Retail loss prevention worked. Now the threat looks different.
If you've read any retail technology article in the last three years, you've seen the number: $112 billion.
It's a real figure that did useful work. It's total industry shrink for fiscal year 2022—the gap between the inventory retailers' books showed and what they actually had, spanning theft, fraud, error, damage, and vendor loss. That gap ran 1.6% of retail sales, or $112.1 billion, and about two-thirds traced to theft, with external theft roughly a third on its own. The rest was process and error. It came from NRF's National Retail Security Survey, last published in 2023.
Four years on, better data exists—and it tells a more interesting story.
What the 2026 numbers show
NRF's The Impact of Retail Theft & Violence 2026, with the Loss Prevention Research Council, found retailers reported a 12.4% decrease in shoplifting incidents and an 8.1% decline in merchandise theft in 2025 versus 2024. NRF credits the stabilization to exactly what you'd hope: sustained investment in technology and employee training.
That's the headline worth carrying into every retail conversation this year: the last investment cycle worked.
The same report maps where the next one goes. Phone scams rose for 69% of respondents, loyalty fraud for 51%, and gift card theft for 42%. Half reported more repeat offenders, 40% more ORC incidents, and 37% more walkout theft.
Retailers didn't just contain a threat. They redrew the map, and the next set of gains sits on the other side of it. Crews that concentrated on grab-and-go are diversifying into the transaction, the return, the loyalty account, and the gift card balance. Hardened entrances did their job. The next layer looks different.
The perimeter moved from the door to the transaction
For twenty years, loss prevention was organized around a physical perimeter: watch the entrance, the aisle, and the exit. The assumption was that loss meant walking out holding something.
Increasingly, loss comes from a transaction that looked legitimate—a fraudulent return against a manipulated receipt, a gift card drained before activation, a self-checkout basket where three of eleven items were never scanned.
Detecting that calls for a different architecture: one that knows what should have happened at the point of sale and compares it against what did. Video, transaction, and inventory data all looking at the same event, at the same time.
Walking the store
At the entrance. A clear theme out of NRF 2026 was that vision AI is stepping into an active role—validating transactions and flagging loss in real time, at the edge, without friction for the shopper. i-PRO brings imaging built for that analytic workload, and March Networks has spent years pairing video with transaction data so a suspicious event surfaces from what happened at the register rather than from scrubbing footage.
Algo IP endpoints, speakers, and visual alerters carry the response side, where deterrence that de-escalates is gaining ground. 63% of retailers report less than half of theft incidents to law enforcement, most often because losses fall below felony thresholds. A well-timed cue prevents more loss, and puts fewer associates at risk, than an intervention at the door.
In the aisle. Exception reporting is only as good as the inventory truth behind it. Brady and Unitech scanning and mobile computing are that ground truth layer. Real-time, item-level accuracy was one of the loudest signals out of NRF 2026, described as the foundation for profitability and store execution. Havis is where detection quality is won: camera angle, scanner position, and a workable mount determine how much usable data reaches the analytics above.
At the lane. Checkout is where the new threat profile concentrates—and where retailers already have the most instrumentation, making it the fastest place to get a return. Elo Touch Solutions, HP, Toshiba, Touch Dynamic are all in the business of the lane. The reframe is that the POS isn't only where loss happens; it's the richest sensor in the building, knowing what was scanned, when, by whom, at what price, with what override. The receipt is the audit artifact at the center of return fraud, putting Star Micronics at the heart of the report's fastest-growing category.
Behind the wall. Real-time validation, edge analytics, and always-on visibility share one dependency: the store has to be up. Eaton power protection is the enabling layer, moving from back-office line item to a precondition for return on everything else—closets built for a POS server and a switch are now asked to support real compute and heat.
What this changes for partners
Lead with the current data. "Your shoplifting exposure is down 12.4%, and here's where the risk moved" lands differently than a four-year-old loss figure, and puts you alongside the retailer's results rather than against them.
Sell the connection, not the box. Any one of video, POS, or inventory alone is a product. Together they're a system, and systems are where partners earn their margin.
The mid-market just opened up. Integrated, AI-driven loss prevention used to be enterprise-only. The operator with thirty to two hundred stores can now deploy what the national chains proved out.
The point
The last investment cycle delivered. A 12.4% decline in shoplifting incidents is what a return on store technology looks like at scale. That same cycle moved the threat toward the return counter, the loyalty account, the gift card rack, and the phone line.
The 2026 data shows where the next round of decisions should go—a far better map than the industry had in 2022.