Retailers are moving fast to make their catalogs agent-ready this holiday season. They’re feeding product data into ChatGPT, Google and other shopping agents customers now use to browse, compare and buy. Most of the work goes toward visibility, improving titles and providing richer attributes.
What many brands may overlook is a key piece of data determines whether an agent sees a product in the first place: whether it’s in stock.
That’s a stockout problem, and it’s not new. What is new is what happens when an agent, instead of a human, is the one hitting it. Retailers who haven’t priced in that difference are going to feel it this peak season more than before.
Stockouts Always Cost You. Now They Cost You Instantly.
Stockouts have long been a common challenge in the retail industry. Katana data found that for small and mid-sized product brands, 69% of stockouts hit an item that had already run out earlier in the same year. The typical brand lost an estimated $21,000 a year to these inventory gaps, with the hardest-hit seeing upward of $268,000 in losses.
Historically, most of that loss was invisible because humans absorbed it. A shopper would hit a sold-out product, call the store or wait for a backorder. More than half of these stockouts are resolved within a day, and the order ends up shipping. AI shopping agents remove that buffer entirely.
An agent doesn’t call to check when something is coming back in stock or wait a day to try again. It’s reading through a retailer’s feed and whatever it says in that moment determines its course of action. If an item is sold out online, even if the warehouse has three days of stock still moving, the agent simply doesn’t put that product in front of the customer. The sale ends up going to whichever competitor has an available, and machine-readable, alternative.
More AI Traffic Means The Same SKUs Sell Out Faster, More Often
Black Friday and the holiday shopping season is exactly when inventory data is under the most stress and AI traffic is heaviest. In 2025, Botify found AI bot traffic for retail grew 5.4x, and Salesforce projects that 20% of all 2026 holiday e-commerce traffic will come from AI chat agents. Product discovery is no longer just a human behavior. It’s increasingly mediated by systems sitting between retailers and their customers.
Product demand spikes resulting from increased AI traffic generate the exact conditions that compound the repeat-stockout problem. AI-driven traffic tends to convert faster than traditional search traffic, because these shoppers already know what they want. That makes them some of the most valuable buyers a brand can hold onto, and it also means hero SKUs sell through faster than replenishment cycles can keep pace.
A Three-Part Test For Machine Readable Inventory
Many retailers will spend time fixing their storefront and feed but leave the inventory data underneath untouched. Taking proactive steps to ensure inventory data is logged and updated in real-time is key to winning in agentic commerce. A few things separate the brands that come out ahead.
First, treat your reorder point as a calculated output, not a number someone typed in once. If a product has run out more than once this year, the threshold is sitting below real sell-through velocity. It should be recalculated from current demand, lead time and safety stock, not carried over from last year’s assumptions.
Second, have one definition of “available” everywhere. If a unit is on backorder, but will ship on time, decide if the system considers it in stock or not. Consistency across every channel is key, because an agent doesn’t know or care that your Shopify admin and product feed disagree. Third, track stock by location. An agent checking availability for a specific buyer is verifying against specific fulfillment locations. An agent checking availability for a specific buyer is checking the location that would actually fulfill the order. A healthy total across the business means nothing if the warehouse shipping that order is empty.
Every brand is racing to be seen by AI agents. That work is wasted if what the agent sees isn’t true. A product feed is only a projection of the system underneath it, and that system is what the agent is really trusting. Going into the highest-stakes selling season of the year, getting real-time inventory into your feeds and holding one picture across every channel is the operational foundation for discoverability.
Ben Hussey is the co-CEO of Katana, an inventory management platform that helps companies manage over $3 billion in sales annually. Ben has led many successful sales and revenue teams, helping businesses enhance their e-commerce, manufacturing, inventory, and order management capabilities while delivering amazing customer experiences. In addition to these roles, Ben spent a decade working for a large telecommunications company, leading commerce initiatives of varying sizes and types — from initiation to delivery and run-time. He’s passionate about the impact software can have on a business and working with high-performing teams to deliver results.
