How do ChatGPT, Perplexity, and Google's AI Overviews decide which businesses to name?

All three work the same way at the core — they pull a set of web pages, then let a language model choose which ones to cite — but they pull from different places and weigh different things. ChatGPT leans on Bing's index, Perplexity on its own index plus live search, and Google's AI Overviews on Google's own index through its Gemini model. So getting your firm named isn't one task; it's being readable and trusted by three separate systems at once. The reassuring part is that the signals they share — a direct, self-contained answer, a clearly and consistently described business, and enough authority to be trusted — are where effort pays off across all of them.

Here is how each one actually works, and where they part ways.

ChatGPT: Bing's index, then a cautious model

ChatGPT's web search is built primarily on Bing's index, a result of OpenAI's partnership with Microsoft, and topped up by OpenAI's own crawler. In practice that means if your site isn't indexed by Bing, it's unlikely to surface in ChatGPT at all — however well you rank on Google. Bing Webmaster Tools matters here in a way most businesses never think about.

OpenAI runs three separate crawlers, and it's worth knowing which does what. OAI-SearchBot builds the search index that powers citations, GPTBot gathers training data, and ChatGPT-User fetches a specific page when a person asks ChatGPT to read it. For visibility, OAI-SearchBot is the one that has to be allowed through your robots.txt.

When it answers, ChatGPT is comparatively sparing with citations and leans toward sources it can attribute with confidence — established, authoritative domains, and pages that answer the question directly near the top rather than burying it. A clear, well-structured page from a trusted site beats a longer, vaguer one.

Perplexity: its own index, and a citation for everything

Perplexity is built differently. It runs on a retrieval-augmented model, keeping its own web index through the PerplexityBot crawler and adding real-time search on top, so there's effectively no knowledge cut-off — a page can be cited within hours of being published. Allowing PerplexityBot through your robots.txt is the entry ticket.

Its defining trait is that it cites a source for everything, with clickable links, in every answer. It also cites far more sources per answer than ChatGPT does, which is quietly an opportunity: more citation slots means less competition for each one, and Perplexity is more willing than the others to surface a smaller, specialised site when it answers the question well.

Two things move the needle most. Freshness — recent content is strongly favoured, and a visible year in a heading helps. And self-contained clarity — because Perplexity assembles its answer from several sources, it rewards a passage that stands on its own under a relevant heading over a point buried in a wall of text. It also tends to look for the same claim across more than one source before trusting it.

Google AI Overviews: Google's index, with a generative layer

Google's AI Overviews are generated by Gemini, grounded in Google's existing search index. When a query triggers an Overview, Google retrieves a large pool of candidate pages, filters them for authority and relevance, has Gemini re-rank and lift the most useful passages, and assembles a summary with a handful of inline citations.

The part that catches people out: ranking in Google's top ten is no longer a reliable route to being cited in the Overview. Through 2025 most cited pages also ranked organically; by early 2026 that overlap had fallen sharply. Being indexed and crawlable by Googlebot is still the prerequisite, but the Overview increasingly rewards something past rank — a complete, self-contained answer, a business entity Google already recognises, and supporting structured data — over a page that merely ranks well.

A note on crawlers: allowing Google-Extended governs whether Google may use your content for its generative features, which is separate from Googlebot indexing your site for ordinary search. You want both open.

Side by side

What they have in common

Strip away the differences and the three rhyme. Each retrieves a pool of pages and then has a model decide what to cite, so crawl access is the price of entry everywhere — block the wrong bot and you're out before the contest starts. Each rewards a direct answer that stands on its own, near a clear heading. And each leans on authority and corroboration: a business described consistently across the web, and trusted in more than one place, is easier for any of them to name without hesitating.

What this means for your firm

The practical lesson is that you can't optimise for one engine and assume the rest follow. The three draw from three different indexes, so a firm can be cited by Perplexity and absent from ChatGPT simply because it's strong in one index and missing from another. Being named consistently means covering all of it: indexed in both Bing and Google, reachable by Perplexity's and OpenAI's crawlers, and described clearly and identically enough that all three trust who you are and what you do.

For a local business that resolves to the same foundations every time — a consistent, well-described entity, genuine reviews and outside mentions that build authority, and content that answers the questions buyers actually ask, plainly and near the top. Do that and you aren't gaming any single engine; you're giving all three the thing each of them is looking for.

If you want to know where your firm stands today, the only honest way is to ask all three the questions your clients would ask, and see who gets named. Most firms have never looked.