Why AI engines cite your competitor instead of your store

When a buyer asks an AI engine for the best product in your category, the engine does not rank a list of pages. It retrieves a handful of sources, reads them, and writes one answer naming two to five brands. Your store is either in that answer or it is not. There is no position two.

This is the part that surprises store owners: the brands in that answer are usually not the ones with the strongest product pages. They are the ones the engine can resolve as entities and corroborate from sources it did not write itself.

What the engine actually does with your category

For a buy-intent question, an answer engine runs its own retrieval, pulls a small set of documents, and generates a response grounded in them. Two things follow from that.

First, the set is small. A page-one ranking gets you ten slots; an answer gets you three or four names. The competition is tighter, not looser.

Second, retrieval favours documents that make the comparison explicitly. A roundup titled "best merino base layers" is directly usable. Your product page, which describes one product and assumes the visitor already chose your brand, is not — there is nothing in it to quote in a ranked answer.

The three things that decide whether you are named

Your brand resolves as an entity

The engine needs a stable object to attach a recommendation to: a consistent name, a consistent description, Organization and Product structured data, and sameAs links to the profiles that confirm the brand exists. A store with no entity markup is a collection of pages, and a collection of pages is not something an engine can recommend by name.

Somebody else has compared you

Answer engines weight independent sources above a brand's own domain, because a brand describing itself is not evidence. If no roundup, review, forum thread, or retailer listing has placed you next to your competitors, there is no document that makes the comparison, and the engine will build it from the ones that do.

Your differentiator appears somewhere other than your homepage

This is the one that stings. Most stores have a real, specific claim — made in the USA, plus-size, mulesing-free, under fifty dollars. They state it on their own site and nowhere else. When a buyer asks for that exact attribute, the engine answers from third-party pages that discuss the attribute, and those pages name whichever brands their authors happened to know.

Why the answer changes every time you check

Answer engines sample from a language model and re-run retrieval on every request, so the same question asked twice minutes apart can return different brands. A single check is not evidence, in either direction — a store owner who sees their name once concludes they are fine, and a store owner who misses once concludes they are invisible. Both are reading noise.

A single check is not evidence. Frequency is.

The only stable measure is frequency: ask the same question several times and count how often the brand appears. That is why a Qubebuild scan runs five prompts five times each and reports a citation rate out of twenty-five rather than a yes or a no.

What to do about it, in order

1. Publish entity markup. Organization on the site, Product and Offer on every product page, AggregateRating where you have real reviews. This is the cheapest step and the one most stores skip.
2. Write the comparison pages nobody has written about you. Your brand against the two competitors buyers actually name. Make the trade-offs explicit, including the ones that favour them — a page that only flatters you is not usable as a source.
3. Go and appear on the domains already being cited. Run a scan, look at the list of sources the engine used, and treat it as a target list. Those specific domains are what the engine reads when somebody asks about your category.
4. Re-measure on a schedule. Changes to your own pages typically show up in answers within two to six weeks. Changes that depend on third-party sources take longer, because they only register once the source publishes and is re-crawled.

None of this is a trick. It is the same work that made a brand findable in classic search, re-weighted for a system that returns one answer instead of ten links.