Every page here answers one question and says when it was last measured. Where a page reports what a model did, it names the provider and the exact model, because a model's behaviour changes silently under the same name.
- Why does each AI model give a different answer about my product?
How much answers about the same product differ between models, where the difference actually sits, and why one model is not a sample of AI. - What can and cannot be measured about AI answers
The parts of an AI answer a brand can measure reliably, the parts it can only estimate, and the claims no measurement supports. - What an official product destination is, and why it is the outcome to measure
The difference between an AI answer that mentions your product and one that sends the buyer to your product's own page — and how to tell them apart. - What a brand controls in AI answers, and what it does not
A map of the parts of an AI answer about your product that you can change, the parts you can only influence, and the parts that are out of reach. - Structured data versus visible text on a product page
What JSON-LD markup and visible page text each do for AI systems, why you need both, and what happens when they disagree. - A PIM and a canonical product record are not the same thing
What a product information management system does, what a public source-backed product record does, and why one does not replace the other. - Is AI crawler traffic the same as AI visibility?
Why a count of AI crawler requests tells you almost nothing about whether AI answers mention your product — measured on our own pages. - How to write product facts that AI can quote
Rules for writing product facts as sentences a model can reproduce accurately: units, limits, variants and what the product does not do. - How to verify AI crawlers in your server logs
A short procedure for checking whether a request that claims to be an AI crawler really comes from that provider, and what to store. - How to read an AI visibility report
Four questions that tell you what an AI visibility number actually counts, before you act on it. - How to publish GTIN and MPN so AI systems can use them
Where identifiers need to appear on a product page, in which form, and how to check the result without special tools. - How to make product facts readable without JavaScript
Why AI crawlers often read a different product page from the one your customers see, how to check it in two minutes, and what to move into the HTML. - How to give every product one canonical URL
Why AI answers link your category or home page instead of the product, and the URL rules that give each product one address worth citing. - How to check what AI says about your product
A repeatable procedure a brand can run in an afternoon: which questions to ask, which models, what to record, and how to read the result. - How to allow AI crawlers without allowing impostors
Firewall and bot-protection rules that let verified AI crawlers through while still blocking scrapers that borrow their names. - Does structured data make AI cite your product?
What Product structured data does for AI systems, what it does not do, and why it works as a second channel rather than a lever. - Does llms.txt help your products appear in AI answers?
What llms.txt is, how often AI crawlers actually request it on our own site, and where the effort is better spent. - Crawled, cited and used are different things
The steps between publishing a product page and an AI answer that relies on it, what evidence each step needs, and why they are so often merged. - Can you trust a GPTBot user agent in your logs?
What a user agent string proves, how often requests claiming to be AI crawlers fail a basic check, and how to tell the difference. - Blocking AI crawlers by mistake
How bot protection and firewall rules end up blocking the AI crawlers a brand wants, and how to check what your site actually lets through. - AI sends buyers to a retailer instead of to you
Why AI answers about your product so often link a retailer's page rather than yours, and when that is — and is not — a problem to fix. - AI repeats an outdated specification of your product
Why models keep stating last season's figures after a product changes, and what to publish so the change is explicit. - AI mixes up two variants of your product
Why answers combine one variant's size with another's weight, and how to make variants unambiguous on the page. - AI links to your site, but not to your product page
Why a model's link to your official site so often lands on the wrong page, how often it happens, and what you can change. - AI gives the wrong price or availability for your product
Why prices and stock levels in AI answers go stale faster than any other fact, and how to publish them so they are treated as observations with a date. - AI crawlers fetch your pages but never cite them
Why a provider can crawl hundreds of your pages and cite none, measured on our own site, and what that does and does not mean. - AI cites review sites instead of your own page
How often model answers about a product point at someone else's page, which kinds of sites they are, and which part of that a brand can influence. - AI cites your page for something your page does not say
Why a citation can point at the right page while the answer states the wrong fact, and how to check what was actually used. - Why AI shows a competitor instead of your product
What happens between a buyer's question and a recommendation, which parts of it you control, and how to check your own products in an hour. - How AI reads product data
What AI systems actually fetch about a product, what they reconstruct, and which parts of that a brand can control.
The records these pages refer to are published at /products, and the evidence rules they follow are at /evidence.