Learn
Signs your store is invisible to AI agents
A store that is invisible to AI agents usually looks completely healthy to its owner. The signs are specific and checkable: product data that only exists once scripts have run, terms published as images, a sitemap that lists collections rather than products, and bot rules written for training crawlers that also catch shopping agents. None of them affect how the store looks to a person, which is why they survive.
A store that is invisible to AI agents almost always looks completely healthy to the person who runs it. The signs below are the ones you can check without tooling, and they share a property: none of them changes how the store looks to a human, which is why they survive redesigns and audits.
Your price only exists after JavaScript runs
Open a product page with JavaScript disabled. If the price disappears, an agent reading your page has no price. It will not wait, it will not execute your pricing script, and it will not infer the number from a currency symbol next to an empty element.
This is the single most common defect and the least visible, because every human who checks sees the price correctly. The fix is publishing it as product structured data rather than changing how the page renders. See how to add product schema for AI agents.
Your policies are pictures
Shipping thresholds and returns windows set in a designed graphic are unreadable. An agent asked whether you offer free returns will answer that you do not say, which is a different answer from no, and worse: it is the answer that loses a comparison to a store that simply wrote it down.
Your sitemap lists collections, not products
A sitemap that stops at category pages tells an agent that you sell things without telling it what. It then has to crawl to find your catalogue, which is expensive, and expensive means skipped.
Your variant picker only answers to a mouse
If choosing a size requires a hover, or the size options only exist once a script has attached them, an agent cannot select a variant. It reaches a product it wants and cannot say which one it wants. See hover only variant picker and variant js required.
Your bot rules were written for a different problem
Most robots.txt rules about AI were written to keep content out of training data. Shopping agents were grouped in because they were listed in the same guide. The rule still says what it said, and it now also means your products are not considered.
The same is true one layer up, at your CDN, where the setting is likely a category toggle somebody chose once. Neither of these announces itself. See robots blocks agent.
Your checkout has never been tried from a cold session
Forced account creation, a popup on the cart page, an add to cart that needs scripts, a challenge at the payment step. Each of these is invisible to you because you are signed in, you have dismissed the popup, and your session is warm. An agent arrives cold every time.
The common thread
Every sign here is a place where your browser was doing work that an agent will not do. That is the whole category. It is also why the absence of complaints is not evidence: the shoppers who would have complained never arrived, and the agents that failed did not file a ticket.
If you want to check these in order rather than at random, how to test if AI agents can buy from your store runs through them in the sequence an agent meets them.
Questions
Would I see a drop in traffic if this were happening?
Probably not, and that is the difficulty. Agent traffic is currently a small share for most stores, so its absence does not show up as a dip. You are looking for traffic that never arrived, which no analytics view reports.
Which of these signs matters most?
Whichever comes earliest in the journey. A bot rule that blocks fetching makes every other defect irrelevant, because nothing gets far enough to encounter them. Fix in the order an agent meets them rather than in the order they are easy to fix.
My developer says the site is standards compliant. Is that enough?
It is a good sign and it is not the same claim. Standards compliance is about how the markup is written. What matters here is whether specific facts, price, availability, variants and terms, are present as data. A page can validate perfectly and still not say what it costs.