Latest / Internet of Things with Fexingo: Connected Devices, Sensors, and Industrial IoT / How IoT Sensors Are Reshaping Retail Inventory Management
Transcript
- Lucas: So, we talk a lot about IoT sensors in factories and warehouses — predictive maintenance, air quality monitoring, all that. But there's a use case that's quietly become one of the biggest deployments in the consumer world, and it's happening in grocery stores. Luna: You mean the smart shelves? I've seen those — the ones that detect when a product is picked up or when stock is low. Lucas: Exactly. But let's get specific. Tesco, the UK grocery chain, has been rolling out a combination of RFID tags and weight-sensing shelves across its stores. They started with a pilot in 2023 in about 50 stores, and by early 2026, they've covered over 800 locations. Luna: That's massive. And the results? I heard something about cutting out-of-stock incidents significantly. Lucas: They reported a 40% reduction in out-of-stock situations for tagged items. And here's the thing — for a grocery store, an empty shelf isn't just a lost sale; it's a customer who might go to a competitor. Industry estimates put the cost of a single out-of-stock event at around 4% of total store sales in lost revenue. Luna: And the sensors also cut down on labor, right? Lucas: Yes. Manual stock-check labor dropped by 60% in stores with full deployment. Employees who used to spend hours walking aisles with handheld scanners now focus on customer service or restocking based on real-time alerts from the shelf. Luna: So what's the tech stack here? Because I imagine it's not just one sensor. Lucas: Right. It's layered. The base layer is passive UHF RFID tags — those thin stickers on product packaging. They cost about 3 pence each, which is roughly 4 cents. Then you have weight-sensing shelves that measure the mass of products on them. The shelf sends data to an edge gateway — a small computer under the shelf that processes the weight change and cross-references it with the RFID tag read. Luna: So the edge gateway decides if a product was picked up or if it's just someone leaning on the shelf? Lucas: Exactly. That's the clever part. The edge gateway runs a lightweight algorithm that filters out false positives. If the weight changes by less than a threshold — say, 50 grams — it ignores it. But if it changes by the weight of a specific product, and the RFID tag leaves the shelf's read zone, it registers a removal. Luna: And then that data goes to the cloud for analytics? Lucas: Yes. The edge gateway sends aggregated data — not every single event, but summaries every few minutes — to a cloud platform. That platform ties into Tesco's inventory management system and their mobile app. So if you're a Tesco app user, you can see real-time stock levels for items in your local store. Luna: That's a nice customer-facing feature. But I wonder about the ROI. 3 pence per tag adds up fast when you're tagging millions of individual items. Lucas: It does. For Tesco, the tags are only used on high-value or high-turnover items — things like meat, cheese, electronics, and seasonal promotions. That's about 15% of SKUs. For those, the savings from reduced out-of-stocks and labor more than cover the tag cost. They estimate a payback period of about 18 months. Luna: So not every item gets tagged. That makes sense. But what about the weight sensor shelves themselves? Those must be expensive to install. Lucas: They are. Each shelf module costs about £200 to retrofit, and a store might have a few hundred modules. But Tesco phased it in — they started with the most problematic aisles, like dairy and chilled foods, where out-of-stocks were highest. That gave them a quick win and funded the expansion. Luna: I've also heard concerns about privacy. If shelves know exactly what you're picking up, does that cross a line? Lucas: That's a fair question. Tesco says the system doesn't tie product picks to individual customers unless they use a loyalty card at checkout. The shelf data is anonymous — it just logs that a specific RFID tag left the shelf. But critics point out that in theory, if you combine shelf data with camera feeds, you could build a detailed picture of a shopper's behavior. Luna: And Tesco does have cameras in stores for security. So is there any integration? Lucas: Officially, no. Tesco states that the shelf data and camera systems are separate and not linked. But privacy advocates argue that the data could be combined later, especially as AI analytics improve. It's a tension that will only grow as these systems become more common. Luna: Let's shift gears a bit — what about smaller retailers? Can a local corner store benefit from this, or is it only for giants like Tesco? Lucas: It's getting more accessible. There are startups offering 'IoT-in-a-box' solutions — preconfigured sensor kits that you plug in and connect to a cloud dashboard. For example, a company called ShelfWise offers a system with weight sensors and a cellular gateway for about $500 per shelf, with a monthly subscription of $50. For a small store with 10 shelves, that's $5,000 upfront and $500 a month. Luna: That's still significant for a small business, but maybe doable if they see a return. What kind of ROI do they see? Lucas: Pilot studies show that small retailers typically see a 20% reduction in out-of-stocks and a 15% reduction in spoilage — especially for perishables. That can translate to thousands of dollars in saved revenue per month. So payback can be under a year. Luna: Right. And there's also the labor angle — small stores often have the owner doing everything, so automating stock checks frees them up for other tasks. Lucas: Exactly. And we're seeing the technology evolve. Next-generation shelves are experimenting with computer vision — a tiny camera embedded in the shelf that recognizes products by their packaging, not just RFID. That could eliminate the need for tagging altogether. Luna: But then privacy concerns get even bigger — a camera watching every item you pick up. Lucas: For sure. That's why companies deploying vision-based systems are emphasizing on-device processing. The camera runs a neural network that only extracts product IDs and quantities, not images of customers. The raw video never leaves the shelf. Luna: So it's edge AI for privacy. That's a smart engineering decision. Lucas: Absolutely. And it's a pattern we're seeing across IoT: process as much data as possible at the edge to reduce bandwidth, latency, and privacy risk. Luna: Before we wrap, I want to touch on something I find fascinating — dynamic pricing. If a shelf knows that milk is expiring in two days, could it automatically lower the price? Lucas: It's already happening. Some stores use IoT shelves to trigger markdowns on near-expiry items. The shelf sends an alert to a manager's tablet, who applies a discount via a mobile app. Some pilot programs have automated that, with the shelf sending a signal to electronic shelf labels to update the price in real time. Luna: That's brilliant for reducing food waste. And it's a direct value proposition that customers can see. Lucas: Right. The environmental angle is strong too. Tesco reported that their smart shelves helped reduce food waste by 12% in the first year across pilot stores. Luna: So the use case is clearly expanding. Where do you see this going in the next couple of years? Lucas: I think we'll see a convergence of RFID, weight sensors, and computer vision into a single sensor module. The cost will keep dropping. Within five years, smart shelves might be as standard as barcode scanners. And the data from these shelves will feed into predictive analytics that not only tell you what's out of stock but when it will run out — and automatically reorder. Luna: That's the holy grail: a truly autonomous supply chain. But we're not there yet. Lucas: No, but the building blocks are in place. And companies like Tesco are proving it works at scale.