Charging Costs to the Right Job: Procurement, Stock, Time & Expenses
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.
There is a number in almost every Kenyan organization that runs projects, and it is called overhead. Some of it is genuinely overhead — rent, insurance, the accounts team. But a large share of it is project cost that arrived without a job attached, got absorbed because it had to go somewhere, and now makes every individual project look better than it was while the business as a whole looks worse than it should.
This is the most common reason a portfolio of profitable-looking projects adds up to a disappointing year. Nothing is being stolen and nobody is incompetent. Costs simply arrive by four different routes, and only some of those routes ask which job they belong to.
The four routes cost reaches a project
Every shilling a project consumes arrives one of four ways. The discipline is identical each time — name the job at the point of entry — but each route fails differently.
-
Procurement — bought specifically for the job
A requisition or purchase order for this project. Easiest to get right and the one most often captured too late: if the project is only named on the invoice, you have lost the commitment for weeks.
-
Stock — issued from your own store
Material you already owned, leaving the store for a site or a job. This is the leak that feels controlled: the issue IS recorded, just not against a project. "Issued to the site" is untraceable the moment there are two sites.
-
Staff time — the largest and least visible
People are usually the biggest project cost and the one that lands nowhere by default, because salary is a monthly total. Time against tasks, costed to the project, is what makes a project cost real.
-
Direct expenses — the small ones that add up
Travel to site, per diems, transport, printing, subsistence. Individually trivial, collectively significant, and almost always coded to a general expense account because the claim form never asked which project.
A cost without a job is not an accounting problem. It is a decision you will make wrongly later, with confidence, because the number you based it on was flattering.
The one-field rule
Almost all of this is solved by a single rule, applied without exceptions: nothing gets committed, issued or claimed without naming a project or being explicitly marked as overhead.
The important half of that sentence is the second half. If "no project" is simply allowed, everything ambiguous drifts there. If choosing overhead is a deliberate, visible selection, ambiguity becomes a question someone answers at the moment they still know the answer. It is the same field either way; what changes is that silence is no longer an option.
Project field optional
- Anything ambiguous goes unassigned, because that is the path of least resistance
- Overhead grows steadily and nobody can decompose it
- Every project reports a margin over its direct costs only
- The business result never reconciles to the sum of its projects
- Allocation happens at month-end, by someone guessing from descriptions
Project field required
- Ambiguity becomes a question at the moment someone knows the answer
- Overhead is what somebody deliberately chose to call overhead
- Project margin includes materials, time and expenses
- Project results roll up to something recognisable as the business
- Allocation is a by-product of normal work, not a monthly exercise
Expect a week of friction
Requiring a project on every issue, order and claim is mildly annoying for about a week and then becomes invisible. Every organization we have watched do this reported the same two things: brief irritation, followed by the first genuinely accurate job costing they had ever seen — and usually one uncomfortable discovery about a client or a type of work they had assumed was profitable.
Shared costs, honestly handled
Some costs genuinely serve several projects. A vehicle covering three sites in a day. A site supervisor across two jobs. A bulk purchase drawn down by four projects over a month. Pretending these belong to one job is as wrong as leaving them unassigned.
Two honest approaches. Split at entry where the split is knowable — a delivery covering two sites is two lines. Or allocate by a stated driver where it is not: supervisor time by days on each site, vehicle cost by trips, bulk stock by quantity drawn. What matters is that the driver is written down and applied consistently, because an allocation rule that changes when the results are unwelcome is not a rule.
For donor-funded work this is a compliance matter rather than a management preference — shared-cost allocation methods usually need documenting and defending to an auditor. The practice is set out in multi-donor fund accounting and, for cross-office costs, in multi-country NGO operations.
The test that tells you whether it works
Pick a finished project and add up everything charged to it: procurement, stock, time, expenses and allocated shared cost. Compare that against what the project earned. Then do the same for every project in a period and compare the total against the business result.
If the sum of your project results looks nothing like your business result once genuine overhead is added, your allocation is leaking — and the gap tells you how much. Most organizations doing this exercise for the first time find the gap is large and sits mostly in staff time. That is uncomfortable, and it is the most valuable number the exercise produces.
What we do and do not do
What AWRA OpsHub does today
- Procurement tied to a project at requisition and order, so commitment is visible before the invoice.
- Stock issues booked to a project, so material leaves the store against a job.
- Time entries costed to the project, whether billable or not.
- Expense claims coded to a project, so travel and per diems land on the job.
- Budget vs committed vs actual per line, on one view, while the job runs.
- Grant and budget-line tagging where the funding source matters.
What it does not do
- We do not invent allocation rules for you. Drivers for shared costs are yours to define and defend.
- We do not compute statutory overhead absorption for financial-reporting purposes — that is your accountant's treatment.
- We do not do BQ measurement or take-off on construction work.
- We cannot fix a project structure that does not reflect how you work — if the project list is wrong, allocation to it will be too.
How costs should be treated for statutory financial reporting is your accountant's call and can differ from the management view described here. Confirm the treatment with them.
Where to start
Do the two easy routes first. Making the project field required on purchase orders and stock issues is a configuration change with a week of grumbling attached, and it typically recovers the majority of the invisible cost on its own. Time comes next, because it is the biggest but needs a behaviour change. Expenses last, because they are the smallest.
Then read the results weekly rather than monthly — project budget vs actual covers the reporting rhythm, and project management software in Kenya the module as a whole.
Our take
Make the project field mandatory on stock issues and purchase orders this month, with overhead as a deliberate choice rather than a default. It is the cheapest change on this list and it usually reveals that one client, one service line or one site was never profitable. That finding is worth more than a year of better estimating.
See every cost land on the right job
Procurement, stock issues, staff time and expense claims all coded to the project that caused them — with budget, committed and actual on one view.
Explore AWRA ProjectsFrequently asked questions
What is the single biggest source of unallocated project cost?
Staff time, by a wide margin. Salaries arrive as a monthly total and land nowhere near a project unless time is captured against tasks and costed, so the largest cost on most jobs is missing from all of them. This is why organizations find that individually profitable projects add up to a business that is not — each project shows a margin over its direct costs while the entire salary bill sits outside every one of them.
How should we handle a cost that serves several projects?
Either split it at entry where the split is knowable — a delivery covering two sites becomes two lines — or allocate it by a stated driver where it is not, such as supervisor time by days per site or vehicle cost by trips. The critical part is that the driver is written down and applied consistently, because an allocation method that changes when the results are unwelcome is not a method. For donor-funded work, expect to document and defend it to an auditor.
Should we force a project on every transaction?
Yes, with overhead as an explicit, visible option rather than a blank. If leaving it empty is permitted, everything ambiguous drifts there and overhead becomes an undecomposable lump. Making overhead a deliberate selection turns ambiguity into a question answered at the moment somebody still knows the answer, which costs seconds and saves a month-end guessing exercise.
How do we know whether our allocation is actually working?
Add up every project result for a period, add genuine overhead, and compare the total against your business result. If they look nothing like each other, cost is leaking out of your projects and the size of the gap tells you how much. Most organizations running this test for the first time find a large gap concentrated in staff time — which is uncomfortable and is precisely the number that makes the case for fixing it.
Our project list does not really match how we work. Does that matter?
It matters more than the allocation mechanics. If projects are defined at the wrong granularity — one project covering work you actually manage as three, or thirty projects for one engagement — then costs will be allocated accurately to a structure that cannot answer your questions. Fix the structure first, at whatever level you actually make decisions and win work, then allocate to it.
