Chapter 1: The data foundation
System vs the shelf, they match less than half the time.
:format(webp))
Retail Research Report
The Store Truth Gap: The Hidden Cost Behind Retail’s AI Plans
Before you lock your AI budget, see what 2,200 store associates know about the three versions of your stores: The one your system reports, the one on the shelf, and the one your customers get when the two don’t match.
- associates surveyed
- 2,200
- countries
- 7
- retail sectors
- 6
- retail experts consulted
- 5
About this research
Every store exists in three versions
Retailers are about to lock their biggest Al budgets yet, and it hinges on one assumption: that the store data feeding it reflects what is actually in the store.
Associates see all three versions, every shift. They are the most complete source of store intelligence truth a retailer has, and most have never been asked.
So in September 2026, Scandit and Sapio Research surveyed 2,200 of them, across seven markets in North America and Europe and six retail sectors.
1
What the system says
The version head office sees. Associates say it matches the shelf less than half the time.
2
What is on the shelf
Associates see it every shift. When the system is wrong, two in three keep unofficial records.
3
What the customer experiences when the two don't match
Head office often sees this only when it becomes a complaint.
Explore Chapters
:format(webp))
:format(webp))
:format(webp))
:format(webp))
:format(webp))
The Store Truth Gap by the numbers
The headline numbers on the state of store data, and what they mean for retailers investing in AI for store operations to drive resilience, efficiency and revenue.
4.4 hrs
4.4 hours a week back
Associates would get back 4.4 hours a week if stock data were always accurate, nearly six working weeks a year. For a US associate at the survey’s average $18 an hour, that is worth around $4,100 a year, more than $41 million for every 10,000 associates.
8%
Only 8% trust the data
Less than 1 in 10 associates trust their store’s stock data completely. Trust is a threshold, not a slope: among associates who say the data is accurate half the time or less, one in four trust it, and that barely moves until accuracy passes 70%.
2 in 3
2 in 3 use unofficial channels to record data
When the system is wrong, two in three associates track stock through at least one unofficial channel: paper notes, group chats, phone photos, a notes app or memory. Four in ten don’t use the official system for it at all.
9 in 10
9 in 10 notice theft every week
9 in 10 associates notice suspected theft in a typical week. One in seven of all associates say they carry on with their task and take no action, so the event never reaches the record.
:format(webp))
Every retailer I speak to is investing in AI across their store estate. Very few can say with confidence that the data they’re building it on reflects what is actually happening on the shelf and across the entire store. Store associates see that gap every shift, so we asked them about it, at scale. Their answers point to a store truth problem more than an AI problem. The returns retailers expect from AI depend on a shared, accurate picture of the store, and a frontline that is set up to act on it.
Samuel Mueller, CEO and co-founder, Scandit
Free companion tool
Take the readiness assessment
The research explores where the store truth gap sits across retail. This companion assessment, created by Andrew Smith, Managing Partner and co-founder of ThinkUncommon, shows where it sits in your business.
The assessment tests three things: purpose, people, and process.
Does your store technology have a clear objective?
Are your stores and people ready to test and deploy it?
Can you get value from it once it’s live?
You get a readiness state for each, and the one move to start with.