Learn / Is my store ready

What happens when an AI agent visits your store

An AI agent arrives at your store cold, fetches a small number of pages, and keeps only what it can read as text or structured data. It does not run your scripts the way a browser does, does not hover, does not wait, and does not have a session. Everything your storefront does to make a page feel good is invisible to it, and everything it needs has to be stated rather than shown.

It arrives without any of the context you assume

There is no cookie, no signed in state, no previously dismissed popup and no remembered currency. Every visit is a first visit. Anything your store shows once and then stops showing will be shown to an agent every single time, including the newsletter modal, the cookie wall and the region selector.

This is why a checkout that works fine for you can fail entirely for an agent. You have not seen your own cold session in months.

It takes the cheapest path to the answer

An agent trying to establish what you sell and for how much will use the least expensive source that answers the question. Structured data beats parsing your markup. A product feed beats fetching pages one at a time. Fetching markup beats rendering the page in a browser.

The practical consequence is that your carefully built product page may never render at all. If your price lives only in a script that runs on render, an agent that stopped at your markup has no price, and it will not escalate to a full browser to look for one.

What a shopper sees against what an agent reads The same product page. A person sees an image, a price rendered by script, and a policy set in a graphic. An agent reads only the text and structured data, so the price and the policy are missing from its view. WHAT A SHOPPER SEES product photo Trail Jacket £129.00 In stock, ships tomorrow Free returns for 30 days WHAT AN AGENT READS image: no alt text Trail Jacket price: not found availability: not stated returns: image only
The same product page, twice. Your store looks fine in a browser, which is why this is invisible to you.

It reads facts, not impressions

Your photography, your typography and your layout carry no information for an agent. What carries information is stated fact: the price, the availability, the variant options, the shipping cost, the returns window, an identifier such as a GTIN that lets you be matched to the same product elsewhere.

A store that shows these and a store that states these look identical to a person and are completely different to an agent. The second one can be compared. The first one can only be guessed at, and guesses lose.

It stops the moment something is unanswerable

An agent does not troubleshoot. If your variant picker needs a hover, it does not try a different approach. If your checkout demands an account, it does not create one. If a challenge appears, it does not solve it. It stops, and from the outside that looks like your store simply not being chosen.

That is the whole reason this is hard to notice. There is no error, no bounce, no abandoned cart in your analytics. There is an absence, and absences do not appear in reports. Signs your store is invisible to AI agents covers the symptoms that do show.

What this means in practice

Almost everything that makes a store agent ready is unglamorous data hygiene rather than redesign. State your facts as data. Make your catalogue readable in one request. Let your checkout work cold, on a narrow viewport, signed out. None of it changes what a person sees, which is both why it gets skipped and why it is cheap.

For the underlying model of what the agent is trying to do across the whole journey, see what is agentic commerce.

Questions

Does an agent load my page like a browser does?

Sometimes, and you cannot rely on it. Some agents fetch and parse markup only. Others drive a real browser and can execute scripts. Designing for the second and hoping is the mistake, because the cheapest path is always tried first: an agent that can get your price from structured data will never render your page at all.

Will it see my discounts and promotions?

Only if they are in the data. A banner announcing a sale is decoration. A price that reflects the sale in your structured data is a fact. If the two disagree, the agent uses the one it can read, and a comparison is decided on that number.

How many pages will it look at?

Far fewer than you would expect, which is why enumeration matters. It is trying to answer a question, not to survey your catalogue, so a store that makes its products readable in one request gets considered where a store that must be crawled does not.

Scan your store