Agentic commerce benchmark by store segment

A group of stores that share something about how they are built or run, rather than what they sell: the currency they price in, how their prices compare with similar stores, how many products they list, who runs their checkout, whether they offer an express wallet, and whether a bot manager stands in front of them. The same store sits in one group for each of those, so a segment is a different cut of the same measurements.

Because what stops an AI agent is often how a store is built rather than what it sells. Two apparel stores can fail at different places because one runs a third-party checkout and the other sits behind a bot manager. Segments let a merchant compare against stores built like theirs.

From what your own storefront showed during its latest scan: your product markup, your catalogue feed or sitemap, the response headers your site sent, and the pages the agent reached on its way to checkout. Ottom makes no extra request to decide a segment and nobody places a store by hand.

No. Segments overlap and travel together: stores behind a bot manager are often larger, and larger stores often run their own checkout. A difference between two segments describes the stores in them. It is a place to look, not a diagnosis, and your own scan report says what stopped the agent on your store.

The price tier segment answers that from measurement rather than opinion: it groups stores as mainstream, premium or luxury by how their prices compare with similar stores in the same currency, and reports how far real AI agents got on each. Pricier stores are also more often behind a bot manager, so read a tier difference together with the bot protection segment.

A store whose median product price is at least 3 times the median of stores in the same category that price in the same currency, with at least 10 such stores to compare against. Prices are never compared across currencies, and nobody picks the stores by hand.

The express wallets segment groups stores by whether their cart or checkout offered an express wallet button to our agent. It describes the stores in each group; it does not show that a wallet causes the result, and Apple Pay never appears in the browser our agent uses.

The checkout segment separates the platform’s own checkout, a third-party checkout service and a checkout the store built itself. Only stores whose agent reached checkout are in it, so its figures describe stores an agent got that far on.

Sometimes. A bot manager decides whether a visitor looks automated, and an AI agent acting for a shopper is automated by design. The bot protection segment compares stores where one answered in front of the store, or a challenge page met the scan, with stores where none was seen. A content delivery network on its own does not count.

It can. An agent has to find the right product before it can buy it, and a large catalogue without clear structure or structured data makes that harder. The catalogue size segment groups stores by how many products they list, read from their own feed or sitemap.

A segment is shown only once enough of its stores have been measured at every stage for a figure that cannot be traced back to a single store. Until then it is left out rather than shown with a caveat. A segment can also be missing because stores in it have not been scanned since segments were introduced.

Every night. Each figure covers the stores measured in the trailing 90 days, so a segment moves gradually as stores are scanned again rather than jumping.

Run a scan of your own store. The report says where the agent stopped and what to change, written so you can forward it to your developer. A paid plan then sets your store against the segments it is in.

The headline figure for each segment is public and free to quote with attribution. The full breakdown, with quartiles and the share of stores passing each stage, is free with an Ottom account and is not open data.