which is exactly right for a store-issue operation. For counter-only items we can feed POS sales into the same formula, so the reorder tile fires on what actually moved.
A handheld barcode scanner is a keyboard that presses Enter — it has no opinion about what you scanned. A phone linked to the desktop sends the barcode to be resolved instead, and gets back an item, an order, or an explicit ambiguity. What that changes at a receiving bay, and where a wedge still wins.
Our tax types are a fixed list of ten, the same ten for everybody. Three exist because Canada needed them — which proves both that the list can grow and that it only grows when a market is loud enough.
Somebody ticked tax exempt on a customer and wanted to know it had taken effect end to end. The one-sentence test that confirms a setting is wired to behaviour, and the afternoon that verifies them across your own system.
A customer pays once. The provider tells you three times, or eleven minutes late, or not at all. Four failure modes, how each is caught, and the one that leaves nothing behind — plus why the retry only runs when somebody presses it.
A customer confirms on Tuesday, you deliver Friday, you invoice Monday. Between those points the commitment lives in a message thread — because there is no document for it. What follows from that, including why the credit check cannot fire any earlier than it does.
A catalogue check compares two numbers somebody typed. It finds the item priced below cost — and misses the discount applied every afternoon, the supplier who crept eleven percent over a year, and the freight that landed three weeks after the goods. Three margin questions, three different places.
Manufacturers barcode the tinned tomatoes. Nobody barcodes the rice you repack, the uniform you assemble, or the part whose label rubbed off — and those are usually the high-margin, high-shrinkage half of the catalogue. Trade numbers versus internal codes, and what to check before you print five hundred labels.
The purchase order goes out on Tuesday. The same item is sitting four kilometres away, above what that branch needs, and has been since March. What the balancing calculation compares, why an item with a reorder point of zero produces no suggestion in either direction, and when buying is still the right answer.
A count taken with the expected figure visible is a confirmation, not a measurement — and a confirmation that agrees tells you nothing. What blind counting reveals, who is allowed to unhide the figure, and why the audit trail on looking matters more than the hiding.
A supplier sends 110 against an order for 100 and the storekeeper books in what arrived, because that is what arrived. Over-receipt is refused here by default with zero tolerance, counting every prior delivery — and short delivery is never refused, which is the more interesting half of the design.
A shop sells and a store issues, and the difference decides which movements the replenishment maths can see. Why a hospital store's reorder points work when the same building's pharmacy counter stays silent — and the ward dimension you have to choose before go-live.
The request is always the same shape: show us what happened, and show us you have not edited it since. What an evidence pack contains including the section people do not expect, exactly what a SHA-256 checksum proves, and the row caps worth knowing before you hand one over.
Four people are certain about four different pumps, and all four are remembering something real, because there are eleven and they are identical. What a custody register records that a list cannot, why "verified" is the action nobody uses, and the one thing about equipment it does not do.
A registration lapses in March and everyone keeps working, because nothing in the working day changes. It surfaces in September when an insurer asks — and the question is about every shift in between. Why the reminder window matters more than the reminder, and what a register that holds actually needs.
A manufacturer withdraws a lot number at four on a Friday. The question is asked in their unit and answered in yours — and if those were never connected at the receiving door, no amount of effort now can connect them. What a batch trace returns, and exactly where it stops.
A tenant asks why their receipt is numbered 58 when last month's was 41. It is a fair question with an answer — and until August 2026 the honest answer for five document types was that the missing numbers belonged to somebody else's business.
Owners rarely leave over a bad month. They leave over the month the statement went out on the eleventh because somebody was on leave. What can be scheduled, why the day of month stops at twenty-eight, and the three reports worth automating.
A tenant with a leaking tap is not going to be given a user account, a password policy and a role. Two ways of taking work from someone who is not a user already exist — for employees and for suppliers — and neither is pointed at your residents. What that costs, and what a portal would actually save.
Arrears chasing works because it escalates — a nudge, a call, a letter, a decision. An automated reminder runs on one organization-wide cadence, so a tenant three days late and one ninety days late get the same email on the same morning. What that one rung is worth, and the tenants it silently never reaches.
A lease promising a response "within one working day" and a system counting twenty-four wall-clock hours agree from Monday to Thursday. On Friday afternoon they part company. Two clocks, which obligation needs which, and the wait that keeps consuming your target.
A contractor bills measured work against a bill of quantities, less retention, on a certificate. A project system bills hours times a rate. Both are called invoicing and no setting turns one into the other — what each does, and how to run them alongside each other honestly.
The second identical warehouse is planned from memory, the third from the second, badly, by whoever still has the file — and every retyping loses the task that was added because something went wrong. What a template carries, and the hour of dating you must do before trusting the programme.
A weekly scaffold check runs for four months and then stops, because recurrence happens on completion rather than on a date. Nothing is overdue and nothing is flagged — the series is absent rather than late. Why that is the right default for a site, and what it costs you.
Pay from a log and the log starts being written carefully — and creatively, by the same people, for the same reason. What a payout reconciles at the moment of posting, why the direct route records a payment rather than making one, and the lock that does not cover the clock you are paying from.
Six activities are behind and everybody has a reason. Two of them probably have three weeks of float and nobody in the room knows which two. The arithmetic that answers it, and the five things it does not account for — including the undated task that silently counts as one day.
A grant carries two clocks. The money clock runs on somebody else's fiscal year and the delivery clock runs on the rain, and in East Africa they are not in phase — which is why the same programme reads as failing in one quarter and exemplary in the next while doing exactly what it planned.
An asset can be missing, or broken, or a third thing your register has no word for: present, working, and not lawfully permitted to operate. The certificate that makes a crane legal is a date on a piece of paper, and the machine gives no sign at all when it passes.
Operations registers the truck once. Your expert-comptable registers it as four things with four different lives, because under the harmonised plan you do not choose your asset classes and a significant machine is not one depreciable object. The two documents agree on the total and disagree on how many things exist — and nothing tells you, ever.
An aging bucket measures collection risk, and that is the whole of it only while the money holds still. West Africa contains both cases at once — a currency pegged at a fixed parity and two that float — so the same report, with the same columns, means two different things depending which office you are sitting in.
A transfer between your own warehouses is an administrative detail — until the two warehouses are the coast and somewhere landlocked, and the stock spends weeks in a state that appears in no warehouse's figure. The workflow handles it well. The reporting has a hole in exactly the wrong place.
Every reorder calculation assumes an order placed today arrives a fixed number of days from today. On an island resupplied by scheduled sailings, the wait depends on when you asked relative to the cut-off — so the same order, placed two days apart, can land weeks apart.
Most operational risk arrives without an appointment, so condition records are treated as housekeeping. A storm season is annual, dated and known — there is a "before", it is on the calendar, and the record you create in it is the one nobody can reconstruct afterwards.
Every cloud system is asked whether it works offline and almost every answer is useless. Here is the specific one: four operations complete with no network and sync later. Knowing precisely which four is the difference between a productive site visit and a wasted flight.
Labour records are built for two readers: the employer and the regulator. Export manufacturing adds a third with commercial leverage — the buyer — and they are not asking whether your numbers are correct. They are asking whether anyone but you could have produced them.
Addition is lossy. Store 4 + 3 + 2 as 9 and you can still charge, reconcile and audit the 9 — but the 4 is gone. Why we store the parts alongside the total, and how to spot the places worth keeping separate before you need the detail.
The card-data question worth asking a point-of-sale vendor is not how the number is encrypted — it is whether there is anywhere to put one. Our payments table has six columns and no card among them, which is the strongest position available and has a price we would rather state than let you find.
American overtime is owed above forty hours in a workweek — a fixed seven-day period the employer designates, which need not start on Monday and need not match the pay period. Our records have no workweek in them at all, and the boundary they do enforce, by locking it, is the calendar month.
A consumer privacy request does not come through a privacy console. It arrives as a support ticket from somebody who has never logged in. Our help desk can identify that person and time the reply; our export and erasure tooling is built around user accounts, which is the one thing they do not have.
A court has told member states to require a system that measures each worker's daily hours. Ours stored a 22:00 to 06:00 shift perfectly and reported its duration as empty — because clock-out is a time with no date, and six separate copies of the same arithmetic each decided an end before a start has no answer. Fixed the day this was published.
European food law asks who supplied a batch and who received it. Our schema answers both and has for months — with two caveats that decide whether the answer arrives in an hour or not at all: the tracking is off by default per item, and deleting a batch leaves its movements pointing at nothing.
When work equipment is used away from your premises, European rules say it must be accompanied by evidence its last inspection was carried out. Our hand-off record captures the custodian, the destination, the condition and the coordinates — and has nowhere to put the one fact that has to travel.
Two people run the same report a week apart and get different figures. Neither made a mistake. The four ordinary mechanisms that keep a closed month moving, why you should not stop any of them, and the three stamps that make a report reproducible without freezing your data.
Three project plans, each realistic, each assuming it has the same engineer. Why nobody ever computes the sum, why a count of open tasks is not a measure of load, and the four contention signals you already have without a capacity model.
A support desk has one door and four kinds of request come through it. Three can be answered; the fourth is work. What happens to your SLA at that boundary — and why the one pause state most desks have records the opposite of what is true.
A customer pays half up front for something you have not made yet. That money is an obligation wearing the costume of revenue — and most systems have nowhere to put it except an invoice for goods you have not delivered. What that costs, and the marker that makes it recoverable.
A customer is at the counter holding something that is not in your system. Refusing that sale is a real control — until somebody rings it as a near-identical product instead, which is the one counter failure that leaves your totals looking exactly right.
The GCC stopped sharing a working week in January 2022, and the two biggest holidays of the year are confirmed days in advance. Both land on the two settings your SLA clock reads — which is why the same performance reads as compliant or late depending on which side of a border you measure it from.
Everyone sells offline capture as the answer to a power cut, and it answers the half that was not the problem. Your till records a sale in the dark; it cannot get a card approved — and in a card-majority market that is where the evening's revenue goes.
Our public holidays repeat on the same date every year. Six of Japan's sixteen are not on a date at all — four are the nth Monday of a month and two are announced annually. The first year is correct, which is the problem.
Everyone audits the submission. The break is almost always three steps earlier, at an adjustment nobody thought of as data entry — and the reason is usually a purchase record that never held a rate.
Six facts are needed to decide whether tax on a purchase is recoverable. An amount is one of them, and it is the least discriminating. Includes a correction to two of our own market pages.
Written by a vendor that does not file your return, has no UK payroll engine and no UK reference customer. Which is a strange sales position and a useful vantage point — there is nothing left to protect except the argument.
Ireland runs five VAT rates at once. Most systems offer one default and a zero override, which covers the two commonest cases and fails silently between them — including, in ours, by charging the standard rate when you ask for a reduced one.
A table with one row per country claims each country has one rate. True for most. For the rest it is a category error — and ours does not merely inform, it writes the value into an organization's settings and never revisits it.
Four exclusions at the top rather than the bottom: no Revenue integration, no Irish payroll, no per-line VAT rate, no Irish reference customer. Then the questions worth asking everybody else.
Five exclusions at the top rather than the bottom: no VAT numbers on the invoice, no VIES check, no reverse-charge treatment, no Belastingdienst connection, no Dutch reference customer. Then the questions worth asking everybody else.
Most tax rules describe a document. A few describe the program that issues it — accredited by the authority, listed on a public register, issuing from series registered in advance. That is a different kind of requirement, and no feature comparison can answer it.
Every system numbers its invoices sequentially and almost none of them can tell you whether the sequence belongs to you. Ours could not: for five document types the counter was reconciled against a number unique across the whole platform. We published that, then fixed it the next day, and both dates are on the page.
One exclusion that is not like the others, and it goes first: we are not a certified invoicing program. Then the four that are ordinary, the check to run before you shortlist anybody, and the questions worth asking every vendor including us.
US sales tax follows the customer. Our customer record holds a latitude to seven decimal places and no state field — and the question turns out to be asked in four places in our own code and answered from the organization every time.
Both are a percentage added at the till, which is where the resemblance ends. One is collected in stages and reclaimed; the other is collected once and avoided with a document. Software built for the first models the second incorrectly rather than incompletely.
We do not do US sales tax at all, and that is the first line rather than a footnote. Then what to ask everyone else, what to ignore, and the five things that should take us off your list.
Our engine computes each Canadian levy against its own base and hands back the breakdown. The document then stores a single blended rate and discards it — so the total is right and the split, which is the only thing either return needs, is gone.
Our table holds Canada at the federal rate, which is correct. It then writes that rate into an organization's own settings and never revisits the question — so a business in a harmonised province is configured at the federal figure and nothing says otherwise.
Five exclusions at the top, one of which rules us out of an entire province. Then the questions worth asking every vendor selling into a country where a sale can carry two taxes owed to two governments.
Zero-rated, exempt and reverse-charged all print as nothing and mean different things for who accounts for the tax and what can be recovered. Our schema has two values for the three, no value for the third — and charges the standard rate if you invent one.
Software is shortlisted on whether the tax comes out right. In the Netherlands a cross-border invoice carries no tax at all, and what makes it lawful is a name, a number and a sentence on the face of the document — three things our own invoice template does not print, though the database holds all of them.
Sales tax on goods is federal in Pakistan and sales tax on services is provincial, levied by five separate authorities. So which government a figure belongs to depends on what you sold — and most systems store the amount without storing the answer.
Taiwan runs two tax mechanisms inside one statute. Most businesses deduct what they paid a supplier; financial businesses cannot. The same payment is a receivable for one and a cost for the other — and nothing about that is a percentage.
The Canary Islands are not on a lower rate of Spanish VAT. They are outside Spanish VAT entirely, paying a different tax called IGIC — and a wrong tax name produces documents that are perfectly correct about the total.
A Brazilian invoice line can carry four or five taxes at once, owed to three levels of government. Ours has one column for tax — which is a counting problem rather than a rate problem, and the reform makes the count go up before it comes down.
The hard compliance question here is procurement, not tax. Ask what the system does on the day a supplier certificate expires, and whether anybody verifies it — the plain no is the good answer.
Japan does not only say what your invoice must show. It says how the consumption tax may be rounded — once per rate band — and the yen has no minor unit. Our own system produces a tax figure of ¥423.4.
If your last return was a claim rather than a payment, one question settles most of the evaluation. We answer it about ourselves first, and the answer is not flattering.
Write down every rate you actually charge before you look at a system. If the list has more than one entry, most tax configuration screens have no shape for your business — including ours.
Ask every vendor where tax on a purchase is recorded, and what happens to a December invoice credited in January. We publish our own answers, and the first one is that the field does not exist.
When every supplier certificate renews on its own anniversary, no review interval works and the failure is silent. What a system can do about it, and the difference between a countdown and a check.
Exporters and zero-rated suppliers reclaim tax rather than remit it. That makes input tax the number the whole return turns on, and it is the number most accounting systems never record as tax.
Jamaica charges 15%, 25% and 10% under a single tax, and the 10% ends on 1 April 2027. Configuration screens ask which country you are in, which for a great many businesses is the wrong question entirely.
A field that saves without error is not by itself evidence of a working feature. The one-afternoon test that separates the two, run on our own tax rate table first — and worth running on any vendor you are evaluating.
A cascading tax charges tax on tax. A credit-invoice tax does not. Liberia switches from one to the other on 1 January 2027, and the five percentage points are the least interesting part of it.
Ask every vendor what their system does with a credit note raised today against an invoice from July 2024. We publish what ours does, which is not enough, which is what makes the question fair.
Twenty-six pay days do not fit into twelve months, and the two left over arrive on a six-monthly rhythm rather than by accident. Most finance systems store a payroll period as a calendar month and never mention it.
Almost every finance system stores one tax rate per country and no date. That is correct until the first credit note against a pre-change invoice, and then it is confidently wrong with nothing to warn you.
Ask every vendor how many payroll runs their system accepts in one calendar month, and what reads a supplier certificate's expiry date at payment time. We publish our own answers, which are one and nothing.
Three invoices below the threshold settled in one transfer are above it. Where withholding is decided at invoice entry the system is wrong in both directions at once, and the certificate that makes the credit claimable arrives later still.
Ask every vendor to compute 1,000 dinars plus FODEC plus VAT and show you the split. We publish what ours does, which is wrong, which is the only reason the question is fair.
One question separates vendors here faster than any feature list: which instrument does this country levy? We had it wrong in our own reference data until August 2026, which is exactly why the question works.
Two levies at 19% and 1% are not a levy at 20%. Where one is inside the other's base the difference is small, one-directional and permanent — and almost every finance system computes it by adding the rates.
A fixed sum per document, and a turnover tax with a floor derived from floor area. Neither is a percentage of a line, both are real obligations — how to record each one today, and the fixed-charge rate types we can add to your workspace.
A sales tax with no input credit is not a smaller VAT. It is a different animal, and treating it as recoverable builds an asset account nobody can ever collect while understating the cost of everything it should have been part of.
Withholding became an advance rather than a final tax, which turns a cost into a receivable — but only against a certificate. Where the certificates live in an inbox, the credit is unclaimable in practice and the same profit is taxed twice.
Where the consumption tax cannot be reclaimed, it has to be charged to a budget line — and a line has a ceiling that a good intention does not raise. The arithmetic is simple and it is discovered late.
Six markets, one currency, and three purchases that get bundled into one conversation. Two of the three are better bought locally, and one of them should be bought before us.
Goods in transit are on your balance sheet and in nobody's stock count. Most systems model a location you operate, so a consignment sitting in a foreign port for eleven days becomes a purchase order — which is a document, not an asset.
Freight was quoted, so freight is in the cost. Demurrage, escort charges and border formalities were not, arrive later, and get posted to expense — which quietly prices every sale made from the consignment.
What the monetary union genuinely gives you, what it does not, and why the bloc your accountant deals with and the bloc your trucks deal with are no longer the same set of countries.
Groups configure their systems for the trading subsidiaries, because that is where the stock and the customers are. Then they discover annually that the entity nobody configured is the one with the evidence obligation.
Real-time fiscalisation moves the tax authority from something you report to afterwards into something standing between you and your own invoice. That is an operations change before it is a tax change.
A market with capable local firms, a genuine fiscalisation obligation and a records problem nobody sells to. Mostly about buying the right three things from the right three places.
Six things get assessed, and every one of them is evidenced by records a business already generates. The failure is almost never that the substance is missing — it is that nobody can produce it without reconstructing the year.
An import payable in Malawi passes through six states. Every accounting system models the first and the last. The four in between are where the money and the risk actually sit.
Exposure is usually explained as a rate that moves. Where payments queue, the rate is not the variable — the interval is, and no hedge addresses an interval.
A small market with a real constraint no software can lift. Mostly about lowering your expectations to the right level and then buying well within them.
Namibian import VAT is charged on the greater of FOB plus ten per cent, or open market value. Your supplier's invoice does not contain that number — and if your landed cost is built from the invoice, neither does your system.
No customs duty, overnight road delivery, a currency at par and an invoice with no VAT on it. Everything about a South African supplier says domestic, and two things that cost money say otherwise.
A small market, a thin local bench, a serious South African vendor presence, and one statutory rule that almost nobody's system applies. How to score the shortlist, including us.
In the CEMAC zone, paying a foreign supplier is a case you prove rather than an instruction you give. Four documents must agree, four different parties create them, and nobody owns the fact that they must match.
A hundred thousand XAF is worth exactly a hundred thousand XOF, always, by arrangement. They are still two currencies under two central banks and two sets of exchange rules — how to keep them apart in your books, and how we hold each one distinctly.
Two purchases are being confused in this market. One has no published specification to build against yet, and the other you could start on this month.
The consolidated accounts are accurate, the auditor signs, and nobody in the building can say what the group is holding today. Those are not contradictory statements.
One event, two records, created independently in two systems, two currencies and two dates — with no single object anywhere that either of them is a version of. That is why tidiness never fixes it.
Written for the holding company rather than the local business — because if your warehouse is in Singapore, you have excellent local options and you should use them.
A short reporting window is an integration problem in a compact market. Across thousands of islands, where much of the selling happens beyond the signal, it becomes a capture problem instead.
Goods spend days between your own warehouses. They have left one count and not entered another, both figures are correct, and the group total is understated every single day.
One question can eliminate a vendor outright regardless of how well they demonstrate, and most buyers ask it last. It is not about features, price or the deadline.
Whether MyInvois applies to you has three separate tests. Only the first is about your own revenue, and the two nobody runs are the ones that produce the surprises.
Sales tax and service tax share an abbreviation and almost nothing else — different bases, different rates, separate registrations, and a scope that recently grew into industries with no history of indirect tax. How to set each one up separately, and keep them that way.
How to buy while the whole market is running a countdown — including how to check whether the deadline everyone is selling you actually applies to your business.
Its validity is measured in kilometres, it can expire while the goods are still moving, and it is checked by a person at a roadside. Almost every failure comes from managing it as a compliance artefact.
The deepest and cheapest software bench in the world is already in your city. Here is how to work out whether you are making the purchase it serves — and what to do if you are not.
Move your own goods between your own warehouses across a state line and you have made a taxable supply. Four obligations follow, all of them triggered by storekeepers and discovered by accountants.
In most growing businesses the only real check on a supplier invoice is that somebody has to type it in and notices things. Structured e-invoicing removes the typing. What replaces the noticing?
Every shortlist in Muscat has become a Fawtara shortlist, and that is the wrong axis. Three layers, only one of which is an open decision — plus the seven things worth scoring and the cases where we would tell you to buy locally.
Fiscal regimes differ enormously in architecture and almost not at all in what they assume about your records. Six data conditions, five of which are configuration decisions made once.
No mandate means the decision has to stand on operational evidence alone. Three honest positions including buying nothing, seven things worth scoring, and what a joint venture's attribution really costs.
Saudi Arabia put the tax authority inside the sale. Oman built a network instead — and quietly handed the buyer a reporting position on documents somebody else created. Corner four is the one with no vendor attached to it.
The tax was never in the price, which is why the recovery hurts. Four documents, created by four parties, filed under four reference schemes — and a worked example of what an ordinary failure rate costs.
A smaller market with the same obligations as a larger one. Why the concentration problem — not fraud — is the real risk, and the five boring structural features that substitute for the people you do not have.
Every neighbouring market is being sold a mandate. Qatar has no VAT and no e-invoicing regime in force — which removes the easiest reason to buy software and leaves only the good one.
An asset register answers what you own. A custody record answers who has it right now. The difference only becomes visible on the day somebody leaves the country — and then it is visible four times over.
No VAT, no e-invoicing, no quarterly reckoning to drag the records into order. Why that makes "does it refuse or does it warn" the highest-weighted question on the shortlist.
Bahrain's characteristic risk is not competitive or operational. It is a cross-border documentary burden sized for a much larger company, held together by one person who is very good at their job.
End-of-service indemnity accrues from month one, steepens after year five, and has two values depending on how the employee leaves. No filing deadline anywhere obliges you to look at it.
One authenticated door instead of a per-vendor list. How the token differs from the HMAC signature, why replay protection matters more through ordinary failure than through attack, and the honest limit on what an inbound event can do.
Sending an HTTP request on an event is a morning's work. Signing it, keying it for idempotency, retrying it and keeping the payload when it finally fails is the difference between a feature and a liability. Ask any vendor what happens after the last retry.
Contractors buy the half that argues for money and improvise the half that spends it. A worked look at where a thin margin actually goes — a variation claim lost to a notice period, and plant hired in that the business already owned.
Clearance ended the all-in-one purchase in this market and nobody updated the shortlists. Where the three seams fall, what to score vendors on, and the cases where we would tell you to buy from a Riyadh vendor instead.
Duty is suspended in a free zone, not waived. The same carton therefore has two true unit costs depending on where it ends up, and most stock systems can only hold one of them. The arithmetic, the failure mode, and how to structure it.
A Dubai entity buys and consolidates; an African operating company clears, distributes and collects. Almost every group runs these on separate systems and reconciles by email — which works right up until somebody asks what the group actually earned on a container.
Phase 1 asked you to generate a structured invoice. Phase 2 asked you to hand it over before you may issue it. That is not a bigger compliance task — it is a different architecture, and it means your system is no longer the record of its own invoice.
The UAE has more competent implementers per square kilometre than anywhere else in this corpus, so the interesting question is not who can do it. It is which of three quite different products you are actually shopping for — and where an international vendor should honestly lose.
IPRES, CSS, CNPS and a collective agreement that varies by sector. The most useful thing a foreign software vendor can tell you about Francophone payroll is where its competence stops.
Our interface is English. Yours is not. That is a staffing question inside the building and a document question outside it — and only one of the two is actually solvable by a vendor.
Bamako is three days from Dakar and the truck is not a rounding error. Corridor distribution, border custody, and why the in-transit window is the number your stock report quietly refuses to show you.
The bon de commande, bon de livraison and facture are a three-way match under different names — and a donor auditor who reads only English is a governance risk nobody budgets for.
Mobile money moved faster here than the till software did. What actually reconciles at close of day when a third of takings never touched the cash drawer — and where our honesty about integration begins.
The CFA franc removes the exchange-rate problem between Dakar, Abidjan and Bamako — and removes nothing else. Why the shared currency makes regional groups underestimate what a border still costs them.
A numbered, nine-class chart of accounts is not a preference in OHADA countries — it is the law. Our ledger has no account numbers at all. Here is why that is the right design, and exactly how the handoff should work.
The CFA franc is pegged to the euro and floats against the dollar. That single fact inverts almost every multi-currency instinct imported from Lagos, Accra or Nairobi — including which exposure is worth tracking.
Dakar buyers are usually shopping for three different things at once and calling all of them "ERP". Separate the operations system from the statutory ledger from the payroll bureau, and the shortlist writes itself.
Abidjan runs more multi-site industry than anywhere else in the CFA franc zone, and the shortlist is usually a French-language incumbent against an English-language platform. What that trade actually costs, priced honestly.
When your customer is in Europe, their audit is your operating system. Traceability that survives a client visit, consumption recorded against the job, and the difference between records that satisfy you and records that satisfy them.
A donor file in English, an approval chain that speaks Arabic or French, and a supplier invoice in neither. Governance across a language boundary, and what breaks when evidence has to be translated after the fact.
Every vendor says they are localized. Almost none are asked which layer they mean. A field guide to the seven layers of localization — and an unusually blunt account of where our own product stops.
A city of twenty-odd million where a delivery van covers a fraction of the ground it would anywhere else, and the working week ends on a different day. Density, custody and the routing of stock through a megacity.
The Egyptian pound and the Moroccan dirham are both "local currency" and almost nothing else about them is alike. Why one demands rate discipline per transaction and the other demands discipline about the euro.
Moroccan buyers are usually choosing between a French-language local incumbent and an English-language international product, and the deciding question is rarely features. Who reads the screen, who reads the contract, and who answers the phone.
South African payroll is a submission regime, not just a calculation. What we run today, the specialist filing partner worth pairing us with, and how to divide the work cleanly between the two.
In Egypt "is it localized?" has five different answers depending on which layer you mean — language, currency, tax rate, e-invoicing, working week. A layer-by-layer buyer's guide, including the one where we tell you to look elsewhere.
Egypt's electronic invoicing regime pairs a certified filing provider with your operations system. How that split works in practice, what each side owns, and what to demand from whoever handles the submission.
A published interruption schedule is a planning input, not just a hardship. What a stoppage mid-run costs you, why the material already issued is the number to protect, and the recovery record that has to survive the outage.
Nine provinces, one promotion, and a customer who checked online before driving to the store. Price consistency, long-leg transfers and the reserved-stock problem that omnichannel creates.
A VAT vendor is judged on the evidence behind the return, not the return itself. What "SARS-compliant software" should mean, what it usually means, and the reconstruction test that separates the two.
Gauteng, the Cape and KZN on one stock position, long inland legs from the port, and capture that survives a scheduled power cut. What national distribution actually demands of a system.
NPO governance is judged on reconstructability — supplier evidence, approvals that held, and funded assets you can still find three years later. Including the honest line on B-BBEE reporting.
Cocoa runs on one of the most structured supply chains in African agriculture — and everything around it does not. What operational records both ends of that spectrum need, and where a general operations system honestly stops.
Ghanaian payroll answers to a revenue authority and a multi-tier pension arrangement, and it is the one area where we recommend somebody else. What good payroll practice looks like regardless of your software.
South African buyers are usually replacing something, not starting from nothing — which makes this a harder purchase, not an easier one. How to tell a system problem from a process problem, what to ask a polished vendor, and the straight answer on SARS, POPIA and payroll.
Accra traders run retail and wholesale out of the same stock at two different prices, then wonder why margin is impossible to read. How to separate the two channels without splitting the business in half.
An audit finding is rarely about a bad purchase — it is about a purchase nobody can reconstruct. The controls that make donor-funded spending provable eighteen months later, and where software honestly stops.
Aggregating cocoa, sesame, cashew or grain means buying from hundreds of smallholders in cash, in the field, on weight and grade — then proving every kilogram to a buyer or a lender. The records that make that provable and financeable.
A vendor-honest guide to choosing operations software in Ghana — what a system must do when the cedi moves and the tax position is layered, how to test every claim in the demo, and the straight answer on GRA e-invoicing, VAT levies and payroll.
Ghana's indirect tax position is layered, not single-rate — and that layering is where software claims turn slippery. What "we handle Ghanaian VAT" should mean, what it usually means, and the questions that tell them apart.
A second shop doubles your revenue and quadruples your blind spots. What breaks when a Nigerian retailer goes from one location to several, and the controls that decide whether the third branch makes money or quietly consumes it.
Nigerian manufacturers carry a cost their bill of materials rarely admits to: the price of making their own power, and the price of an input bought at a rate that has since moved. What it takes to know a unit cost before you price it.
Nigerian payroll answers to state revenue authorities, pension administrators and more besides — and it is the one area where we say plainly that our system is not the answer today. What payroll demands, and where our boundary honestly sits.
In Lagos the distance between your warehouse and your customer is measured in hours, and stock spends most of its life in the gap. What inventory software must do where traffic, van sales and multiple warehouses hide your real position.
Procurement is where an organization's money leaves and its reputation is decided. The five gates that make spending defensible to a donor, a board or a lender — and the honest limits of what any system can enforce.
A vendor-honest guide to choosing operations software in Nigeria — what an ERP must actually do when the naira moves, how to test every claim in the demo, what to budget, and the straight answer on FIRS e-invoicing, VAT and Nigerian payroll.
Nigeria is digitizing tax administration fast, and "we handle FIRS" has become the most casually thrown-around claim in software demos. What VAT and naira operations actually require of your records, and how to tell a records claim from an integration claim.
Everyone knows which supplier is difficult; almost nobody can prove it. The 50/30/20 weighting, why acknowledgement is worth measuring, and the one definition inside the score you must check before quoting it at anybody.
The transactions that cost you money are ordinary, small and buried among thousands of correct ones. What the pricing and financial radars actually watch, why explicit rules beat a black box, and how to stop alerts becoming furniture.
Every business system has a chat box now, and almost none do what users assume on first sight. Guidance assistant versus data agent, what ours answers today, and the one question that tells you which any vendor is selling.
Every inventory system offers a forecast; few explain how the number was produced. The blended velocity, the risk grade that compares cover against lead time, the settings that change it, and the four cases where it is confidently wrong.
Most access problems are ordinary people holding permissions nobody meant to give them. Designing roles around jobs, why approval must be a separate permission from action, and the quarterly review that keeps it honest.
An audit trail only earns its keep when somebody reads it. What a usable entry records, why exports matter as much as edits, the edit-after-approval query worth running monthly, and what a trail cannot do.
Two things account for most account risk in a small business: passwords one breach away from public, and accounts belonging to people who left. Why an employee record and a user account are not the same object.
Spreadsheets are not the problem, and businesses that move because somebody told them to move badly. Five triggers that genuinely mean it is time, two that look like triggers and are not, and the six free weeks before you shortlist anything.
The dominant payment rail across much of the continent and the worst-reconciled item on most balance sheets. Five structural reasons the statement never matches, the pattern that fixes it without an integration, and which single provider we actually connect to.
Nobody can honestly publish a per-country price list. What genuinely changes the number as you cross the continent, a three-year worksheet that makes any two quotes comparable, and our own pricing stated openly rather than behind a form.
A tariff preference is worth nothing to a business that cannot quote in the buyer's currency, cost goods honestly through the border, or produce the records proving where they came from. Operational readiness, and a hard line at customs.
Three questions vendors answer with the wrong facts. "Is it cloud" is not about uptime, "does it work offline" is not about caching, and "where is our data" is not a yes-or-no. Including the twenty-minute demo script that settles the second one.
Registers fail because they record what an organization owns rather than who is holding it. Custody, verification and disposal as the three things that matter — and a precise line between an asset register and a maintenance system.
The most over-claimed category in African business software. Why the HR layer genuinely travels and the payroll calculation genuinely does not, what "configurable tax tables" means when a budget changes, and our coverage stated as one country.
Buyers use "multi-currency" to mean three unrelated things and vendors answer whichever one they can. Separating transaction, reporting and entity currency across a continent of pegs, managed floats and redenominations.
"Best" cannot be answered as a product ranking across fifty-four regulatory environments. What can be answered: the five things that genuinely change country to country, the much larger set that does not, and how to test a regional claim in a demo.
Most inventory advice written for African businesses is inventory advice with a map on the cover. Four conditions genuinely change the priority order — and they push the same three capabilities to the top of every list, in every country.
Donor-funded and commercial buyers are judged by different people and fail the same two ways: a control that warned instead of refusing, and evidence nobody can reassemble. Threshold design, supplier document expiry, and the boundary at tendering.
Grants awarded in dollars, spent in kwacha, reported in a template nobody else uses, and audited two years later by someone who was not there. The reconstruction test, the advance regime that survives it, and an honest line on mobile money.
Three currencies in one region behaving three different ways. Why the right software design is the one with no opinion about monetary policy, and the four habits that make margin visible instead of estimated.
Vendors sell "Southern Africa" as one market. It is four buying problems wearing one label — a replacement decision in South Africa, a support-depth decision in Botswana, a currency-discipline decision in Zambia, and a record-what-happened decision in Zimbabwe.
Three countries vendors treat as one bloc, and the three things that actually separate them: how the currency behaves, how far the revenue authority has gone with electronic invoicing, and how many people near you can implement what you sign.
The businesses that supply and service mines run the hardest inventory problem in the region — critical spares nobody can quantify, consumables issued without paperwork, and tools that leave with a contractor. Including where an asset register stops and maintenance software begins.
A group with staff in Nairobi, Kampala, Dar and Kigali runs one HR function and four payroll regimes — and pretending they are one is how a statutory liability sneaks in. What consolidates cleanly, what stays national, and the honest truth about which statutory payroll any system automates. Hedged.
AfCFTA and the EAC customs union are lowering the barriers to trading across African borders — but a tariff advantage is only real if your operations can follow. What cross-border trade demands of your system, and an honest line on where software stops and customs, tariffs and rules-of-origin begin. Hedged.
Four countries, four standard rates, four electronic-invoicing regimes that do not talk to each other. An orientation for a group trading across all four — what differs, what your system should handle, and why "confirm with each authority" is the only honest way to write this. Hedged; not tax advice.
The owner wants one honest answer: how did the whole group do? Getting there without a quarterly spreadsheet ritual — the monthly group pack, a currency policy that holds, and the clear line between management consolidation the system produces and statutory consolidation your auditor signs.
Sourcing across borders adds a foreign currency, a longer lead time and a customs step to every purchase. The procurement disciplines that keep control when your supplier is in another country — and the honest line on where operations software stops and a clearing agent begins.
When you buy in dollars, sell in shillings and move stock across three currencies, the rate you paid is rarely the rate in your books. The disciplines that let an East African trader state a true cost, protect a margin and reconcile across currencies.
A regional group is several legal companies that trade with, lend to and share costs with each other — and the money moving between them is where clean books go to die. How to set up multiple entities in one system, and where operations software stops and statutory consolidation begins.
Schools and colleges run real operations behind the classroom: procurement, boarding supplies, laboratory and ICT assets, transport fleets and tight budgets answerable to boards and parents. The disciplines that keep an education institution accountable.
East Africa's hotels, lodges and camps run on thin food-and-beverage margins and remote locations — from a Nairobi city hotel to a Serengeti camp to a Rwandan gorilla lodge. The disciplines that protect cost and control supply across hospitality.
A vendor-honest guide for businesses and groups operating in more than one EAC country — why one system beats a copy per country, what "one ERP across the region" actually requires, and the straight answers on tax, e-invoicing and payroll that no platform automates identically across every border.
SACCOs, VSLAs, Umurenge SACCOs and MFIs are how much of East Africa saves and borrows — and they run on member trust. The operations and governance disciplines that keep a member-owned institution accountable, alongside its core banking system.
Clinics, hospitals and health programs share a hard inventory problem: medicines that expire, cold chains that cannot break, equipment that must work, consumables that vanish. The disciplines that keep a health facility supplied, safe and accountable.
Humanitarian logistics is supply-chain management with the difficulty turned up: fragile routes, prepositioned stock, cold chains, and cargo whose late arrival has human cost. The disciplines that keep a relief supply chain accountable across East Africa's hardest environments.
A retail or distribution chain that crosses East African borders faces the same shrinkage and thin margins in every market — plus currencies, corridors and connectivity a single-country tool never planned for. What it takes to run one chain across the region.
Coffee, tea, cashew, cotton, dairy, horticulture — East Africa's cooperatives and agribusinesses share one operational spine: member trust, seasonal cash, perishable produce and export currency. The regional view of software that holds it together.
The DRC's eastern provinces trade through Rwanda, Uganda and the port of Dar es Salaam more naturally than through Kinshasa. For importers and distributors on the corridor, the disciplines that make dollarized, cross-border trade controllable.
Burundi is small, landlocked and trade-dependent, running on the franc and the corridors through Rwanda and Tanzania. For a Bujumbura SME, the disciplines that turn thin margins and a tight currency into a business you can actually see.
Burundi, South Sudan, Somalia and the DRC joined the EAC into one market — but they are frontier economies, not Nairobi. What actually matters when currency is unstable, infrastructure is thin, and the honest answer to "is it localized?" is no.
In South Sudan, Somalia and the DRC the US dollar often matters more than the local currency, and where the local unit circulates it can move fast. The disciplines that let a business or program state its true cost and reconcile when the money itself is unstable.
South Sudan and Somalia host some of the largest humanitarian operations on earth, in some of its hardest conditions. The disciplines that hold a field program together when connectivity, banking and security cannot be assumed.
Foreign currency is scarce, the birr moves, and the rate you paid is rarely the rate in your books. The disciplines that let an Ethiopian business state its true cost, protect its margin, and reconcile across currencies — instead of discovering the gap at year-end.
Ethiopia is East Africa's manufacturing ambition and its agricultural heartland at once. The disciplines that turn raw materials into finished goods without losing the cost, and that hold a cooperative together from farm to sale.
Ethiopian payroll is less about clever software and more about clean discipline: PAYE and pension handled consistently, records that reconcile, and honest expectations about what any system actually automates. A practical, hedged guide.
Addis is Ethiopia's trading engine — importers, distributors and retailers moving goods to the regions. What inventory and distribution software must do when your stock is imported, your currency moves, and your customers are a day's drive away.
Addis is one of Africa's great hubs for NGOs and development agencies — and donor-funded procurement is judged twice: by your finance team, and by an auditor who was not in the room. The controls that make both pass, across currencies.
Rwanda's hotels, lodges and tour operators run on tight margins and high standards. The operational back-office — F&B stock, procurement, cost control, assets — that protects the margin, and an honest line on where AWRA stops and a hotel PMS begins.
A vendor-honest guide to choosing operations software in Ethiopia — what an ERP must do for an Addis-and-regions business, how forex and multi-currency reorder the checklist, what to budget, and the straight answer on VAT and e-invoicing.
What VAT and Ethiopia's move toward electronic invoicing actually require of a business system — with an honest line between what operations software should do, what it should not claim, and what stays with the Ministry of Revenue and your accountant.
PAYE and RSSB contributions are where Rwandan payroll gets exacting. What a system genuinely helps with, what it does NOT automate the way Kenya's statutory payroll is, and why "confirm with RRA/RSSB" is the most honest line in this guide.
Cooperatives are central to Rwanda's economy and held to a high governance bar. The operational disciplines — member intake, transparent ledgers, procurement, assets — that keep a cooperative's trust, and an honest line on where operations software stops and core banking begins.
RRA's EBM is one of the most digital tax regimes in the region — and one of the most over-claimed lines in software demos. What EBM and 18% VAT actually require, what an operations system should and should not claim, and how to buy on evidence.
For Kigali retailers and distributors, most of the cash is on shelves and in stores — and that is where the least visibility lives. What inventory software must do here, and how to tell a real capability from a demo.
Rwanda holds public and donor-funded spending to a high bar, and the audit trail is where that bar is met or missed. What procurement software must enforce — competition, approvals, three-way matching, grant-tagged spend — and where the honest limits are.
Dar es Salaam is the gateway for a landlocked region. The operational disciplines of port-to-corridor trade: landed cost, goods in transit as a location, custody down the chain, and the fleet as a cost centre.
A vendor-honest guide to choosing operations software in Rwanda — what an ERP must actually do for a Kigali-and-upcountry business, how to test each claim, what to budget, and the straight answer on EBM, RRA and RWF.
Institutions own more than their books show — and lose more of it than they admit. What belongs in a fixed asset register, why donor-funded and member-owned equipment needs it most, and how to keep one that survives an audit.
Tanzanian payroll is less about clever software and more about clean discipline: PAYE, NSSF, SDL and WCF handled consistently, records that reconcile, and honest expectations about what any system actually automates. A practical, hedged guide.
Cashew, coffee, cotton — Tanzanian agribusiness runs on member trust, seasonal cash and produce that loses value by the hour. The operational disciplines that hold a cooperative or produce business together, from intake to payment.
Retail and wholesale in Dar run on thin margins and fast stock. What inventory and point-of-sale software must actually do for a shop counter, a wholesale depot or a multi-branch chain — and how to tell a real capability from a slick demo.
Donor-funded procurement is judged twice — by your finance team and by an auditor who was not in the room. The controls that make both pass: competitive sourcing you can prove, three-way matching, sub-grantee oversight, and a record that is a by-product of the transaction.
Coffee and tea move through Uganda in a chain of weigh-ins, grades, advances and member payments — most of it still on paper. The operational disciplines that turn a cooperative's intake season into clean, trusted records.
A vendor-honest guide to choosing operations software in Tanzania — what an ERP must actually do for a Dar-and-upcountry business, how to test each claim, what to budget in TZS, and the straight answer on TRA, EFD receipts and VAT.
The most oversold line in Tanzanian software buying, explained straight: what TRA's EFD receipting and 18% VAT actually require, what an operations system should and should not claim, and how to buy on evidence rather than a confident tone of voice.
PAYE, NSSF and Local Service Tax are where Ugandan payroll gets fiddly. What a system genuinely helps with, what it does NOT automate the way Kenya's statutory payroll is, and why "confirm with URA" is the most honest line in this guide.
SACCOs run two businesses at once: members' money, and their own operations. This is about the second one — procurement, assets, branches, payroll — and an honest line on where an operations system stops and core banking begins.
Distributors and retailers lose more to untracked stock than to almost anything else. What inventory software must do for a Kampala warehouse or an upcountry shop floor — and how to tell a real capability from a demo.
Growth in Uganda is multi-location, and multi-location runs on trust exactly until it cannot. The stock-control disciplines — per-location stock, governed transfers, branch-owned counts, consolidated reporting — that let head office see today, today.
Donor-funded procurement lives or dies on the audit trail. What procurement software must enforce for a Kampala head office running field projects — competition, approval chains, three-way matching, grant-tagged spend — and where the honest limits are.
A vendor-honest guide to choosing operations software in Uganda — what an ERP must actually do for a Kampala-and-upcountry business, how to test each claim, what to budget, and the straight answer on EFRIS, VAT and UGX.
The most oversold line in Ugandan software buying, explained straight: what URA's EFRIS and 18% VAT actually require, what an operations system should and should not claim to do, and how to buy on evidence rather than a confident tone of voice.
AWRA POS is the point-of-sale module of AWRA OpsHub — a fast cashier screen wired to inventory, cash control, multi-branch selling and shift reconciliation. What it does and how it fits the suite.
AWRA Projects & Tasks is the delivery-management module of AWRA OpsHub — projects, tasks, milestones, Gantt planning, agile sprints, time tracking and budgets. What it does and how it fits the suite.
AWRA Intelligence Insights is the analytics layer of AWRA OpsHub — signals across inventory risk, vendor reliability, revenue momentum and financial stability from data you already generate. What it does and how it fits the suite.
AWRA Helpdesk & Ticketing is the support module of AWRA OpsHub — tracked tickets, department queues, SLA timers, threaded conversations and self-service intake portals. What it does and how it fits the suite.
AWRA Accounting is the finance module of AWRA OpsHub — journals, ledgers, reconciliation, budgets, AR/AP aging and QuickBooks sync, posted from real operational events. What it does and how it fits the suite.
AWRA Sales is the revenue module of AWRA OpsHub — customers, quotations, invoices, payments and POS with real-time inventory deduction. What it does and how it fits the suite.
AWRA Assets is the asset-tracking module of AWRA OpsHub — a register with custodians, movement tracking, QR and barcode labels and audit-ready reports. What it does and how it fits the suite.
AWRA HR & People is the human-resources module of AWRA OpsHub — employees, leave, attendance and configurable statutory payroll (PAYE, NSSF, SHIF), with employees as records, not logins. What it does and how it fits the suite.
AWRA Inventory is the stock-control module of AWRA OpsHub — real-time levels across warehouses, approval-controlled transfers, counts, quality holds and reorder alerts. What it does and how it fits the suite.
AWRA Procurement is the purchasing module of AWRA OpsHub — requests, approvals, RFQs, purchase orders, three-way matching and budget control in one procure-to-pay workflow. What it does and how it fits the suite.
Small money, large risk — the place where controls are weakest and leakage most common. The simple imprest system that keeps small cash honest without smothering it.
The routine that proves your books and your bank agree — and the most reliable way to catch errors, missed transactions, and fraud early. Why frequency is everything.
When an invoice is wrong you do not edit it — you issue a credit or debit note. Which one, who issues it, and why editing the original is the mistake to avoid (including under eTIMS).
A SKU is how your system knows what an item is; a bin location is how a human finds it. Get both naming systems right and stock control becomes almost easy.
The filing system for every shilling your business touches. Get its structure right and every report writes itself; get it wrong and you fight your own numbers for years.
The checkpoint between your transactions and your statements — a list that proves the books balance. What it catches, and the serious errors it never will.
Buy the same item at different prices, then sell one — which cost do you use? The answer is an accounting choice that quietly changes your profit, stock value, and tax.
Money you already spent, sitting on a shelf, earning nothing and costing more every month. How to define dead stock, find it before it fossilizes, and clear it.
Nobody schedules the day they outgrow their books — you just do more work for worse answers. The seven signs your accounting tool has become the bottleneck, and what they mean.
The first big fork in any ERP decision, usually answered by habit. Who owns the servers, what the cost really looks like, and who carries the burden of keeping it running.
Why assets lose value on paper and the two methods most businesses use — a worked comparison, and what belongs to accounting versus tax.
Profitable businesses run out of cash all the time — here is why. What working capital is, how the cash conversion cycle measures the gap, and the three levers that shorten it.
The document that turns "the delivery arrived" into a fact the system can trust — what a GRN records, why it must be independent of the buyer, and how it anchors the three-way match.
The oldest control in the book and the one small teams quietly abandon — the four functions it splits, the dangerous combinations, and how to keep it without more headcount.
How much to order at once — the EOQ equation, the two costs it balances, a worked example, and the tidy assumptions that make it lie in the real world.
The fastest read on whether stock is an engine or an anchor — the formula, days-on-hand, sane benchmarks by sector, and the averaging trap that hides dead stock.
Three sourcing documents that get used interchangeably and should not be — what each asks, when each fits, and the cost of comparing incomparable bids.
Cash accounting records money when it moves; accrual records it when earned or owed. The choice changes what your profit means and whether your books can be trusted.
The two numbers that decide whether you run out or tie up cash — the reorder-point formula, statistical safety stock, and where each quietly drifts wrong.
Telling the system the shelf was right and the record was wrong. Done well it is an audit trail of reality; done casually it is where inventory fraud hides.
The gap between ordering and shelving stock drives almost every stockout, bloated buffer, and emergency order. What it really costs and how to tame it.
A requisition is an internal request to buy; a PO is an external commitment to a supplier. Confusing them is how organizations get bound to spend nobody approved.
Count continuously or count occasionally and trust the math between? The choice decides how fresh your numbers are, what shrinkage you can see, and how much month-end hurts.
Not every item deserves equal attention. Rank stock by the value it ties up, guard the vital few, and stop wasting control on the trivial many.
The recipe your product is measured against — what a BOM contains, single vs multi-level, the three jobs it does, and starting from zero this week.
Money spent against time elapsed — the fastest health check a grant has. The formula, the four patterns, why donors read it the way they do, and the 30-minute monthly review.
From "we need this" to "the supplier is paid" — the eight P2P steps, the control at each, and the leak that opens when any step is skipped.
The authoritative list of everything durable you own — the fields that make it useful, the register-vs-depreciation-schedule distinction, and the disciplines that keep it alive.
The annual count finds a year of errors at once and explains none of them — how cycle counting works, ABC scheduling, blind counts, and replacing the shutdown weekend.
FIFO issues the oldest stock first; FEFO issues whatever expires soonest — they sound interchangeable and are not. Where FIFO quietly fails, and what FEFO requires.
Landed cost is what a unit actually cost by the time it reached your shelf — the formula, the components that go missing, and a worked example where 40% margin turns out to be a loss.
Rwanda expects documentation by default — procurement chains, living asset registers, and donor funds where the record is a by-product of the transaction, not a scramble.
Four registrations, four currencies, one donor report — country dimensions, cross-border cost allocation, inter-office balances, and consolidation without the quarterly ritual.
The control that stops you paying for goods you never received at prices you never agreed — how the match works, tolerances, and two-way vs four-way variants.
Ugandan SMEs digitized payments a decade before operations — EFRIS pressure, the reconciliation gap, and the stock-first sequence Kampala businesses are using in 2026.
Zonal offices as real nodes, offline field capture, dual-jurisdiction records, and the fleet as a program cost — operations for programs from Dar to Kigoma.
The long chain — landed costs at the port, regional depots, in-transit as a location, van routes, and agent consignment — with the leak at every link named.
The Kampala–field gap, sub-grantee money as accountable advances, and mobile money at program scale — donor fund discipline for Uganda's NGO sector.
QuickBooks keeps the books honest — but the ledger only learns what happened after the money moved. Where accounting-only software stops, what the ERP layer adds, and why you keep the accounting tool you trust.
The airplane-mode test, sync queues and conflict rules, two-minute capture design, and the nightly heartbeat — what offline-first actually means, and which four operations survive it here.
What each fund type is, where NGOs mix them by accident — salary borrowing, shared costs, exhausted lines — and the structure that keeps the wall intact.
When the lab result fails or KEBS calls, holds freeze suspect stock in one action and the batch chain traces to customers in minutes — build it before the call.
The van is a warehouse that drives away; the site is a store with no walls — load-out transfers, nightly variance, job-costed materials, and crew custody.
How a grant budget lives through its life — approved lines, commitments, monthly burn-rate reviews, realignment deadlines, and a closeout that matches to the shilling.
Donor-funded equipment is held in trust — the register fields, custody rules, verification rhythm, and disposition reports that prove the trust was kept.
Three donors, five grants, one ledger — dimensional accounting, shared-cost allocation without double recovery, and reports in every donor's format.
The offering is the hardest money to handle: cash, collected in trust, counted by volunteers. Getting it right protects the money, the counters, and the confidence of everyone who gives.
How to operationalize order tracking, quality checks, quotation analytics, and supplier performance in one procurement command layer.
How AWRA Reports & BI Studio helps teams create governed analytics for inventory, procurement, sales, and finance.
How AWRA Workflow Automation standardizes task routing, retries, approvals, and escalation for enterprise operations teams.
Step-by-step guides and operational playbooks for inventory, procurement, sales, and accounting teams.
Key operational KPIs to track across inventory, procurement, assets, mobile field work, finance, governance, and reporting.
Industry-specific insights for retail, education, NGOs, healthcare, logistics, asset-heavy teams, and governed operations.
Data-driven scoring, AI readiness, and insight workflows for modern operations teams.
Security, governance, audit readiness, role access, mobile evidence, and compliance practices for operational teams.
How to evaluate inventory platforms that connect procurement, assets, mobile workflows, governance, and reporting.
The home market, and the only country in the statutory rule engine — eTIMS transmission, M-Pesa collection and the payroll maths are ours rather than a connector to somebody else's.
EFRIS is URA's, not ours — we hold the record it is built from, and say so before the shortlist, not after.
Corridor stock out of Dar, landed cost on the inland leg, and a straight answer on EFD fiscalization.
Tight governance expectations, cross-border stock to Goma and Bujumbura, and EBM left to an approved provider.
Birr, forex allocation and real landed cost — written for a market where the rate you got is the whole story.
Lagos distribution at scale: multi-location stock, naira volatility recorded rather than smoothed, FIRS left to FIRS.
A levy stack you maintain, MoMo reconciled rather than integrated, and procurement that refuses above a threshold.
A replacement decision, not a first system — which layer actually failed, and whether the ledger stays.
ETA e-invoicing and an Arabic right-to-left interface are both builds, not shipped features. Everything operational is.
DGI declarations stay with your practitioner; stock, procurement, assets and project cost are what we sell.
The third purchase most Dakar buyers underestimate — and an English-only interface we raise in call one.
Judge us on the depot upcountry: two-way traceability, corridor stock, and OHADA books left with your expert-comptable.
A kwacha that moves against your cost base, and margin decided at the receipt rather than on the price list.
Record what happened, hold no opinion — the transaction stored in its own currency at the rate actually applied.
The calmest currency in the region and the thinnest implementation bench, so the support answer comes first.
Dollar-pegged, and better served by local implementers than any market on this list — so we lead with the one thing they cannot see: the African end of your trade lane.
Nobody here sells you one system any more — ZATCA clearance settled that — so the only honest question is which seams you can live with, and we are unusually specific about which side of each one we sit on.
Fawtara is a five-corner network rather than a clearance gate, which quietly hands the buyer an obligation as well as the seller — so this is the one Gulf market where the honest pitch is about accounts payable, not invoicing.
The one Gulf market where nobody can sell you a deadline — no VAT, no e-invoicing mandate in force — which makes it the only place buyers get to choose a system on whether it runs the business rather than on whether it clears a document.
The smallest market in the Gulf and one of the most export-facing, which means a cross-border compliance surface run by a finance team of three — so the risk here is documentary rather than operational.
No VAT and no e-invoicing means the usual Gulf argument is unavailable here — what is distinctive instead is a workforce that is overwhelmingly expatriate, which turns payroll into a balance-sheet problem and custody into an operational one.
The one market here where the compliance software is cheaper and better than anything we would build — so this page is about the part that is not compliance: every time your own goods cross a state line, you have made a taxable supply.
The mandate everyone is selling you a countdown to — where the interesting question is not when it starts but whether it applies to you at all, because that has three different answers and only one of them is your own turnover.
A reporting window measured in days, and thousands of islands where a good part of the selling happens somewhere the network does not reach — which makes a tax rule into an operations problem.
A market where almost nobody needs an operations system for Singapore — but a great many holding companies need one for everywhere else, and the reporting, the consolidation and the decision all happen here.
Money leaves this currency zone as a documented case rather than a bank instruction — so the file behind a purchase order, not the purchase order, is what decides whether your supplier gets paid.
No customs duty from South Africa, the rand at par, and an import VAT charged on a value your supplier's invoice does not contain.
Where a payable can be approved, funded and still unpayable for months — so the exposure is the wait, and your ageing report is measuring the bank rather than you.
Where the company is resident in one country and the business happens in five others — and the records are the only thing that can tell a regulator which was which.
One currency, one chart of accounts, six tax authorities — and stock that spends a fortnight in a country you do not operate in, under a duty nobody has paid yet.
Where the tax on what you buy is a cost, and the tax withheld from you is the receivable — the opposite of both instincts.
One country, three port administrations — so the same goods from the same supplier land at two different costs.
Where five levies share no base and no moment — and our own totals add rates that Tunisia stacks.
Paid every fortnight, twenty-six times a year — against a payroll table with a unique key on the month.
Where 9% and 15% are both correct answers, and the only thing that decides between them is a date.
Five months from a tax that changes mechanism rather than rate — against two date columns of ours that nothing reads.
An invoice is a document the state signs before your customer sees it — and ours is a document we print.
The only market on this site where the compliance obligation is something the buyer has to do — and the thing it wants acknowledged is a delivery.
One tax, three rates — against a reference file whose unit is a country when Jamaica's unit is a sector.
Where the VAT return is a claim rather than a payment, and the tax you are claiming is on the side of the ledger we do not hold.
The one market here where the hard question is procurement rather than tax — and the one where the answer was already built.
One currency across six tax jurisdictions, and every one of them draws the tourism line somewhere different — including the one that abolished it.
The first market here whose hard problem is a form rather than a rate — and the one where we can count exactly how much of that form our own records can fill.
Five VAT rates running at once, and a product that gives an invoice line two options — the organization's rate, or nothing.
Destination-based tax needs a destination, and our customer record holds a latitude to seven decimal places and no state.
In some provinces a sale carries two taxes owed to two governments, and our invoice line has one column for tax.
The buyer's VAT number and the words "btw verlegd" are what make a cross-border invoice lawful here, and our invoice prints neither.
Portugal certifies the software, not just the invoice — and our product is not on the register, which is a fact about us rather than a gap on a list.
The law specifies how to round consumption tax, and the yen has no minor unit — so this is the first market where our arithmetic is the thing that fails, not our document.
An invoice line here carries several taxes at once, and ours carries one — which is a counting problem rather than a rate problem, and the transition makes the count go up before it comes down.
Three tax systems inside one country, and two of them are not VAT at all — so the question here is not what the rate is but which tax you are under, and our profile holds one name per country.
One statute runs two tax mechanisms side by side, and the difference is not a rate — it is whether tax you paid a supplier is money you get back or money you spent.
One business can owe sales tax to five different revenue authorities depending on what it sold, and our invoice line records a rate with nobody attached to it.
Grants, procurement, assets and payroll in one governed system, priced in KES.
Requisition to receipt as one auditable chain, so the donor file assembles itself.
Grant and budget-line identity captured at entry, not reconstructed at report time.
Relief and programme stock with receipts, issues, counts and reasons auditors accept.
Donor-funded equipment with named custody, grant tags and end-of-project handover.
PAYE, NSSF and SHIF handled, with programme staff cost allocated to the right grant.
Who builds this, where we are, and how a Nairobi-based team supports an operation like yours.
Approval thresholds that refuse, RFQ comparison and three-way matching before payment.
The one market where the statutory engine is ours and we keep the rates current.
Tills that move stock, shifts that reconcile, and M-Pesa matched to the sale.
Every asset with a custodian, a location, a condition and a verification history.
One price and one stock position across branches, with promotions that expire on schedule.
Fees, procurement, stores and assets in one place, with the board pack generated from it.
Batch, expiry and FEFO issuing enforced in the query — plus where an HIS should take over.
True landed cost, quality holds that hold at 2am, and traceability in both directions.
The operational half of a SACCO — and a straight answer on where a core banking system belongs.
Stock movement captured with no signal, van stock that reconciles, and jobs evidenced where they happened.
Store issues, outlet-level variance and procurement that survives a busy service.
Site stores, plant custody, subcontractor spend and cost per project as work happens.
Time, budget and recoverability per engagement, with the evidence attached.
Contributions, restricted funds and spending approvals a congregation can be shown.
Billing, arrears and maintenance today — and plainly what the property layer still lacks.
Member intake at the scale, input credit, and real cost carried into finished goods.
Nothing matches your search yet — try another keyword, topic, or page type.
From the archives
Beyond the blog
Answers to common questions about AWRA OpsHub, inventory, procurement, accounting, and integrations.
Watch quick demos to see AWRA OpsHub in action across inventory, procurement, and finance.
Monthly operations insights — no noise, unsubscribe anytime.
SubscribeNow screening
The posts explain how operations should work. These recordings show the product doing it — stock, purchase orders, approvals and the reporting on top, in real screens.
16 August 2026
Demonstration of Google API scope usage for AWRA OpsHub (awraops.com), a B2B procurement and operations platform.
Watch on YouTubeHelp Center
Run inventory, procurement, assets, sales, and field work with approved AWRA guidance for setup, migration, integrations, security, pricing, and support.
Demo booking
Share a few details so we can send the confirmation email and route your request to the right AWRA team.
Privacy choices
We use necessary cookies for secure sessions. With your permission, we also use cookies and browser storage for preferences, analytics, and demo engagement. Privacy Policy
AWRA OpsHub Cookie settings
Necessary cookies stay on for login, CSRF protection, and security. You can choose the optional categories below.
Overview
AWRA uses cookies and browser storage to keep public pages secure, remember selected preferences, measure website performance, and manage demo engagement prompts.
You can choose whether functional and marketing engagement storage apply. Analytics measurement is always active in this AWRA setup.
These settings apply to AWRA public website experiences such as the homepage, feature pages, pricing calculator, blog, help center, and request-demo page. Authenticated dashboard and vendor portal sessions still rely on required security cookies.
For more context on privacy handling, open the Privacy Policy.
Required Cookies
Required cookies and storage support basic website delivery, secure sessions, request protection, and remembering the consent choice itself.
These cannot be switched off from this manager because disabling them would break login/session behavior, form protection, or the ability to remember the privacy choice you save.
Functional Cookies
Functional storage improves the public website experience by remembering interface choices, helper states, dismissed notices, and short-lived interaction preferences.
Turning this off does not stop secure required cookies or analytics. It only limits optional convenience memory.
If disabled, AWRA may show some helper prompts again or forget non-essential display choices. Core public pages, contact forms, and request-demo forms still work.
Analytics Cookies
Analytics helps AWRA understand public page performance, traffic patterns, and content usefulness so we can improve the marketing website.
This category does not by itself enable demo popups, exit-intent prompts, or advertising pixels. Those are controlled by Marketing & engagement.
In this AWRA setup, Google Analytics and Google Tag Manager measurement are treated as mandatory website measurement and remain active.
Marketing & engagement
Marketing and engagement storage supports demo prompts, exit-intent prompts, campaign attribution, and future advertising pixels.
When disabled, AWRA will not auto-open demo or exit-intent popups. CTA buttons can still open a form because that is a direct visitor action.
When enabled, AWRA can remember that a visitor already saw, dismissed, or submitted a demo prompt so the same popup is not repeated aggressively.
Public assistant
Ask AwraIQ about features, pricing, onboarding, login, integrations, security, demos, mobile apps, automation, reports, or support.
Try asking
