A new study tracking 375 product brands across Shopify, Amazon and other sales channels finds that stockouts (especially on best sellers) carry a bigger price tag than many companies realize, and that burden is set to rise as shopping shifts toward AI assistants.
The study, titled “The cost of invisible stock in the AI era,” was done by Katana, the cloud-based software company that provides inventory management and ERP solutions, which found that a typical brand lost an estimated $21,000 per year to periods when top products were unavailable to sell. Researchers found that the impact was unevenly distributed: the top 25 percent of brands lost about $82,900 annually, while the hardest-hit 10 percent saw losses topping $268,000.
More from WWD
“Most brands never see the full number,” the report’s authors said. “It shows up a few days at a time, across the year, and reads as the cost of doing business.”
And that cost is about to grow. “Shopping is moving into AI tools, and quickly: traffic from AI tools to U.S. retailers grew 393 percent year over year in early 2026,” the report stated. “Instead of browsing a store, buyers ask ChatGPT or Google for what they want, and the tool returns a short list of products that fit and are ready to buy. In agentic buying, there is no shelf to browse or a prompt to consider something similar. A product the tool reads as unavailable is a product the shopper never sees.”
The findings suggest the problem is not a rare glitch, but a persistent pattern concentrated in a small slice of a catalog. Across the measured period, 69 percent of stockout losses fell on the top 10 percent of brands, and most brands experienced repeat failures. More than half of the stockouts involved items that had already run out earlier in the year.
On average, the best-selling products studied went out of stock roughly 14 times per year for about two days each, leaving items unavailable for around a month total—losses often missed in standard reporting because many stockouts are quickly resolved, from backorders or short production runs, with sales continuing in the background.
With AI-driven search converting about 42 percent better than traditional search, even brief gaps could shift purchases immediately to competitors, shrinking the “forgiveness window” that once allowed brands to recover after a delay.
The report’s authors also said stockouts work differently with AI tools. “In a store or on your website, a shopper who hits an out-of-stock product can pick something similar or come back in a day or two,” the authors said, adding that with AI tools, buyers describe what they want, “and the tool only shows products that match and are available to buy right then. If your product reads as unavailable, it doesn’t make the list, and the shopper won’t even know it was an option.”
