UK frozen food grocer, Iceland, has partnered with SAI, the leading active intelligence solution for stores, to significantly reduce shrink while honing operational effectiveness across its UK store estate.While the ONS reported earlier this month that instances of shoplifting had declined by -4% year-on-year, levels of in-store theft continue to remain high across the sector, with over half a million (507,086) incidents reported last year. In 2025, Iceland estimated that theft cost its business around £20 million in losses annually, with its Executive Chair, Richard Walker, saying that shoplifting was not just impacting its margins, but was also negatively affecting its ability to keep prices low and reinvest in staff wages.
Operating over 1,000 UK stores, including 700 High Street locations, Iceland wanted to reduce shoplifting without adding unnecessary friction, excessive labour or extra operational burden across its bricks-and-mortar estate. Having previously trialled other loss prevention (LP) solutions, such as weighted shelves, Iceland wanted to move towards a more proactive solution to tackle theft, as its Head of Loss Prevention & Corporate Audit, Andy Edwards, explained:
“We wanted to become the second choice on the High Street to steal from, and in order to do that, we needed to be more proactive. We needed a solution that could work alongside us, rather than behind us.”
Using its patented Visual Language Model (VLM), the SAI One platform blends computer vision with GenAI to turn traditionally ‘passive’ store surveillance systems into an active, operational platform for real-time store intelligence. Proprietary AI innovation within the platform analyses live data from video feeds using the advanced VLM to build a comprehensive, contextual picture of how stores actually operate. This then surfaces meaningful operational signals, insight cues and timely staff alerts that drive estate-wide performance.
Deploying SAI One to overlay multi-modal visual AI onto its existing camera infrastructure, Iceland was able to derive real-time intelligence across self- and assisted-checkouts, aisles and entrances, delivering high-accuracy detection across shrink and operational issues.
At the checkouts, the solution uses Vision AI to detect missed scans, partial payment and walkouts, while in-aisle it targets and detects concealment, consumption in store and shelf swipe. High-fidelity real-time alerts generated by the platform are sent out to colleagues’ handheld terminals (HTTs) within 8 seconds of suspicious activity being confirmed, with imagery and footage of the individual and location within the store helping staff respond more quickly and effectively.
Since rollout, the solution has helped deliver an 80% reduction in loss, while reducing ‘false’ alerts by 50%. And, with fewer false alerts, staff confidence and adoption of the system has prompted a 90% alert response rate. It has also helped free colleagues to focus their time on operational execution and CX delivery, rather than manning false alarms.
“We’ve been delighted with the progress that we’ve made, and the journey that we’ve been on. SAI is playing a huge part in [improving] our worst 100 stores, which cost us about a third of our total shrink number,” Edwards added. “The platform is not just helping us reduce shrink and making Iceland one of the hardest places to steal from on the High Street, it’s also helping colleagues on the shopfloor from a productivity perspective. It’s really accelerated us to being at the forefront of AI use in loss prevention.”
Beyond reducing shrink, the SAI One platform has helped Iceland improve colleague confidence, supported safer and more informed interventions and helped store teams spend more time running the business and focusing on customers, rather than become obsessed with surveillance.
“Iceland shares our belief that AI should make stores smarter, safer and easier to run. Together, we’re proving that retailers can reduce shrink, empower colleagues and improve operational performance, all from a single platform,” said Som Sinha, Founder and CEO at SAI. “We’re excited to continue building on these results and helping shape the next generation of intelligent retail operations at Iceland.”
Looking ahead, Iceland is exploring how it uses the SAI solution to further enhance its LP efforts, including Revisit Alerts, which identify and warn store staff about previous offenders returning to stores and Nearby Store Broadcast Alerts that use the SAI One platform to send warnings and information to other locations.
The retailer is also exploring how it uses SAI’s in-aisle theft data to move into responsible Facial Recognition Technology (FRT) to keep proven repeat offenders out of stores as well as its wider crime prevention, incident reporting and case management activities.
SAI is working with leading retailers to redefine how FRT is used responsibly, integrating it as a controlled, evidence-backed component within a broader Vision AI platform, rather than as a standalone surveillance tool. SAI’s charter for responsible FRT use includes guardrails and commitments that: guarantee no images of innocent shoppers remain on FRT databases and are removed in real-time (taking no longer than one second); give customers the right to review evidence at the point they are being denied access to a store; and guide sanctions based on the severity of previous offences.
