Perplexity answers look different from other AI assistants. Instead of one synthesized paragraph with a few inline links, you get a short answer with numbered sources sitting right next to the claims. That layout tells you something useful: the citation is part of the product, not an afterthought. If your pages do not show up in those numbered slots, you are largely invisible in that answer, even if the model clearly knows something about your category.
This guide covers what Perplexity's own documentation says about its crawlers, what kinds of pages tend to earn citations, and how to check your current position without guessing.
The two bots that matter, and why they are not the same
Perplexity publishes a bot guide at docs.perplexity.ai/guides/bots. Two user agents do most of the work you care about.
PerplexityBot surfaces and links websites in Perplexity search results. Perplexity states it is not used to crawl content for foundation models. In other words, this is the crawler tied to the search product — the one whose access determines whether your pages can be retrieved and cited.
Perplexity-User fetches pages when a user asks Perplexity a question. It generally ignores robots.txt, because the user requested that fetch. This is the same pattern you see elsewhere: an agent acting on a live user request is not the same thing as an indexing crawler.
That distinction has a practical consequence. Blocking PerplexityBot in robots.txt to keep content out of some pipeline also removes you from the search surface where citations happen. Blocking Perplexity-User is largely beside the point, since it may ignore the file anyway when a person is waiting on the answer.
One more detail worth knowing: Perplexity says robots.txt changes can take up to 24 hours to take effect. If you flip a rule and immediately test, you may be looking at stale behavior.
And a reminder about what robots.txt is. Under RFC 9309, a crawler follows the most specific group that names it and ignores the * group in that case; within a group, the longest matching rule wins. It expresses intent. It is not access control.
A quick robots.txt sanity check
If your goal is to be cited, the usual mistake is over-blocking. Look for a group naming PerplexityBot with Disallow: / and ask whether that was intentional. If it was not, you have found the cheapest possible fix.
What Perplexity-style answers tend to reward
Perplexity does not publish a ranking formula, and anyone who claims to have reverse-engineered one is guessing. What you can reason about is the shape of the output: a direct answer, followed by numbered sources that support specific claims.
That shape favors pages that are easy to lift a claim from.
Direct answers near the top. A page that opens with three paragraphs of background before answering the question gives a retrieval system nothing clean to quote. State the answer, then explain it.
Specific, checkable statements. "Most teams see results in about six weeks" is quotable. "Results vary" is not. If you use a number, be able to defend it — and if it is illustrative rather than measured, say so on the page.
Freshness signals that are real. Dates that update because the content changed, not because a template rotated them. Pages that reference current versions, current pricing, current product names. Perplexity answers frequently read as current-state summaries, so stale pages lose to current ones on the same topic.
Clear entity naming. If your brand is "Acme" and your page says "we" and "our platform" throughout, a system assembling an answer has less to work with. Name the brand and the product in the body text, not just the title tag.
Comparison and definition content. "What is X," "X vs Y," and "best X for Y" formats map cleanly onto the questions people ask. These are also the formats where a competitor can be cited instead of you, which makes them worth owning.
Where the numbered sources come from
Not every cited source is a page you wrote. Perplexity will cite review sites, forums, documentation, news, and competitor pages. This is the part that surprises founders: you can have a strong site and still lose the citation to a third-party comparison article that mentions you in passing.
Which means Perplexity SEO is partly about your own pages and partly about the pages other people write about you. If a widely-read comparison page describes your product inaccurately, that inaccuracy can travel into answers.
Crawling is not the same as being chosen
A common assumption is that if PerplexityBot can crawl you, you will be cited. Crawling is a prerequisite, not a selection mechanism. Plenty of crawlable pages never appear as sources, because they do not answer anything cleanly or do not look authoritative on the topic.
So separate the two questions:
| Question | What to check |
|---|---|
| Can Perplexity reach and index my pages? | robots.txt rules for PerplexityBot, page renderability, no accidental noindex |
| Would Perplexity want to cite this page? | Direct answer up top, specific claims, current information, clear brand naming |
Most teams fix the first one in an afternoon and then spend months on the second, which is the correct order.
Testing your position instead of assuming it
You cannot see Perplexity's internal source selection, but you can see its output. Our free AI visibility check generates three buyer questions for a brand and category, asks Perplexity with web search enabled, and shows whether the brand is named, which competitors are named, and which sources are cited. It gives you 3 free checks per day. It does not track over time, so treat it as a snapshot: run it, read the cited sources, and note which of them are pages you control.
Two things to watch for in the results.
First, whether you are named but not cited. If the answer mentions your brand while sourcing a third-party page, your problem is page quality or third-party coverage, not crawling.
Second, which competitors appear and from which sources. If a competitor is consistently cited from their own documentation or a comparison page they wrote, that is a content format you can copy. If they are cited from a review site, that is a different project.
If you want to understand how this differs from other assistants, How ChatGPT Search Picks the Sources It Cites covers the ChatGPT side, and Google AI Overviews vs ChatGPT: Two Different Games explains why AI Overviews behave differently again — they are part of Google Search and governed by normal Googlebot crawling and indexing, not a separate opt-in.
A note on what is not documented
Perplexity does not publicly document its ranking signals, how it weights freshness, or how it decides between two equally relevant sources. Anyone presenting a definitive list is filling gaps with inference. What is documented is the bot behavior above, and that is where you should start, because it is the part you can verify.
The rest is experimentation: publish a page in the format the answer seems to want, wait, and check whether it gets picked up.
Do this next
Pick one question your buyers actually ask — ideally a "what is" or "X vs Y" query where you currently lose. Run it through the free AI visibility check, write down every cited source, and mark which ones you control. Then rewrite your closest matching page so the direct answer sits in the first two sentences, the brand and product are named in the body, and every specific claim is one you can stand behind. Recheck after a few days; robots.txt and index changes can take up to 24 hours to settle, so do not judge the result within the hour.