How Perplexity Picks Its Sources and Which Brands It Cites

Perplexity is built around live search. In normal use it searches the web, reads a handful of pages and writes an answer with numbered citations beside it. That makes it one of the most transparent AI engines about where its information comes from, and it suggests a brand’s presence on crawlable, retrievable pages counts for more here than anything a model memorized during training.

For how the other engines compare, see how ChatGPT, Gemini, Claude and Perplexity decide which brands to mention. What follows separates what Perplexity documents, what studies have measured, and what is reasoned from both.

How does Perplexity find the pages it cites?

Documented facts, from Perplexity’s crawler documentation:

• PerplexityBot is “designed to surface and link websites in search results on Perplexity.” Perplexity states it “is not used to crawl content for AI foundation models.”
• Perplexity-User fetches pages when a person’s request needs them. Perplexity says that “since a user requested the fetch, this fetcher generally ignores robots.txt rules.”
• Changes to robots.txt may take “up to 24 hours” to be reflected.

Reasoning. This is a classic case of retrieval-augmented generation: find pages first, then write from them. For a brand, it means two things. Your pages and the pages about you must be crawlable and indexed. And the content on those pages, not your reputation in some abstract sense, is what Perplexity reads when it answers.

How big is Perplexity?

Reported by its CEO. Aravind Srinivas said Perplexity handled about 780 million queries in May 2025, growing more than 20% month over month. No official 2026 figure has been published that we could verify.

That is far smaller than ChatGPT or Google’s AI features. Reasoning. A tool that puts sources in front of every answer suits people who like to check them. For B2B, SaaS and considered purchases, that kind of audience can matter more than its size suggests, though we know of no published data on Perplexity’s user mix.

Which sources does Perplexity cite most?

Industry study. Profound analyzed 680 million AI citations between August 2024 and June 2025. For Perplexity, Reddit made up 6.6% of all citations and 46.7% of its top-10 source share. YouTube was 2.0% overall and 13.9% of the top-10 share.

Two cautions. First, “top-10 source share” means share among the ten most-cited domains, not of all citations; most citations are spread across a long tail of other sites. Second, these mixes move. Semrush saw Reddit’s presence in ChatGPT answers drop from about 60% to about 10% within weeks in 2025. A source that dominates today may not tomorrow. Our guide to the sources AI cites most tracks these shifts across engines.

Reasoning. The long tail is where most brands live. Specialist publications, niche blogs, local news, industry sites and comparison pages make up the bulk of what Perplexity actually reads. A brand described accurately across many of those pages has many small chances to be retrieved, rather than one big bet on a single platform.

How accurate is Perplexity about sources?

Independent research. The Tow Center at Columbia tested eight AI search tools on 1,600 queries about news articles. Across the tools, more than 60% of answers were incorrect; Perplexity was wrong 37% of the time, the best result in the group. The study also found tools citing syndicated copies instead of originals and retrieving content from publishers that had blocked their crawlers.

A wider European Broadcasting Union study across ChatGPT, Copilot, Gemini and Perplexity found 45% of AI answers about news had at least one significant issue.

These studies were about news, not brands. No credible study we know of measures AI errors about companies specifically. But the lesson carries over: a citation next to a claim does not guarantee the claim is right. If Perplexity describes your brand wrongly, see our guide on fixing outdated or incorrect AI answers.

What does this mean for brands that want to be cited?

Reasoning, built on the facts above.

1. Allow PerplexityBot. It is not a training crawler, so blocking it mostly removes you from answers.
2. Be on pages that answer real questions. Perplexity retrieves pages for the question asked. An article that mentions your brand in the context of “best CRM for property managers” is retrievable for that question; a generic directory listing usually is not.
3. Keep facts consistent. When Perplexity pulls three or four sources into one answer, matching facts across them reduce the chance of a muddled description.
4. Take part in communities honestly. Answering questions as yourself, with your affiliation stated, is fine. Astroturfing Reddit with fake accounts breaks platform rules and, where it amounts to fake reviews or endorsements, can break consumer law (general information, not legal advice). Our comparison of brand mentions and community marketing covers the trade-offs.
5. Spread across many relevant publications. Because sources rotate, breadth protects you better than any one placement.

Suppose a small accounting software firm in Ontario is mentioned in eight articles about bookkeeping for contractors, on regional business sites and trade blogs. When someone asks Perplexity “what accounting software do Canadian contractors use,” several of those pages are plausible retrievals. None of them guarantees a citation on any single run, and answers vary run to run. Across many runs, though, the brand has a fair chance of appearing, and that is measurable.

How is Perplexity different from ChatGPT for brands?

Studies plus reasoning. The two engines lean on different sources. In Profound’s data, Wikipedia led ChatGPT’s top-10 source share while Reddit led Perplexity’s. ChatGPT also often names brands without linking them: BrightEdge found it mentions brands 3.2 times more often than it cites them. Perplexity, by design, puts its sources in front of the reader for nearly every answer.

That changes what a good outcome looks like. In ChatGPT, being named in the text is the main prize. In Perplexity, the page that is cited also gets seen, and readers who like to check sources may click it. So the quality of the pages that mention you matters twice: once for whether Perplexity retrieves them, and again for what a reader thinks when they open one. A thin or obviously promotional page can hurt more here than in an engine that hides its sources. Our article on where ChatGPT gets its information covers the other side of this comparison.

What is unknown?

Perplexity does not publish which search index or indexes it queries, how it ranks retrieved pages, or how it decides which brands to name. We do not claim to know. The useful approach is the one that works for every engine: measure how often you appear across a set of real buyer questions, run several times, as set out in our guide on measuring AI visibility.

What to do next

• Confirm PerplexityBot is allowed in your robots.txt.
• Ask Perplexity ten questions your buyers ask, three times each, and note which sources it cites and whether you appear.
• Get a free AI visibility check to see where you stand across Perplexity, ChatGPT, Gemini and Google’s AI answers.
• If Perplexity and ChatGPT already cite or name you, growing how high ChatGPT places you is the job of The Blue Ocean GPT.