Building the business case for stocktaking automation.

What to measure before you start

A business case built on assumed numbers gets picked apart in the first review. Before you write anything, pin down four figures for your own site: what the current stocktake costs in labour, including any agency staff brought in to hit the date, and how much time is spent searching for pallets that are not where the system says they are. Add how often stock accuracy causes a real operational problem, a short shipment, a delayed order, or a stopped line, and what your current accuracy figure actually is, measured, not assumed. Most sites have never had to state that last one precisely. Get a real number before anyone puts a solution in front of you, because "we think it is roughly right" does not survive a finance review.

If you do not have a clean baseline, the causes worth checking for first are covered in causes of stock discrepancies, and a fuller framework for what accuracy actually measures is in the inventory accuracy guide.

The numbers finance will challenge

Three lines get pushed back on more than any others. Expect it, and have an answer ready.

• "Labour efficiencies released" is soft. Any figure for time saved that has not yet been redeployed into something measurable will be treated as aspirational, correctly. Keep it separate from the hard savings in your headline number, and state clearly what you plan to do with the released hours.
• Downtime avoided is hard to attribute cleanly. If your case leans on avoided production stoppages, finance will ask how you know the stoppage was caused by a stock discrepancy rather than something else. Only use this line where the causal link is already documented, not inferred.
• "Current cost" is often understated. Most sites quote only the visible labour line and miss the lost pallet search, the write-offs and the agency spend booked elsewhere. Under-stating your current cost makes the case look weaker than it is, so build the full picture before you present it, not just the obvious line. See what a manual stocktake actually costs for how the full figure is usually built.

The payback arithmetic

Work the arithmetic in the direction finance actually checks it: from the cost of the solution back to what the current process must be costing for the payback to hold. On Sentispec's published Site tier, €9,950 setup plus one quarter of the €39,550 annual fee is €19,838 of cost in the first three months. For a three month payback to be genuine, manual stocktaking at that site has to be costing about €79,000 a year, roughly €6,600 a month, before the project starts.

That is the test to apply to any payback claim you are handed, including your own: state the assumption behind it explicitly, and check your own numbers against it rather than assuming the headline applies to your site. If your current cost is below that threshold, your payback will run longer than three months, so say so in the paper rather than let the review discover it. Real sites range widely: across the ten deployments we publish, break even has landed anywhere from one month to six months or more, and what drove the difference at each site is published on the benchmark page.

What to put in front of a CFO

A CFO wants three things and does not want a fourth. They want the current cost, stated conservatively and with its basis shown. They want the new cost, including that nothing has to be capitalised, since the recording unit is leased inside the annual fee rather than purchased. And they want the payback period stated as a range, not a single confident number, with the assumption behind it visible. What they do not want is the vendor's efficiency story leading the case. Lead with the hard labour and search cost numbers, and let the softer productivity upside sit underneath as a secondary point, not the headline.

What to put in front of an operations director

An operations director cares less about the payback period and more about whether the count will actually work on their floor without disrupting the shift pattern they run. Lead with what a run actually involves: a device picked up by a forklift the site already operates, driven at normal travel speed, with no changes to racking and no new labels. Show how long deployment takes and what it demands of their team during it, and be specific about what the count does and does not verify, since an operations director will spot an overclaim faster than a CFO will. Bring the discrepancy types the system actually reports, not a generic accuracy percentage, because that is the language their team will be working in day to day.

Questions to ask any vendor, including Sentispec

Whichever vendor you are evaluating, these questions separate a real case from a pitch.

• What exactly is included in the published price, and what is not? Ask specifically whether hardware, support, upgrades and training are all inside the headline figure, or whether any of them are quoted separately later.
• What does deployment actually require from our team, and for how long? Get a day by day account, not a total. A vague "a few weeks" answer usually means nobody has run it enough times to know.
• What does the accuracy figure measure, exactly? Ask what counts as a match, whether empty locations are included, and whether the figure is quoted after go live on a working site or from a best case demonstration.
• Can we see the evidence behind a claimed result, not just the headline? A vendor confident in its numbers will show a real discrepancy list or a real recording, not only a percentage on a slide.
• Where does this method not work well? A vendor who cannot describe the conditions their method depends on is worth pressing harder. For a method that rides on a forklift, the dependency is a driver and a forklift for the length of a run, both of which most sites already have.

The comparison between automated and manual counting is on automated stocktaking against manual counting, and the published price behind the arithmetic above is on the pricing page.