Perplexity SEO: how to get cited in Perplexity's answers
Perplexity crawls and cites live. What PerplexityBot actually does, which SEO myths don't hold up, and how to track whether your brand shows up in its answers.

The short answer: Perplexity crawls your site with its own bot (PerplexityBot), cites sources with real URLs in every answer, and has no secret markup requirement. The basics are the same old SEO basics: crawlable, factually accurate, well structured. What is genuinely different from ChatGPT or a Google AI Overview is how you measure it, and that difference is exactly what this article is about.
We already wrote about how to rank in ChatGPT and about ranking in Google AI Overviews. Perplexity deserves its own article, because it's the only one of the four platforms GEO-Crafter measures (ChatGPT, Gemini, Claude and Perplexity) that returns its own separate list of source URLs with every single answer, and that has practical consequences for how you can track visibility there.
What Perplexity does differently
Perplexity isn't a chatbot answering from trained memory alone: for most questions, it actively fetches web results and builds an answer with citations around them. Perplexity doesn't publicly document its exact internal pipeline; several independent analyses describe it as a form of retrieval-augmented generation, but the precise steps (how many sources, exactly how ranking works) aren't confirmed by Perplexity itself. What is confirmed: the answer you get comes with clickable footnotes to the actual pages used, not just a summary with no sources attached.
PerplexityBot: what the documentation actually says
Perplexity runs its own crawler, PerplexityBot, and the official documentation (docs.perplexity.ai) is reasonably concrete about it: the bot honors robots.txt directives and won't index the full or partial text of sites that disallow it via robots.txt. Just as important, and often confused with training data for large language models: according to the current documentation, PerplexityBot isn't used to collect content for training AI foundation models. That's a policy statement from Perplexity itself, not something you can verify from the outside, and like any company policy the text can change; check the current documentation if this matters for your decision.
Blocking it uses the standard syntax:
User-agent: PerplexityBot
Disallow: /
That's the same rule as for any other crawler: add it and the bot stops coming in. That removes your own site as a source Perplexity can cite directly, though your brand could in theory still get mentioned through other, still-accessible sources. Not blocking it is the precondition for showing up as a source yourself, not a guarantee that it happens.
Perplexity-User: the second, lesser-known crawler
Alongside PerplexityBot, there's a separate user agent called Perplexity-User, which Perplexity's own crawler documentation lists as distinct from PerplexityBot: where PerplexityBot crawls ahead of time and builds an index like a classic search engine, Perplexity-User fetches a page at the moment a user's question points to it. Exactly how that user agent behaves around robots.txt is described more briefly in the official documentation than you'd like, and third parties fill that gap with assumptions that don't agree with each other. Don't take any single claim about it, including anything in this article, as a settled fact, and check the current text at docs.perplexity.ai if it matters for your situation.
Is there special markup for Perplexity?
There's even more uncertainty here than with Google's AI Overviews, where Google at least explicitly says "no." As far as can be found in the official documentation, Perplexity hasn't made its own public statement about llms.txt or about required schema.org markup. Llms.txt itself is a third-party proposal (llmstxt.org), not a standard officially recognized by Perplexity or any other major AI platform. A single test by Search Engine Roundtable, published in early 2026, deliberately placed made-up data inside JSON-LD schema on a page and found no clear signal that ChatGPT or Perplexity used that schema data the way schema.org markup is designed to be used. It was one controlled test on one page, not a broad measurement, and informal experiments like this age quickly. Treat it as a useful correction to the assumption that schema automatically buys you "AI readability," not as the final word. Practically: build schema.org markup for ordinary SEO reasons (rich snippets, clarity for search engines), not in the hope that it drives your Perplexity citations.
What demonstrably works
The research that introduced the term generative engine optimization (Aggarwal et al., 2023) tested nine concrete content changes, such as adding citations, backing claims with statistics, and sharpening writing style, and found a visibility gain of up to 40% in their test setup (a best case, not an average, and the effect varied by domain). That research didn't look at Perplexity specifically, but at generative search systems in general; the core line (clear, citable claims, backed up with numbers, a direct writing style) is exactly what a system that cites live sources, as Perplexity does, can act on. Concretely, with no guarantees: make sure your page is crawlable and indexable, answer the question directly and early in the text, back up claims with a number or source instead of a vague statement, and make entities (who you are, what your product does) explicit rather than implicit.
The Publishers Program: context, not an action item for most sites
Perplexity launched a Publishers Program in July 2024: publishers share in the revenue when Perplexity earns money from an interaction that uses their content, with titles like TIME, Der Spiegel and Fortune as first partners. That's worth knowing, but it isn't practical SEO advice: the program runs through separate agreements with individual publishers, not a signup form any site can fill in. For most businesses that want their brand to show up in Perplexity's answers, this doesn't change anything about the basics above.
How to track whether you show up in Perplexity's answers
This is where Perplexity diverges most clearly from a Google AI Overview. For an AI Overview you have Search Console, with a dedicated report and a search appearance filter (see ranking in AI Overviews); for Perplexity there's no comparable official dashboard for site owners. Two routes remain:
- Manual: ask a fixed set of buying questions from your market directly to Perplexity, in a new session with no conversation history, and log whether and how your brand gets mentioned. We describe this approach in more detail in what is AI visibility and how do you actually measure it.
- Via the API: Perplexity's own API returns structured citation fields with actual source URLs on every answer, not just plain text. OpenAI's standard chat API doesn't do that the same way unless browsing or a separate retrieval layer is switched on. That makes it easier, with Perplexity, to check programmatically whether a specific URL from your site was actually used as a source, instead of only guessing from the wording of the answer.
Perplexity, ChatGPT and a Google AI Overview side by side
| Perplexity | ChatGPT | Google AI Overview | |
|---|---|---|---|
| Live web crawl per question | Yes, almost always | Only with browsing enabled | Yes, runs on the existing search index |
| Source URLs in the answer | Yes, as separate annotations/footnotes | Variable, not always visible | Yes, as clickable links |
| Official tracking report for site owners | None | None | Search Console (AI Overview report) |
| Own crawler with public robots.txt rules | PerplexityBot (+ Perplexity-User) | GPTBot | Googlebot (existing rules) |
An example (illustrative)
Say a B2B services company notices competitors show up more often in Perplexity's answers to a buying question in their market. They check via the API which source URLs Perplexity is actually citing, and find that a competitor's page backs up the same claim with a concrete, recent figure early in the text, while their own page only gets to that evidence in the third paragraph. After moving that evidence up and adding a source reference, they measure again a few weeks later. This example is illustrative and isn't a result you should automatically expect: AI answers are non-deterministic, and OpenAI itself describes its own models' output that way explicitly (developers.openai.com), which holds for generative search systems in general.
How we measure this
Our paid scan asks the same buying question in parallel to ChatGPT, Gemini, Claude and Perplexity, with a question set you freeze. For Perplexity, we use the actual source annotations the API returns to determine which URLs were really used, instead of only scanning the plain text of the answer. No ranking guarantee, no potential score: a coverage percentage based on real answers. Repeatable over time under matching conditions, with the caveat that AI answers are non-deterministic: any single result can vary from one call to the next, which is exactly why we look across multiple measurements instead of one snapshot. Curious where you stand right now? Request access or take a look at the GEO audit first.
Frequently asked questions
What is PerplexityBot, and should I block it? PerplexityBot is the crawler that indexes pages so Perplexity can link to them in answers, similar to a regular search engine crawler. Perplexity states that this bot is not used to collect content for training AI models. You can block it via robots.txt (user-agent PerplexityBot, disallow), but then you also stop appearing in Perplexity's answers.
Do I need llms.txt to show up in Perplexity? No, and there is no official statement from Perplexity naming llms.txt as a requirement. It's a third-party proposal (llmstxt.org), not an established standard. Ordinary crawlability and indexability remain the baseline.
Can I track whether my brand shows up in Perplexity's answers? Not through a dashboard from Perplexity itself: there's no Search Console equivalent. You can check manually with a fixed set of questions, or use a tool that calls the Perplexity API; unlike ChatGPT, that API returns the actual source URLs with every answer.
Is Perplexity SEO different from SEO for ChatGPT or Google AI Overviews? The basics (crawlable, factually accurate, well structured) overlap for the most part. The difference is in measurement: an AI Overview runs through Google Search Console, while Perplexity has no comparable official report but does cite a list of source URLs with every answer, which makes it measurable in its own particular way.