AI-powered shelf intelligence could help retailers recover lost sales, improve execution… And creates a new battle over retail data
Instacart (NASDAQ: CART) is expanding its artificial intelligence capabilities in grocery retail with the acquisition of Arpalus, a computer vision company focused on real-time shelf intelligence, the company announced July 16.
The San Francisco-based grocery technology company said the acquisition will strengthen its ability to help retailers improve inventory accuracy, enhance ecommerce fulfillment and gain greater visibility into product availability inside stores.
Instacart, which supports more than 1,500 retail banners and nearly 100,000 stores through its platform, is positioning shelf intelligence as another tool in its expanding suite of grocery technology solutions. Terms of the transaction were not disclosed.
For decades, grocery retailers have had a detailed understanding of the final step of the shopping journey: the transaction. They know what consumers purchased, when they purchased it and how frequently they return.
What retailers have known far less about is everything that happens before checkout: whether a product was actually available, whether it was placed correctly, whether a promotion was executed properly and how many potential sales were lost because a shopper never found what they were looking for.
Grocery’s Next Data Battle
That information gap is becoming grocery’s next major technology opportunity.
Instacart’s acquisition of Arpalus, a company specializing in AI-powered shelf intelligence, represents the latest move by a major grocery technology provider to bring artificial intelligence deeper into store operations. The technology uses computer vision to analyze shelf conditions in real time, helping retailers identify out-of-stocks, improve inventory accuracy and better understand what is happening inside the physical store.
The significance of the deal extends beyond Instacart. It reflects a broader industry shift: the grocery shelf itself is becoming a valuable source of operational data.
A typical 41,650-square-foot supermarket carries more than 33,000 unique items, creating a massive challenge for retailers managing inventory, labor and space allocation. That challenge has become even more important as consumers have become more price sensitive and retailers have adjusted store layouts and product assortments.
Average shelf volume has contracted, with stores operating approximately 7% less efficiently than in previous years. For an industry operating on average net margins of roughly 1.6%, small improvements in execution can have a meaningful financial impact.
One of the biggest opportunities is reducing out-of-stocks. Retailers lose an estimated 4.1% of total sales due to products being unavailable when consumers want to buy them. In many cases, the issue is not a supply chain failure but an execution problem inside the store. Industry estimates suggest as much as 90% of stockouts are caused by poor in-store replenishment rather than a lack of product availability.
The product may be somewhere in the building – but it is not where the shopper expects to find it. The problem plagues young CPG companies and established brands alike.
A Problem For In-Store Fulfillment and Consumers Alike
That challenge has become even more significant as grocery retailers expand digital ordering. When an online shopper sees an item listed as available but receives a substitution or discovers it is unavailable, the result is frustration for the consumer and lost revenue for the retailer. Shelf intelligence technology aims to close that gap by creating a real-time view of store conditions.
For retailers, the potential applications extend beyond ecommerce. AI-powered systems could help identify missing products, pricing errors, incorrect placement, promotional compliance issues and opportunities to improve labor allocation. That information could also reshape how manufacturers measure the performance of their brands.
Consumer packaged goods companies spend billions competing for shelf space, displays and promotional visibility, but much of their understanding of in-store execution still comes from periodic store visits, field representatives and retailer reporting.
Real-time shelf data could provide a much clearer picture of what is happening after a product leaves the distribution center and before it reaches the checkout.
Manufacturers could know whether displays were executed correctly, whether products were available, whether competitors gained shelf space and whether promotions were actually delivering the intended results. The physical shelf could become as measurable as a digital advertising campaign. That possibility also raises a larger industry question: who owns the data generated by the shelf?
Retailers may view shelf conditions as proprietary operational information. Manufacturers may argue they need access to verify execution on products they invest millions of dollars marketing. Technology companies such as Instacart could find themselves positioned between both sides as providers of the infrastructure connecting retailers and brands.
The debate could eventually mirror the rise of retail media networks, where retailers transformed consumer purchase data into a significant new advertising business. The shelf may become the next valuable retail data asset. The biggest challenge, however, may not be technical. It may be consumer acceptance.
The Slippery Slope Of Shelf Surveillance
As retailers introduce more AI-powered tools throughout the shopping experience, they will need to balance operational benefits with growing concerns about privacy, transparency and the role of technology inside stores.
Consumers are likely to embrace technology that makes shopping easier – fewer out-of-stocks, more accurate online orders and better availability. They may be less comfortable with technology that feels like surveillance or unnecessary monitoring.
The message to shoppers will need to be simple: AI is not being deployed to watch the customer. It is being deployed to make sure the product they came for is actually there.
Instacart’s acquisition of Arpalus is another sign that grocery’s next technology battle will not only happen online. It will happen in the aisle. The first generation of grocery data focused on what consumers bought. The next generation may focus on what prevented them from buying.
For an industry built on operational precision and razor-thin margins, the shelf may become its most important source of intelligence.