ROI on Inventory Software, Measured Not Modelled
Most ROI cases for inventory software are built on a percentage somebody made up. Four things that are genuinely measurable, how to get a before-figure for each, and the honest reason two of them will be smaller than any vendor told you.
Vendor ROI calculators work by asking you to accept a percentage. "Reduce stock holding by 20%." "Cut stockouts by 35%." The percentages are not invented from nothing — they are drawn from somewhere real — but they are applied to your business without any knowledge of it, and the output is a number with your currency symbol on it and no evidential connection to you.
The alternative is slower and defensible: pick four things you can measure, get a real before-figure for each, and accept that the case will look less impressive and be true.
The four measurable effects
| Effect | How to get the before-figure | Honest size |
|---|---|---|
| Cash released from dead stock | Your first proper count, valued. Anything with no movement in twelve months. | Large, and one-off. Usually the biggest single number and it does not recur. |
| Shrinkage found and stopped | Variance on the first count versus the second count three months later. | Moderate and recurring. The number people most overstate. |
| Time recovered | Time the daily statement reading, the stock queries, the month-end reconciliation. Actually time them. | Moderate and reliable. Boring, easy to defend. |
| Stockouts avoided | Count them for a month before go-live. Every time a customer wanted something you did not have. | Small at first, grows. Depends on reorder points being tuned, which takes 60 days. |
[Dead stock](/glossary/dead-stock) is the big one and it only pays once
Almost every Kenyan SME store with more than a few hundred lines is holding between 8% and 20% of its inventory value in things that have not moved in a year. Nobody knows this before the first count, because dead stock is defined by the absence of movement and an absence does not appear in any record you were keeping.
Finding it is genuinely valuable and the value has a specific shape: it is a one-off release of cash, not a recurring saving, and it is only realised if you actually do something — discount it, return it, write it off, stop reordering it. A dead-stock report nobody acts on has released nothing.
A 1,200-line store, first year, measured rather than modelled
Against a first-year all-in cost of roughly KES 260,000 to 460,000, that is a real return — and note what makes it real: three of the four numbers came from counting something rather than from a percentage. The fourth was left blank on purpose, because a made-up figure in an otherwise honest case discredits the whole thing.
The two numbers to be sceptical about
Shrinkage reduction and stockout reduction are the two figures vendors lean on hardest, and both are smaller in practice than in the model.
-
Shrinkage: the first count finds it, the system does not stop it
A count reveals the gap. What closes it is receiving against orders, adjustments needing a reason and a second approver, and somebody reading the exception report weekly. All of those are habits the system enables and none of them is automatic. Businesses that install software and change no habits see shrinkage return within two quarters.
-
And some of it was never shrinkage
A meaningful share of your first count variance is bad record-keeping rather than loss — units received and never entered, issues recorded twice, a unit-of-measure confusion between cartons and pieces. That part vanishes on the first count and cannot vanish again, so counting it as a recurring saving overstates year two badly.
-
Stockouts need tuned reorder points, and tuning needs data
Reorder points typed in at configuration are guesses. They become useful at around day sixty, when there is consumption history to check them against. So the stockout benefit is real and it starts in month three, which matters if you promised it in month one.
-
And reorder point here is typed, not derived
There is no consumption-velocity calculation that sets it for you. It is a number a person maintains, informed by reports. That is a genuine limitation and it means the stockout benefit depends on somebody doing the tuning — which is part of the two hours a month.
Software finds the money. Habits keep it. A business that buys the first and skips the second gets one good quarter and a renewal it resents.
Building a case somebody will believe
Five weeks before you ask for the budget
- Time three things. How long the daily payment-matching takes, how long a stock query takes, how long month end takes. Write down the minutes, with dates.
- Count the stockouts. A tally sheet at the counter for one month. Every time somebody wanted something you did not have. This is the single most persuasive number in any inventory case and almost nobody has it.
- Count one shelf properly. Not the whole store — one high-value shelf, blind, against whatever record exists. The variance on that shelf is your evidence that the rest is worth counting.
- Price the dead stock you already suspect. Every storekeeper can point at the slow-moving corner. Value it. That number is usually enough on its own.
- Then present four measured numbers and one blank. The blank is more credible than a guess, and whoever approves budgets in your organisation has seen enough vendor calculators to know the difference.
One thing the system will not compute for you
There is an inventory turnover and days-on-hand report, and a dead-stock report, and an ABC ranking by consumption value. What there is not is a cash-conversion-cycle figure, a DSO or a DPO — the inputs all exist and the three ratios are not calculated. If your business case rests on working capital days, you will be computing that from exports rather than opening it.
Our take
Build the case on dead stock, count variance and timed hours, and leave stockouts blank until you have counted them for a month. Be explicit that the dead-stock number is one-off and that a share of your first count variance was bad records rather than loss — the case survives that honesty easily, and it is the version that will not be thrown back at you in year two.
Find the money first, then decide
Dead stock and ageing, turnover and days-on-hand, ABC ranking by consumption value, and blind counts with variance recorded line by line — the four things a defensible case is built from.
See plans & pricingFrequently asked questions
How do we calculate ROI on inventory software?
From four measurable effects rather than a vendor percentage: cash released from dead stock, shrinkage found and stopped, time recovered from tasks you have actually timed, and stockouts avoided. Get a real before-figure for each — count a shelf, time the daily reconciliation, tally stockouts for a month — and accept that the resulting case looks less impressive and is true.
What is the biggest single return?
Dead stock, almost always. Most stores above a few hundred lines hold 8–20% of inventory value in items that have not moved in a year, and nobody knows before the first count because dead stock is defined by an absence of movement. It is a one-off release of cash rather than a recurring saving, and only if you act on it — discount, return, stop reordering.
Why should we be sceptical about shrinkage savings?
Two reasons. First, a count reveals the gap and the system does not close it — what closes it is receiving against orders, adjustments needing a reason and a second approver, and somebody reading the exception report weekly. Second, a meaningful share of first-count variance is bad record-keeping rather than loss, and that part vanishes once and cannot vanish again, so treating it as recurring overstates year two badly.
When do stockout reductions actually start?
Around month three. Reorder points typed in at configuration are guesses; they become useful at about day sixty when there is consumption history to check them against. Reorder point here is a number a person maintains rather than one derived from consumption velocity, so the benefit depends on somebody doing that tuning.
What should we measure before buying?
Time the daily payment-matching, a typical stock query and the month-end close, with dates. Tally every stockout at the counter for one month — this is the most persuasive number in any inventory case and almost nobody has it. Count one high-value shelf blind against whatever record exists. And value the slow-moving corner your storekeeper can already point at.
