The Self-Checkout Text Nobody's Talking About

You fired the cashier to cut costs, but self-checkout quietly opened a new door for thieves. Shrink runs 30 to 60% higher at the busiest unstaffed lanes, and almost nobody's watching. This week, the camera-based AI that finally catches it mid-scan, straight from the man behind the research. Hi, I'm Claire, your host. Thanks for joining this week's Retail Reality Check, a podcast on the latest tech installs in retail and consumer product goods. We provide hard data so you can make smart decisions based on IHL Group's research. Speaking of gaps nobody's watching, if you want to go deeper on inventory visibility, we have a webinar replay ready for you right now on demand. It's called Spend Smart: How Smart Technology and AI Are Closing the Inventory Visibility Gap. Retail research experts from IHL and Epicor break down what AI readiness actually looks like for the mid-range specialty retailer, and why the fix starts with fundamentals, not another dashboard. Watch it anytime, the link is in the description below. Let's start with the theme running through basically every story this week. Checkout used to be the place retailers competed to disappear. This week, it became the place they compete to see. Scandit just launched a self-checkout loss prevention tool, Real-Time Vision AI, that uses the cameras already sitting above your self-checkout lanes to flag missed scans, skipped items, carts full of stuff nobody paid for, and abandoned transactions. On-screen prompts nudge the shopper to fix it themselves. And if that doesn't work, it escalates to an attendant. Scandit says this recovers or deters more than 75% of self-checkout losses. Here's why that number matters. IHL's own shrink research found that stores with heavy self-checkout use see shrink rates 30 to 60% higher than staffed lanes. Retailers call it the SCO tax. But when you deploy AI specifically to catch scan avoidance and ticket switching in real time, the research shows 374% ROI over 3 years and roughly $88,000 in prevented loss per store every year, with payback landing inside 6 months for most deployments. Diebold Nixdorf is showing the same instinct on the ordering side. They just modernized the kiosks at Dave& Buster's. Multi-payment capability, Powercard recharge built right in, hooked into the existing point of sale through an open API. Wait times dropped, throughput went up, and the rollout hit 99.8% uptime. That's the kind of operational proof point that turns a pilot into a chain-wide rollout. And it's not just the big chains. Eagle Stop, Good 2 Go, and Moto Mart, 3 convenience store operators, are all adding self-checkout lanes Eagle Stop is building it into store remodels in Missouri. Good To Go is putting in 3 lanes per store. Moto Mart is being the most cautious of the 3, piloting at only about 10% of its 84 locations. And honestly, given that 30 to 60% shrink number we just talked about, that caution looks smart. You want to see what happens to your loss numbers before you roll something out chain-wide. Same pattern— AI paying for itself with data you already have shows up somewhere you might not expect. Williams-Sonoma. Their AI assistant Olive, tied into their loyalty program The Key, is now driving 9 times the revenue on a personalized visit compared to an average one. IHL's research backs up why that number is realistic and not a fluke. Retailers using AI and machine learning saw 16% higher sales growth and 134% 4% higher profit growth in 2024, and 56% of retailers are already pointing GenAI specifically at personalization. The gap isn't the technology, it's whether you have clean first-party data to feed it. Last one for the deep dive. Tapestry, the parent company of Coach and Kate Spade, now lets you buy directly inside a Google Gemini conversation. See a Coach bag while you're chatting with Gemini, hit buy, check out with Google Pay, done. That's built on something called the Universal Commerce Protocol, a bet on composable, headless commerce architecture. IHL's data says that bet is a smart one. Retailers already using AI and machine learning in their demand chain operations are 81 times more likely to have already adopted headless commerce. So Tapestry isn't just chasing a shiny new checkout button. They're sitting on the architecture that makes it possible in the first place. Same story, 5 aisles of retail. The technology to fix the leak, whether that's shrink, slow checkout, generic personalization, or a clunky purchase flow, already exists and is already proven. The question isn't whether it works, it's who deploys it before their competitor does. Now let's move fast through the rest of this week's headlines in our lightning round. Canadian retailers are getting squeezed by SKU proliferation— more colors, sizes, and marketplace-only listings than their systems were built to track. IHL Group research pegs that mismatch— inventory distortion worldwide— at $1.7 trillion a year. Walmart is rolling out tap-to-pay at Walmart and Sam's Club stores by year-end, opening up to Apple Pay, Google Pay, Samsung Pay, and contactless cards. Amazon, meanwhile, is pouring billions into AI infrastructure, including a Qualcomm deal worth up to $60 billion in data center chips. Wilson, the tennis and sporting goods brand, runs its entire global supply chain, from raw materials to the racket in a pro player's bag, on SAP's connected technology stack. And they're preparing for a major SAP S/4HANA transformation starting in 2027. Instacart launched an AI shopping assistant called Clementine, building recipes and carts from real-time inventory data across nearly 2,200 retail banners. Shipt answered with Ask Shipt, which builds a customized cart from a text prompt or a photo in seconds. Coca-Cola's CokeBuddy ordering platform uses an AI feature called Perfect Basket to recommend what small retailers should stock, And how much? Across 39,000 outlets in Malaysia, 83% of retailers went with the AI's recommendation. Costco just expanded its Uber Eats delivery partnership from 17 states to 47, putting nearly 600 warehouse locations on the platform for members. Walmart is adding Papa John's to its Express Delivery lineup, the first national restaurant brand fulfilled entirely from locations outside Walmart stores. And customers can bundle a pizza order right in with their groceries. Wakefern Food Corp., the cooperative behind ShopRite and Fairway Market, added more than 375 Northeast supermarkets to Uber Eats, with discounts up to 30% off a customer's first order. Rugs Direct rolled out a conversational AI shopping assistant called Ask Cleo, built with Constructor. And shoppers who use it are converting to add to cart at 8 to 9 times the rate of everyone else. Skechers is building one of Europe's largest automated footwear logistics centers, a 230,000 square meter facility in Belgium with NAP technology, more than 3 million storage locations capable of moving 10 million pairs of shoes a month. Philippine 7 Corporation, the operator of 7-Eleven in the Philippines, is replacing spreadsheet-based planning with Relex forecasting and replenishment technology across 27 distribution centers. And a 10-store restaurant brand was losing $20,000 a month even while hitting labor targets because hours were spread evenly instead of matching actual traffic. AI scheduling from SynergySuite found the gap and unlocked $20,000 to $35,000 a month just by moving the same hours around. So here's the question I'll leave you with. Every story we covered today— shrink prevention, kiosks, personalization, the AI checkout button— none of it is experimental anymore. It's live, measured, and paying back inside a year in most cases. The real question is whether you're the retailer deploying it or the one reading about a competitor who already did. If you want to go deeper, head over to ihlservices.com. There are more than 40 additional stories from this week alone, plus free white papers on everything we talked about today: shrink, checkout AI, and personalization. Thanks for listening to Retail Reality Check. We'll see you back here next week.