To rank in Perplexity AI, you have three jobs. First, let Perplexity’s crawler, PerplexityBot, into your site, in robots.txt and in your CDN or firewall, so your pages get into its search index. Second, write pages whose individual passages answer a question cleanly, because Perplexity retrieves and ranks pieces of pages, not just whole pages. Third, keep those pages accurate, current and backed by named sources, and cover the follow-up questions a full answer needs. None of this guarantees a citation: Perplexity’s answers change from run to run and depend on the model and search mode. It removes the reasons Perplexity would pass you over.

The channel is big enough to bother with. Perplexity CEO Aravind Srinivas said at Bloomberg Tech in June 2025 that Perplexity handled about 780 million queries in May 2025, growing about 20% month over month (Search Engine Land). And Perplexity shows its sources: its help center says each answer includes numbered citations linking to the original pages, so a citation is a visible, clickable credit.

Everything below comes from Perplexity’s own documentation and published research unless it says otherwise. For the ChatGPT and Gemini side, see How to Get Cited by ChatGPT, Perplexity, and Gemini.

How Perplexity finds, ranks and cites sources

Perplexity runs its own search engine. In a September 2025 research post, its search team described the system behind both its answer engine and its Search API:

  • The index. It tracks over 200 billion unique URLs. PerplexityBot is the crawler, and a machine-learning model decides which URLs to index and when to re-index them, based on how important a page is and how often it is likely to change. Perplexity’s help center adds that it also works with third-party crawlers to build the index.
  • Parsing into passages. A content-understanding module extracts the meaningful content of each page and splits it into “self-contained spans,” each of which can be retrieved and ranked on its own at query time. Pages built around lists and tables may get more formulaic extraction rules.
  • Retrieval and ranking. Each query searches the index lexically (by words) and semantically (by meaning) and merges the results. Filters then remove “clearly non-responsive or stale content,” and several ranking stages, ending with cross-encoder reranking models, narrow the set. Scoring happens at both the document and the sub-document level, so the answer model gets the most precise passages rather than whole pages.

Perplexity then writes an answer from those passages and cites them. How many sources it draws on depends on the mode. Its Pro Search help page describes standard search as fast answers from one or two sources, and Pro Search as running multiple searches and synthesizing dozens of sources. Pro Search is no longer a toggle: Perplexity applies it automatically up to the limits of the user’s plan. Deep Research, per the help center, “performs dozens of searches automatically” and “reads hundreds of sources” to write a report.

Two practical consequences follow. A page that blocks PerplexityBot can’t compete. And the unit that competes is the passage: a page earns a citation because one of its passages is among the best answers to the query, or to one of the sub-searches Pro Search runs.

Perplexity does not publish a list of ranking factors. Guides that name “factual density” or “entity relationships” as Perplexity’s criteria aren’t quoting Perplexity.

Let PerplexityBot in, and know what Perplexity-User does

Perplexity’s crawler documentation lists two user agents:

User agentWhat it doesFollows robots.txt?
PerplexityBotCrawls pages to “surface and link websites in search results on Perplexity.” Not used to crawl content for AI foundation models.Yes
Perplexity-UserVisits a page when a user’s question needs it, and may link it in the answer. Not a crawler.”Generally ignores” robots.txt, because a user requested the fetch

To be eligible, allow PerplexityBot. If no rule blocks it, you’re already fine. If you want to state it explicitly:

User-agent: PerplexityBot
Allow: /

Blocking it doesn’t make you disappear entirely. Perplexity’s robots.txt help page says PerplexityBot won’t index the text of a site that disallows it, but Perplexity “may still index the domain, headline, and a brief factual summary.” The same page says Perplexity doesn’t build foundation models, so your content won’t be used for AI model pre-training if you allow PerplexityBot.

Check your firewall and CDN, not just robots.txt. Bot protection can block crawlers you meant to allow. Perplexity recommends allowing PerplexityBot “and permitting requests from our published IP ranges,” and gives setup steps:

  • Cloudflare: a custom rule that matches each user agent against its own published IP ranges, with the action set to Skip, placed above any rule that blocks or challenges. Perplexity notes that Bot Fight Mode can’t be skipped this way and may still challenge its bots.
  • AWS WAF: an IP set per bot, combined with a User-Agent string match, in an Allow rule evaluated before your blocking rules.

The IP ranges change, so Perplexity publishes them as JSON files linked from the crawler page and recommends refreshing your rules from them automatically. Matching both the user agent and the IP also keeps impostors out, since anyone can claim to be PerplexityBot.

This matters more on Cloudflare. In August 2025, Cloudflare said Perplexity was using undeclared crawlers to get around no-crawl rules, de-listed it as a verified bot and added managed rules to block that crawling. Perplexity’s help center now says the feature that let users summarize a blocked URL has been disabled and that “PerplexityBot only crawls content in compliance with robots.txt.” Whatever you make of the dispute, check that your bot settings let the declared PerplexityBot through.

The AI Citation Readiness Checklist covers the same check for OpenAI, Anthropic and Google crawlers.

Write passages Perplexity can lift whole

Because Perplexity ranks spans as well as pages, write so that the passage answering a question can stand on its own:

  • Answer first. Put the direct answer, number or definition in the first sentence or two under each heading, then explain.
  • One claim per paragraph. A paragraph that mixes three ideas is a weak match for each of them.
  • Name things in full. Write “Perplexity’s Pro Search” rather than “it” when a passage might be read on its own.
  • Use tables and lists for comparable facts. Prices, specs, steps and side-by-side comparisons are easier to extract as a unit. “Tool A is great and Tool B is also great” gives an answer engine nothing to cite. “Tool A: $29 per user, two-day onboarding. Tool B: $49 per user, two-week onboarding” does.
  • Give specifics with sources. A claim with a number, a date and a named source can be checked and attributed. “Studies show 73% of marketers…” can’t.

The best-known research on writing for answer engines backs the last point, with caveats worth knowing. In the GEO paper by Princeton and IIT Delhi researchers (Aggarwal et al., KDD 2024), rewriting source pages to cite sources, add quotations or add statistics raised their visibility in answers by roughly 30–40% on the authors’ GEO-bench benchmark, using a simulated answer engine. The authors also ran the methods on Perplexity.ai, on 200 test queries. Because Perplexity didn’t let them choose the sources, they uploaded the source text as files. There, quotation addition did best on their word-count measure (about 22% over baseline), statistics addition raised their subjective-impression score by 37%, and keyword stuffing did 10% worse than no change on that word-count measure.

Read that carefully. The Perplexity test measured how Perplexity used text it had already been handed, in 2023. It says nothing about how Perplexity finds or ranks your page today. It does support writing that is specific, sourced and quotable, and it gives no support for keyword stuffing.

Two quick exercises help. Take any paragraph on a key page and count the claims a reader could check: a number, a date, a named source. If the count is zero, rewrite it. Then test your quotes. “Our CEO believes AI is transforming the industry” could come from anyone. A usable quote follows this template: “[Specific, checkable observation with a number],” said [Name], [Title] at [Company], in [named public source, date]. If you could swap in any other executive and the quote would still work, it’s too generic to cite.

Cover the questions behind the question

Pro Search and Deep Research break a question into several searches, so a full answer draws on passages about the sub-questions too. For a query like “best payroll software for a 20-person restaurant,” the sub-questions include pricing, tip handling, point-of-sale integrations and setup time. A page, or a small cluster of linked pages, that answers each of those clearly gives Perplexity more chances to cite you than a single overview.

Three habits help:

  • Map the follow-up questions. Run your main queries in Perplexity, note the sub-topics its answers cover and the follow-ups people ask, and make sure each has a clear heading and answer somewhere on your site.
  • Publish what only you have. Your own pricing, specs, policies, survey results or benchmark data make you the primary source, and anyone else repeating those numbers is a secondary summary. That’s a stronger position than paraphrasing what ten other pages already say.
  • Build comparison pages with explicit criteria. A comparison answer needs criteria and data for each option. Pages that state both give Perplexity something to cite for each row.

Keep pages current, honestly

Perplexity’s ranking pipeline filters out stale content early, and its indexer schedules re-crawls around how often a page is likely to change. Outside data points the same way. In a July 2025 Ahrefs study of about 17 million citations by ChatGPT, Perplexity, Gemini, Copilot and Google’s AI Overviews, compared with Google’s organic results, pages cited by Perplexity averaged 1,166 days since publication, against 1,416 days for Google’s organic results. Perplexity’s in-text citations also tended to list newer pages first. That’s a correlation, not proof that updating causes citations, and Ahrefs notes that pages cited by AI assistants were still 2.9 years old on average.

The practical rule: update key pages when the facts change, with new figures, new examples and corrected details, and change the modified date only when the content actually changed. The AEO Content Refresh Playbook covers how to run a review cycle.

What Perplexity’s source labels tell you

Some Perplexity citations now carry a shield icon with one of three labels: Government, Academic or Trusted. According to Perplexity’s help page on source labels, a review process rates whole websites, not pages, using questions such as whether the site corrects its mistakes, says who wrote each piece, and keeps news separate from advertising and opinion. Perplexity says partnerships, payments and other business arrangements don’t affect labels, and that most domains have no label, which “does not mean a site is low quality.”

Perplexity doesn’t say labels affect which sources get cited, so don’t treat them as a ranking factor. The criteria are still a useful checklist for any site that wants to be quoted: named authors, a visible corrections practice, and sponsored content clearly marked. Third-party coverage helps for a similar reason: reviews, industry press and directory listings give Perplexity independent pages that describe you. See Entity SEO for AI Knowledge Graphs and Review Management for AI Visibility.

Comet, the API and the publisher program

Comet is Perplexity’s Chromium-based browser, available for Mac, Windows, iOS and Android. Its help center says it uses Perplexity as its primary search engine and has buttons to ask about or summarize the page you’re on. Comet isn’t one of the user agents on Perplexity’s crawler page, so there’s no separate Comet rule to set in robots.txt. For site owners, Comet mostly means more searches start in Perplexity, where the same index and ranking apply.

The API. Developers, and many AI-visibility tools, reach Perplexity through its API. Perplexity’s migration guide says support for Sonar Chat Completions ended on September 27, 2026, with synchronous and streaming requests being reformulated as Agent API requests; the Sonar model is still offered inside the Agent API. A separate Search API returns ranked web results without a written answer. Perplexity’s API FAQ says the API “uses the same search system as the UI with differences in configuration,” and the model may differ too, so API results can differ from what people see in the app. Keep that in mind when a tracking tool reports your Perplexity visibility.

The publisher program. Perplexity’s Publishers’ Program (July 2024, starting with TIME, Der Spiegel, Fortune and others) promised partners a share of ad revenue from answers that referenced their content. In August 2025 it announced Comet Plus, a $5 subscription (included in Pro and Max) whose revenue goes to participating publishers, “minus a small portion for Perplexity’s compute costs,” allocated by human visits, search citations and agent actions. Launch partners included Condé Nast, Le Monde, the Los Angeles Times and the Washington Post. These programs are for publishers with content to license and are about getting paid for citations; Perplexity doesn’t say membership changes which sources it cites.

How to measure your Perplexity visibility

  • Fix a prompt set. Write 20–50 questions your buyers actually ask, including comparison and “best X for Y” questions. Run them in the Perplexity app and record whether you’re cited, which URL, and which competitors appear. Run each prompt more than once, because answers vary.
  • Re-run on a schedule. Monthly is enough for most sites. Judge the trend over several runs, not a single result.
  • Check your server logs. Look for PerplexityBot (is it crawling your important pages?) and Perplexity-User (are users’ questions pulling your pages in?). Verify the IPs against Perplexity’s published ranges.
  • Watch referrals. Sessions referred from perplexity.ai in your analytics show who clicked through from a citation.
  • Treat API-based tools as a proxy. They’re useful for scale, but spot-check them against the app, since Perplexity says API results can differ.

How to Measure AI Citations covers the tools, and Competitor Analysis for AI Citations covers benchmarking against rivals.

What to skip

  • Keyword stuffing. It did worse than doing nothing on the GEO paper’s word-count measure in its Perplexity test.
  • Fake freshness. Changing dates without changing content.
  • Formatting folklore. There’s no evidence that a particular punctuation pattern, such as “Institution (Year): finding,” helps Perplexity parse a claim. Clear, sourced sentences are what count.
  • Perplexity Pages tactics. Perplexity’s help center says the “Create page” feature is temporarily retired.
  • Schema as a Perplexity lever. Perplexity’s documentation for site owners doesn’t mention structured data. Keep markup accurate for other reasons, but don’t expect it to win Perplexity citations. The structured data guide covers what the evidence shows.

Prefer to self-assess first? Download the 25-item AI Citation Readiness Scorecard (PDF), a printable self-assessment across schema, content architecture, authority, measurement and conversion. Free, instant download, no email required.

Want it done for you? The free AI Visibility Snapshot tests 10 prompts on ChatGPT, Google AI Overviews and Perplexity, with a short note on where you’re cited and where you’re not. The Citation Audit tests 30–50 locked prompts across six surfaces (ChatGPT, Google AI Overviews, Perplexity, Gemini, Claude and the OpenAI Responses API) and delivers competitor gaps, schema and content fixes and a prioritized 60–90 day roadmap in 5 business days. It’s $1,500, fully credited toward an Implementation Sprint or retainer signed within 30 days. We don’t guarantee citations or rankings. See pricing.

Request a free Snapshot or a Citation Audit →


FAQ

How does Perplexity decide which sources to cite?

Perplexity searches its own index, which tracks over 200 billion URLs, built by its crawler, PerplexityBot. It combines word-based and meaning-based retrieval, filters out stale or non-responsive results, and ranks candidates in several stages at both the page and the passage level. The answer cites the passages it used. Perplexity doesn’t publish ranking factors, so the reliable levers are being crawlable, answering clearly in self-contained passages, and being specific, sourced and current.

Should I allow PerplexityBot in robots.txt?

Yes, if you want to appear in Perplexity’s answers. Perplexity’s crawler documentation recommends allowing PerplexityBot and permitting its published IP ranges in your firewall or CDN. Perplexity says PerplexityBot is not used to crawl content for AI foundation models. If you block it, Perplexity says it may still index your domain, headline and a brief factual summary, but not your page text.

What is the difference between PerplexityBot and Perplexity-User?

PerplexityBot crawls the web to build Perplexity’s search index and follows robots.txt. Perplexity-User fetches a specific page when a user’s question needs it, and Perplexity says it generally ignores robots.txt because a person requested the fetch. To control whether you appear in search results, use rules for PerplexityBot.

How long does it take to show up in Perplexity?

There’s no fixed timeline. A page can be cited once PerplexityBot has indexed it and it’s among the best matches for a query. Perplexity’s indexer schedules crawls by how important a page is and how often it is likely to change, so timing varies by site. Run the same prompts before and after publishing, and judge the trend over several runs.

Do Perplexity’s source labels affect rankings?

Perplexity doesn’t say they do. Labels (Government, Academic or Trusted) go to whole websites after a review that checks things like corrections, bylines and separating news from advertising. Perplexity says payments and partnerships don’t affect labels, and that most sites have no label. The criteria are still good practice for any site that wants to be cited.

Does Perplexity’s Comet browser change how I optimize?

Not much. Comet uses Perplexity as its main search engine, so the same index, crawler rules and ranking apply. Comet isn’t listed as a separate crawler, so there’s no extra robots.txt rule to set. Make sure PerplexityBot can reach your pages and that your key passages answer questions clearly.

Sources

Related reading: The AI Citation Readiness Checklist, How to Get Cited by ChatGPT, How to Get Cited by Claude, How to Optimize for Google AI Overviews, How to Measure AI Citations.