Scandit Self-Checkout Loss Prevention is a retail loss prevention software that uses real-time vision AI to detect loss at self-checkout, accidental or intentional, as it happens.
It recovers or deters more than 75% of self-checkout losses through soft nudges, associate alerts, and per-store loss visibility.
Detection is item-based only: no facial recognition, no biometric identification.
It runs on a per-station unit, works with existing cameras, and needs no in-store server or purpose-built AI hardware.
A shopper waves a cereal box past the scanner. There’s no beep or scan, and the event goes unnoticed. The busy shopper misses it while scanning a basket full of items, and the one associate keeping an eye on multiple lanes does too. A small, unintentional incident. But repeated across many checkout transactions every day, across multiple stores, it becomes one of the costliest and hardest problems to solve in retail today.
Self-checkout moved scanning out of the cashier's hands and into the shopper's. Checkout got faster and more flexible for shoppers, but loss rose for retailers. The question every retailer now faces is how to win that loss back without slowing the lane, adding more resources, or further stretching store associates, and without treating honest shoppers as suspects. That is the problem we built our Self-Checkout Loss Prevention Solution to solve.
Here is a quick overview of how it works, and why we designed it the way we did.
Why self-checkout created a new kind of loss
Self-checkout is still seeing continued growth amongst many retailers, especially grocers, who have a 96% adoption rate.
According to a 2026 ECR research report on retail loss, self-checkout handles 54% of transactions in the stores that offer it. As scanning shifted to the shopper, a new category of loss opened up. The same report highlights that self-checkout loss has climbed to 28.5% of all retail loss, up from under 10% in 2018.
Retailers are in a bind. On one hand, they need to enable customers to quickly, frictionlessly, and conveniently complete transactions at the self-checkout stations.
On the other hand, they need to find solutions to protect their revenue and mitigate against these losses - whether it’s staffing checkout lanes, using weight scales to check that items have been scanned, or gates that require the shoppers to scan a receipt to exit. But these solutions typically add friction, expense, and detract from the customer experience - effectively breaking the self-service promise.
Typical loss patterns account for the bulk of lost revenue
Most of that loss comes from three patterns: items missed at the scanner, items left in the basket or trolley, and walkaways where a shopper leaves mid-transaction. Together they account for 97% of loss. Walkaways are the most expensive, averaging €88 each, while missed scans are simply the most common.
When it comes to missed scans, many are honest mistakes. That is why we talk about loss, which covers both intentional and unintentional incidents.
A single unscanned item in a full basket, three lanes over, is almost impossible to see by eye. That is the heart of the problem: most missed scans simply are not observable by an associate, who is there to help shoppers, not to watch every lane. The answer has to be real-time detection and correction, and solving it that way recovers far more revenue with far less friction.
Why loss is hard to spot and prevent
When we started building out our solution, three problems stood out.
The first is the variety of loss patterns. Loss shows up in many ways: a missed scan, a barcode swap, an item slipped into a bag, or items left in the basket, across a huge range of products. Detection has to hold up in busy real stores without ever flagging a shopper's own belongings.
The second is privacy, which is a significant priority for retailers looking to stay compliant, reassure their customers, and avoid creating the wrong kind of headlines. A camera pointed at a checkout sits squarely within GDPR, and, in markets with strong worker representation, within the reach of works councils. A system that identifies shoppers in order to do its job creates stringent compliance requirements that can be as challenging to tackle as the loss issues it’s trying to solve.
The third is deployment. Retailers run large numbers of self-checkout lanes across mixed hardware and store formats. A solution that needs months of custom setup per site, or a rack of dedicated equipment, rarely makes it past a pilot.
The solution works with existing camera infrastructure - whether built-in, off-the-shelf, or standard security cameras above the self-checkout station - eliminating the need for purpose-built AI hardware, which also avoids single-vendor supply-chain limits.
Powered by vision AI, our solution watches the scan session and correlates what it sees with the live data coming from the scanner and the point-of-sale (POS) system.
It reads item movement, the path an object takes through the scan zone and bagging area, and correlates it with the scan events. It does not use facial recognition, build or store a biometric template, or recognize a shopper from one visit to the next.
The solution’s entire job is to answer specific questions related to typical loss patterns: did the item that just moved get scanned? Was an item left in the basket? Did the shopper walk away without paying?
Depending on the answer, the system will then flag the event. Processing runs locally at the lane on a per-station unit, so the analysis happens where the transaction happens. That per-station unit makes deployment easy and scalable, and supports any store format, from a 2-lane store to a hypermarket.
Most missed scans are honest mistakes, so the shopper simply gets a gentle on-screen nudge to rescan the item, with overhead footage showing what was missed. ECR's research finds shoppers self-correct in 80 to 97% of cases once prompted, with no associate involvement, and the lane keeps flowing
If a shopper leaves without completing payment (a walkaway), the system escalates. An alert goes to an associate, who can see what happened and decide how to act. Every step, from a nudge alone to an associate alert to a video-backed review (where a clip is shown to the associate), is configured to the retailer's own policy.
The combined effect is meaningful. When deploying our Self-Checkout Loss Prevention Solution, more than 75% of self-checkout losses can be recovered or deterred through soft nudges, associate alerts, and per-store visibility, with the majority resolved before anyone has to step in, freeing associates to focus on customers rather than routine checks.
Privacy by design
Privacy is critical in loss prevention, and that’s why we very intentionally built the solution with it in mind from the very start. It is baked into how the system works. Because detection is item-based, there is no facial or biometric data to protect in the first place.
Video is processed on the station, and the design is built to support retailers in meeting their GDPR obligations, backed by Scandit's ISO 27001:2022 certification. Because the system identifies no one, it avoids by design the practices the EU AI Act prohibits, such as biometric categorization and emotion recognition. We build to the Act's requirements and give retailers the materials to support their own transparency obligations.
Where footage is used to improve the solution, it is masked first, and anything that cannot be reliably masked is deleted rather than kept. We have written about this in more depth in a dedicated piece on privacy by design at self-checkout, for teams who want the full picture.
A lot of loss prevention is slow and costly to roll out. Physical fixes like weight scales or receipt gates need new hardware at every lane, and alternative camera-based systems often depend on dedicated AI cameras or a proprietary in-store server that is costly to buy, house, and maintain.
Scandit needs none of that: detection runs on the per-station unit already described, with no server and no purpose-built AI hardware to buy into.
Together they address the two moments that most often create friction at the lane and pull an associate back to the self-checkout lanes: an age-restricted item, and typical loss patterns like missed scans. And both are powered by vision AI.
Age Verified Self-Checkout enables retailers to automate and verify self-checkout purchases of age-restricted items to reduce lane delays, remove avoidable associate intervention, and improve compliance. On-device vision AI age estimation and ID scanning let shoppers confirm their age via their smartphone selfie, meaning 80% of age checks are resolved in seconds with no associate intervention needed.
With both solutions combined, Scandit provides a more autonomous self-checkout model that supports age verification compliance, protects against loss, and frees up associates for higher-value work instead of routine checks.
The bottom line
Retailers should not have to choose between stopping loss and keeping self-checkout fast and welcoming.
Our Self-Checkout Loss Prevention Solution flags the patterns behind most self-checkout loss as they happen, lets shoppers put things right themselves, and does it without identifying anyone.
See how it works on the solution page, or talk to an expert about running it on your own lanes.
Self-checkout loss prevention is retail loss prevention technology that detects loss at self-checkout lanes as it happens, then prompts the shopper or alerts an associate to correct it. Scandit Self-Checkout Loss Prevention uses real-time vision AI to spot the three patterns behind 97% of self-checkout loss: missed scans, items left in the basket, and walkaways - and recovers or deters more than 75% of it. Detection is item-based, so no shopper is identified.
Vision AI watches the scan session through a camera above the lane and correlates what it sees with the scanner and point-of-sale data, flagging any item that moves through the checkout without a valid scan. Scandit's solution answers simple questions in real time: did the item get scanned, was one left in the basket, did the shopper leave without paying? Processing runs locally at the lane on a per-station unit, so detection happens where the transaction happens.
No. Scandit Self-Checkout Loss Prevention uses item-based detection only, with no facial recognition and no biometric identification of shoppers. Video is processed on the station to support retailers' GDPR obligations, backed by ISO 27001:2022 certification, and because the system identifies no one, it avoids by design the practices the EU AI Act prohibits.
No dedicated server is needed. Scandit runs detection on a per-station unit that works with existing cameras, whether built-in, off-the-shelf, or security cameras above the lane, with no purpose-built AI hardware. That keeps total cost of ownership low and lets it scale across any store format, from a 2-lane store to a hypermarket.
Retailers can recover or deter more than 75% of self-checkout losses with Scandit Self-Checkout Loss Prevention, through soft on-screen nudges, associate alerts, and per-store loss visibility.* Most cases resolve before an associate has to step in: once nudged, shoppers self-correct in 80 to 97% of cases (ECR, 2026).
:format(jpeg))
:format(jpeg))
:format(jpeg))
