Perplexity SEO Optimization: How to Get Cited by Perplexity

Perplexity SEO is the work of getting your pages used as cited sources inside Perplexity's answers — and the fact that reorders everything else is that Perplexity operates two different fetchers under two different rule sets. PerplexityBot is the one that builds search results: Perplexity's own crawler documentation describes it as “designed to surface and link websites in search results on Perplexity”, and recommends allowing it in robots.txt and permitting requests from its published IP ranges. Perplexity-User is the one that visits a page while a person's question is being answered, and the same page states that “since a user requested the fetch, this fetcher generally ignores robots.txt rules”. So robots.txt is the lever for being surfaced, not a reliable lever for being quoted — and the control that most often blocks a site from Perplexity is not robots.txt at all. It is the web application firewall in front of it, which is why Perplexity publishes WAF allowlisting instructions of its own.

What Perplexity SEO actually means

The phrase covers two jobs that are worth separating before anyone spends money. One is optimising your own site so that Perplexity retrieves and cites it. The other is using Perplexity as a research tool inside an SEO workflow. This page is about the first, which is also what people mean when they search for perplexity ai seo, perplexity seo optimization or how to get cited by Perplexity.

Framed that way the job is narrower than the vocabulary suggests. Perplexity publishes no submission form, no Perplexity-only ranking factor, and no schema type that buys admission. What it does publish is a crawler policy, two named user agents with different robots.txt behaviour, published IP ranges for each, and firewall configuration guidance. Those four things are the entire documented surface a site owner controls. Everything beyond them — passage structure, source quality, topical coverage — is inference from how the answers are assembled rather than from anything Perplexity has committed to in writing, and this page marks the difference each time it matters.

That distinction matters more here than on a conventional search page. Perplexity SEO advice circulates as confident checklists, and a good deal of it is guessing dressed as mechanism. The reliable core is small, so it is worth knowing exactly where it ends.

PerplexityBot and Perplexity-User: two fetchers, two rule sets

This is the structural fact almost every guide compresses into a single sentence about “blocking the bot”, and compressing it is what produces the wrong decision. Perplexity documents two distinct agents:

Agent	What Perplexity says it is for	robots.txt
PerplexityBot	“designed to surface and link websites in search results on Perplexity”. Perplexity adds that it “is not used to crawl content for AI foundation models”.	Obeys it. Perplexity recommends allowing it if you want to appear in results.
Perplexity-User	Supports user actions: when someone asks a question, “it might visit a web page to help provide an accurate answer and include a link to the page in its response”. Also stated not to be used for training foundation models.	“Since a user requested the fetch, this fetcher generally ignores robots.txt rules.”

Two consequences follow, and they point in opposite directions. If your aim is to be found, robots.txt is the right lever and PerplexityBot is the agent to allow — Perplexity's recommendation is explicit. If your aim is to stay out, robots.txt alone will not achieve it, because the user-triggered fetcher is documented as generally ignoring it. Publishers who disallowed PerplexityBot expecting silence and then saw their pages cited were not seeing a violation; they were seeing the second agent behaving as documented.

Both agents publish their IP ranges as JSON — perplexitybot.json and perplexity-user.json on perplexity.com — and Perplexity's guidance is to verify by user-agent string and source IP together, not by user-agent alone. That is ordinary anti-spoofing hygiene, and it is also the detail that makes the firewall section below work. Perplexity also notes it “may take up to 24 hours” for a robots.txt change to be reflected, so a same-day retest proves nothing either way.

How Perplexity builds an answer, and where a citation comes from

Perplexity's help documentation describes the sequence plainly: it interprets the question, “searches the internet, gathering information from authoritative sources like articles, websites, and journals”, compiles the relevant material into an answer, and attaches “numbered citations linking to the original sources”. Content, it says, “is sourced from the web in real-time as you ask your questions”.

Two practical readings come out of that. First, the unit that gets cited is a passage, not a page. An answer is assembled from material lifted out of sources, so a page that buries its answer inside a long unlabelled argument offers less to lift than one that states the answer under a heading naming the question. This is the same structural pressure that shapes generative engine optimization generally, and it is inference from how answers are built rather than published guidance — but it is inference with a low cost of being wrong, because the same structure helps ordinary readers.

Second, and more often missed: the model is not the retriever. Perplexity's model selector lets Pro users route a question to GPT-5, Claude 4.6 Sonnet and others, and Pro Search and Research run more searches and read more sources than a standard query. None of that is something a publisher can influence. Choosing a different model does not change which pages Perplexity can reach. Tuning your site for “the Claude model in Perplexity” is not a coherent goal; being reachable, indexed and quotable is.

Documented, versus widely believed

Four claims circulate as fact. Here is what Perplexity's own material supports.

Believed: blocking PerplexityBot in robots.txt removes you from Perplexity. Documented: it removes you from the crawler that surfaces and links sites in Perplexity's results. It does not stop Perplexity-User, which Perplexity states generally ignores robots.txt because a person asked for the fetch. Treat the two outcomes separately when deciding policy.

Believed: Perplexity is a front end on Google's or Bing's index. Documented: Perplexity documents its own crawler whose stated purpose is to surface and link websites in Perplexity's search results, and describes searching the web in real time. It publishes no statement that its web results are supplied by a third-party index. That is a notable contrast with Microsoft, which states directly that Copilot is “Powered by Bing's search index” — and it is the reason Copilot SEO reduces to Bing indexation while Perplexity SEO does not reduce to anything so simple. Note the shape of this claim: it rests on the absence of a published dependency, so treat it as the current documented position rather than a permanent architectural fact.

Believed: there is a schema type or an AI text file that gets you cited. Documented: nothing of the kind appears in Perplexity's crawler documentation, which covers user agents, robots.txt, IP ranges and firewall configuration and stops there. Structured data may earn its place for other reasons; it is not an entry ticket here.

Believed: a permissive robots.txt is sufficient access. Documented: Perplexity itself publishes WAF allowlisting steps for Cloudflare and AWS, which only makes sense because firewalls block its agents on sites whose robots.txt permits them. This is the gap that produces the most confusing symptom in Perplexity SEO — an allow rule in the robots file and no citations at all.

How to rank on Perplexity

Ordered by what can actually stop you, hardest blocker first. The first three are documented mechanism; the last two are structural inference, marked as such.

1. Allow PerplexityBot in robots.txt. Perplexity's recommendation, in its own words, is to allow the agent “to ensure your site appears in search results”. Check for a blanket disallow, and check any inherited rules from a platform default or a security plugin. Allow up to 24 hours before judging the result.
2. Allow both agents through the firewall, by user agent and IP. Covered in full below. This is the step that is skipped, and it is the one that silently costs the most.
3. Do not spoof-proof yourself into a block. Verify against the published IP JSON rather than user-agent string alone, and refresh those ranges on a schedule — Perplexity states they are updated regularly and should be treated as the source of truth.
4. Write answers as self-contained passages. Inference. Put the answer to a question directly under a heading that names the question, in a passage that survives being read on its own. This is the same discipline that wins featured snippets, and it costs nothing if the inference turns out to be wrong.
5. Be the kind of source the answer needs. Inference. Perplexity describes gathering from “authoritative sources like articles, websites, and journals” and citing “reputable news organizations, academic publications, and established content sources”. That is a description, not a ranking system, and no one outside Perplexity can quantify it — but it points the same direction as ordinary E-E-A-T work rather than away from it.

What is deliberately absent from that list: any Perplexity-specific markup, any submission step, any file to place at the root of your domain. If a checklist offers you one, ask which Perplexity document it came from.

The control most sites miss: the firewall, not the robots file

Perplexity's crawler page carries a full WAF configuration section, and the fact that it needs one is the finding. A modern edge firewall — Cloudflare's bot controls, AWS WAF, or a managed rule set from a host — can refuse Perplexity's agents while your robots.txt cheerfully allows them. The robots file is a request; the firewall is a wall.

Perplexity's own guidance is to build an allow rule that combines two conditions rather than either alone: user agent contains PerplexityBot or Perplexity-User, and the source IP is in the ranges published at the two JSON endpoints. For Cloudflare that is a custom rule under Security → WAF set to Allow. For AWS WAF it is IP sets plus string-match conditions, combined in rules placed at higher priority than the blocking rules, and associated with the Web ACL. Perplexity's stated best practice is to combine the user-agent match with IP verification precisely so that an allow rule cannot be abused by a spoofed agent string.

Two operational notes worth carrying into a real change window. Firewall changes propagate on their own schedule, so verify from logs rather than assuming. And this is the first thing to check when the symptom is no citations at all rather than fewer citations than a competitor: an access problem is binary and looks nothing like a content problem. If you are running a crawlability review for conventional search, adding the two Perplexity agents to it costs almost nothing.

What you can actually measure

Be honest about the instrumentation, because this is where Perplexity is weakest for practitioners. Perplexity does not publish a webmaster console. There is no Perplexity equivalent of the citation reporting Microsoft added to Bing Webmaster Tools, so nobody can hand you a verified count of your Perplexity AI citations.

What is available is indirect and still useful:

• Server and edge logs. The two user-agent strings are published in full, so PerplexityBot and Perplexity-User hits are identifiable in raw logs. This answers the access question — are we being fetched at all — which is the question that most often actually matters.
• Referral traffic. Perplexity links its cited sources, so cited pages can appear as referrals from perplexity.ai in analytics. Volumes are small and this undercounts badly, because a satisfied reader never clicks. Read it as a presence signal, not a traffic channel.
• Sampled prompting. Ask Perplexity the questions your pages target and record which domains it cites. It is manual, it varies run to run, and it is a sample rather than a measurement — but it is the only direct read of the outcome. Recording it on a fixed schedule is what turns it into a trend instead of an anecdote. General-purpose LLM visibility tools automate this sampling; they do not have privileged access either.

Anyone offering you a definitive Perplexity ranking report is reporting a sample. That is not a reason to ignore it — it is a reason to treat the sample size and the schedule as part of the method.

Perplexity versus ChatGPT, Gemini and Copilot

The four answer engines differ mainly in what they publish about themselves, which changes how much of the work is documented and how much is guesswork.

• Copilot is the most tractable, because Microsoft states it runs on Bing's index and ships citation reporting in Bing Webmaster Tools. Indexation is the entry ticket and you can see the result.
• Gemini sits inside Google's ecosystem, so Google's own crawling and indexing documentation carries most of the weight.
• ChatGPT publishes crawler documentation of a similar shape to Perplexity's, with distinct agents for indexing and for user-triggered fetches.
• Perplexity publishes a crawler policy and firewall guidance but no webmaster console. Access is well documented; outcome is not observable. That asymmetry is the whole character of Perplexity SEO — you can prove you are reachable and you cannot prove you are cited.

The overlap between all four is larger than the differences, which is the useful conclusion. Reachable, indexed, structured, credible pages serve every one of them, and no engine has published a control that rewards work done only for it.

Questions we hear

Is Perplexity good for SEO?

As a traffic channel, it is small and hard to attribute — Perplexity answers questions in place and links its sources, so a cited page earns visibility more reliably than clicks. As a target for optimisation it is reasonable, because almost all of the work is work you would do anyway: be crawlable, be reachable through your firewall, answer questions in self-contained passages under headings that name them. As a research tool inside an SEO workflow it is a different question again, and the honest answer is that its value there is the citation trail rather than the prose.

Does Perplexity use Google?

Perplexity does not document any dependency on Google's index. It documents its own crawler, PerplexityBot, whose stated purpose is to “surface and link websites in search results on Perplexity”, and describes searching the web in real time as questions are asked. Compare Microsoft, which states plainly that Copilot is “Powered by Bing's search index” — Perplexity publishes no equivalent statement. Because this rests on the absence of a published claim, treat it as the current documented position rather than a settled architectural fact.

How does Perplexity choose sources?

Perplexity has not published a ranking system, so nobody can answer this mechanically. What it does say is that it gathers “from authoritative sources like articles, websites, and journals” and that answers are “supported by citations from reputable news organizations, academic publications, and established content sources”. That is a description of the output, not a specification of the selection. Everything more specific you will read — passage length thresholds, freshness weightings, schema preferences — is inference. Some of it is sensible inference; none of it is documented.

What is PerplexityBot, and should I allow it?

PerplexityBot is Perplexity's crawler, identified by the user-agent string Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; PerplexityBot/1.0; +https://perplexity.ai/perplexitybot). Perplexity states it is “designed to surface and link websites in search results on Perplexity” and “is not used to crawl content for AI foundation models”. If you want to appear in Perplexity's results, allow it — that is Perplexity's own recommendation — and allow its published IP ranges through your firewall as well.

How do I get cited by Perplexity?

Clear the access gates first, because nothing else matters until they are clear: allow PerplexityBot in robots.txt, allow both agents through the WAF by user agent and IP, and confirm from logs that fetches are arriving. Then make the page easy to lift from — the answer stated under a heading that names the question, in a passage that reads correctly on its own. The first half is documented mechanism. The second half is inference, and worth doing regardless because it is the same thing that serves readers and conventional search.

Does blocking PerplexityBot keep my content out of Perplexity?

Not on its own. Disallowing PerplexityBot removes you from the crawler that surfaces and links sites in Perplexity's results, but Perplexity states that Perplexity-User “generally ignores robots.txt rules” because a user requested the fetch. If exclusion is a genuine requirement — a legal or licensing one rather than a preference — robots.txt is not the whole answer and the decision belongs with whoever owns that requirement.

Go deeper

Related: augmented intelligence SEO · ChatGPT SEO · Gemini SEO · Copilot SEO · LLM visibility tools · what GEO means in SEO.

Want the crawler-access audit, the firewall check and the citation sampling run for you? See how AI search visibility works as a module, or request a free AI-powered SEO audit as a first read.

Last updated 22 August 2026. Sources: Perplexity Crawlers documentation (docs.perplexity.ai), Perplexity Help Center — “What is Perplexity?” and “How does Perplexity work?”, both last updated 1 May 2026.