Why AI Engines Name Your Competitor Instead of You

If ChatGPT names your competitor and not you, we can’t point you to one proven reason, and we haven’t found a study that isolates one. But some causes are documented by the companies themselves, and some are supported by research. This post separates the two and says which findings are only associations.

Can the engine reach your site?

This is the cause with the clearest documentation. OpenAI lists several crawlers with different jobs, and three matter here. OAI-SearchBot is “used to surface websites in search results in ChatGPT’s search features”, and sites that opt out “will not be shown in ChatGPT search answers, though can still appear as navigational links”. GPTBot crawls content that may be used to train OpenAI’s models. ChatGPT-User is used “for certain user actions in ChatGPT and Custom GPTs”, so robots.txt rules may not apply, and OpenAI says it “is not used to determine whether content may appear in Search”. For managing search opt-outs, OpenAI points to OAI-SearchBot.

Perplexity describes a similar split. PerplexityBot is “designed to surface and link websites in search results on Perplexity”, and Perplexity recommends allowing it in your robots.txt if you want to appear. Perplexity-User fetches pages when users ask, and “generally ignores robots.txt rules” because a user started the request.

For Google, a page must be indexed and eligible to be shown in Search with a snippet before it can be a supporting link in AI Overviews or AI Mode. Google says there are no additional requirements beyond that.

So check your robots.txt for OAI-SearchBot and PerplexityBot, and check that your key pages are indexed in Google. A blocked crawler is a reason you can find and fix. It is not the only reason, and it may not be yours.

If your site is new, how long any of this takes is a separate question. It is covered in how long until AI engines find your new website.

What the model already knows

In a September 2026 preprint, Edward Malthouse and colleagues asked six models for product recommendations in five categories: GPT-5.5, GPT-5.4 Mini, Gemini 3.1 Pro Preview, Gemini 2.5 Flash, Claude Opus 4.7 and Claude Sonnet 4.6. They used the providers’ APIs with web search and other retrieval tools switched off, so the answers reflect the models themselves, not a live lookup. Each request asked for up to five brand names for a bare category, such as a hiking jacket.

Many large, established brands were not named at all. L.L.Bean, Eddie Bauer and REI Co-op received no recommendations for hiking jackets, and Craftsman and Black+Decker received none for cordless drills. Being big did not guarantee being named.

What did the naming go with? The authors found prominence associated with a brand’s wider visibility: mainly Google search interest, then online brand conversation. The plain correlation between search interest and how prominently a brand was ranked was .63, on transformed values. In the authors’ combined model, news mentions, advertising spend and Wikipedia page views added little on top of those two. On their own, all five measures correlated positively with prominence. The authors call this an exploratory, observational result that says nothing about cause, and they add that it is not a reason to optimise Google Trends directly.

Ahrefs, an SEO software vendor, published a broadly consistent pattern from 75,000 brands, using Spearman correlations against how often a brand was mentioned in AI responses. Its own brand-tracking product, Brand Radar, supplied the data, and that product competes with tools like Bevizio. Branded web mentions correlated at 0.664 for ChatGPT, 0.709 for AI Mode and 0.656 for AI Overviews. Backlinks and URL rating were very weak. Ahrefs’ own caveat: “correlation isn’t causation”.

How you ask changes who gets named

The same preprint also tried detailed prompts that described a buyer’s situation, using two hand-picked brands as an illustration rather than a systematic comparison. Craftsman was not named in any bare-category list for cordless drills. With detailed needs-based prompts it appeared in 35.4% of 240 recommendation lists. L.L.Bean, for hiking jackets, rose to 5.4%. The authors call the effect “limited”, but it shows that missing from the answer to “best X” is not the same as never named.

That is why the question matters. The one you test should be the one your buyers would ask, with their use case in it.

What the evidence doesn’t show

Both studies are observational. Neither shows that raising your search interest or your number of mentions makes an engine name you. The preprint tested models through APIs in fresh sessions with no user history. Its authors note that real users have chat histories that could change the results. We found no sign that it has been peer reviewed.

A different preprint looked at how rarely small software brands surface at all; it is covered in do AI engines ignore small software? What the evidence says about improving your chances is in how to rank in ChatGPT: what is actually known.

See it for your own product

Bevizio measures. It doesn’t write or fix anything for you. What it shows is who the engines name instead of you and which sources they used. The free check gives you one answer from each of three engines (ChatGPT, Perplexity and Google AI Overviews) to start with, and the live demo shows a real account over time. Because answers vary from run to run, read them as counts, as why we ask the same question 8 times a week explains. The step-by-step version of checking by hand is in Does ChatGPT recommend your product?