“I check the shelf, then I fix the screen,” an associate in a mid-sized grocery store told researchers running a seven-country survey — and that habit explains why stock records match the shelf only 46 percent of the time, as of November 13, 2026. The survey, run by Sapio Research for Scandit among 2,200 store associates at retailers with more than 1,000 employees in the US, UK, Canada, France, Germany, Italy and Poland, landed in the middle of peak planning, when a wrong record costs a sale rather than a line in a report. For anyone accountable for distribution coverage across a store estate, it reframes availability as a data problem first and a stocking problem second.
Where the working hours actually go
The cost is absorbed by the shift rather than by a budget line. Associates who took part estimate they could reclaim more than 4.4 hours a week, nearly six working weeks a year, if the record matched the shelf; for a retailer with 10,000 store associates the survey puts the equivalent labour at more than 41 million dollars a year. None of that appears as an inventory cost. It appears as slower service, shorter opening hours on the tasks nobody measures, and a promise at checkout that the shelf cannot keep.
Two smaller numbers describe the same loss from the aisle side. Associates spend an average of 4.1 hours a week answering customer questions that store technology should be able to answer, and are asked to physically confirm whether an item is in stock 5.85 times a week, with more than one in five doing it more than ten times. That is the shape of a system that is trusted enough to be consulted and not enough to be believed.
Why half-right data is the expensive kind
Trust arrives late, not gradually
The survey's most useful finding is a threshold rather than an average. Trust barely moves while accuracy sits under 50 percent, and it rises sharply once accuracy clears 70 percent, which means a chain can improve from 46 to 60 percent and see no change in behaviour on the floor. Below the threshold, staff do the sensible thing: they walk the aisle, correct the record, or simply help the customer in front of them. The national average therefore hides two very different operating models inside the same estate.
What a hand-written tally changes
When the record and the shelf disagree, 42 percent of the associates who take action bypass the official system entirely, and 68 percent use at least one informal method: paper notes, team chats, phone photos, a personal list. The result is a shadow store that head office cannot see and a forecasting model cannot read, and it is precisely the layer that automated replenishment and online availability promises are being built on top of.
Report the aisle, not the country
| Indicator | What the survey reports | Why it matters operationally |
|---|---|---|
| Record matches the shelf | 46 percent of the time | Half of all lookups need a physical check |
| Associates who fully trust the data | 8 percent | Trust is rebuilt by hand, not by the system |
| Hours each associate could reclaim | More than 4.4 a week | About six working weeks a year per person |
| Annual labour equivalent, 10,000 associates | More than 41 million dollars | The gap is already a payroll cost |
| Associates who bypass the official system | 42 percent of those who act | The shadow store never reaches a dashboard |
| UK associates who fully trust the data | 3 percent | Confidence varies sharply by market |
National availability figures hide the only number that matters operationally, which is availability inside the trade area a given store serves. Two stores with the same national score can behave completely differently: one serves a dense residential catchment where a missing pack size loses the basket, the other an office district with a lunchtime spike where the same gap is invisible by two in the afternoon. Splitting the metric by store cluster is what turns it into a replenishment instruction rather than a monthly grade.
Cluster reporting in practice
A workable routine fits on one page. Report availability weekly by cluster rather than monthly by country. Separate a true stockout from a listing gap, because a product in the building but not sellable online is a batch update, not an order. Measure replenishment cycle time from shelf-empty detection to shelf-full, and tie delivery promises to store-level inventory so the checkout page never offers what the shelf cannot support. None of this needs a new fulfilment network; it needs one version of the truth shared by store, warehouse and delivery.
What to do before the next peak
The four weeks around peak widen the gap fastest, because promotional plans, service capacity and fulfilment capacity all have to move together. A campaign that drives demand a store cannot serve converts a marketing win into a refund and a review, and the record that would have caught it is the one an associate is currently keeping on paper. Auditing a sample of clusters before the calendar turns is cheaper than explaining afterwards why the promise was made.
- Pick five clusters and measure match rate weekly, not monthly.
- Log every manual correction for two weeks and count the repeat offenders by SKU.
- Check whether label printing and point-of-sale updates share a release step.
- Give click and collect a promise the store can keep, then compare collection conversion.
FAQ
Is on-shelf availability the same as in-stock rate?
No. In-stock asks whether the store holds the item; availability asks whether a shopper can buy it, which includes the listing, the label and the price file.
Why does accuracy below 70 percent change so little?
Because staff cannot tell which records to trust, so they verify everything. Above the threshold they start relying on the system instead of the aisle.
Does buying more scanners close the gap?
Not on its own. Participation in pilots matters more: associates who helped test the tools reported using them at 74 percent against 61 percent for those never asked.
Where should a small chain start?
With one cluster and one week of manual corrections. That produces a defensible list of exceptions without buying anything.
