You ask ChatGPT for the best tool in your category and it names three companies. None of them is you. One of them is a competitor you know you beat on product. This is not a bug and it is not personal. The assistant is doing exactly what it was built to do, and the reason it named them is knowable and fixable.

How an AI assistant actually picks who to name

When someone asks a buyer question, the engine builds its answer from three things. First, what it learned in training, which is a compressed memory of a large slice of the public web up to a cutoff date. Second, what it retrieves at answer time, which for engines with live search means its own web index, plus places like Reddit, review sites, and roundup articles it pulls in to ground the reply. Third, how easy your own pages are to read and quote once it gets to them.

A competitor wins a spot when it is present in those inputs and you are not. That is the whole mechanism. The model is not judging your product. It is assembling an answer out of the material in front of it, and your competitor is in the material.

The concrete reasons a competitor gets named instead of you

In practice, the brand that gets recommended tends to have several of these going for it. Yours may have none.

  • They are cited in the sources the engine reads. When an assistant answers a "best X" question, it often leans on G2 and Capterra listings, Reddit threads, and third-party roundup posts. If your competitor is in those and you are absent, the engine sees them and not you. In our own scan of the AI-visibility-tool category, 60 buyer questions across 7 engines produced 372 answers, and the engines drew on roughly 20 recurring sources, with Reddit the single most-cited. Miss the sources that repeat and you miss most of the answers.
  • They have pages that answer the buyer question directly. If a competitor has a page titled for the exact thing the buyer asked, written in plain buyer language, the engine can lift a clean sentence from it. If your closest page is a vague homepage or a feature grid, there is nothing quotable to pull.
  • They are mentioned across the web more. More mentions in more places means the model saw the name more often during training and is more likely to retrieve it now. Repetition across independent sources reads as consensus.
  • Their pages are clear and structured. Straight headings, plain descriptions, a real pricing page, and schema markup make a brand easy to parse and easy to quote. A site that hides its offer behind marketing language is hard to summarize, so it gets summarized less.

Notice that none of these is about being the better product. They are all about being findable and quotable. You can lose to a worse competitor purely because it is easier for the machine to read.

What to do about it

The fixes follow directly from the causes. There is no trick here and no shortcut.

  • Get into the cited sources. Find the review sites, Reddit threads, and roundups the engines actually read for your category, then earn a real presence in them. A claimed G2 profile, honest reviews, and a genuine answer in the thread where buyers ask do more than a month of homepage edits.
  • Publish pages that answer real buyer questions in buyer language. Take the questions your buyers ask and write a clear page for each one. Use the words they use. Give a direct answer near the top so the engine has a clean sentence to quote.
  • Earn mentions. Get named in the posts, comparisons, and lists where your category is discussed. Every independent mention is another place the engine can pick you up.
  • Add structured data and clean up your pages. Clear headings, a plain pricing page, and schema markup make you easy to parse. This is unglamorous and it works.

What you cannot do is bribe or prompt-inject your way in. There is no ad slot inside the answer, and hidden text telling the model to recommend you gets ignored or filtered. This is earned presence. The brands that get named did the work of being present where the engine looks.

How measuring it works

You cannot fix what you cannot see, and guessing which sources matter for your category wastes months. The way to know is to run the buyer's real questions across the engines and watch what comes back.

That is what Saymetry does. We take the actual questions your buyers ask, run them across the major engines several times each, and record who gets named and which sources powered each answer. You end up with a plain picture: here are the questions where a competitor wins, here are the sources doing the work, here is where you are absent. From there the gaps are obvious. You stop guessing and start closing the specific sources and pages that decide your category.

FAQ

Can I pay ChatGPT to recommend me?

No. There is no ad slot and no paid placement inside the answer. The engine names brands it can find and quote from its training data and the sources it retrieves. You earn your way in by being present in those sources, not by paying the model.

How often do the answers change?

Often, and not on a fixed schedule. The same question can return a different brand list from one run to the next, and answers shift again when models are updated or when the pages and threads they read change. A single result is a snapshot, not a fixed ranking. This is why measuring means running each question several times and looking at the pattern.

How is this different from SEO?

SEO tries to rank a page in a list of blue links. AI visibility is about whether an assistant names you inside a written answer. The work overlaps, because clear pages and outside mentions help both, but the target is different. You are optimizing to be quoted and recommended, not to sit at position three.

My competitor is smaller than me. Why do they win?

Size does not decide it. Presence does. A smaller competitor that is discussed on Reddit, listed in review sites, and has pages that answer the exact buyer question will get named ahead of a larger brand that is quiet in those places. The engine rewards what it can read, not what is biggest.