Getting Cited by Perplexity: What You Control and What Is Chance

Perplexity documents two agents. Its crawler page describes them as of October 2026:

• PerplexityBot crawls for Perplexity's search index. Perplexity states it is "not used to crawl content for AI foundation models."
• Perplexity-User fetches pages when a user's request triggers it. Perplexity states this fetcher "generally ignores robots.txt rules."

That's the documented part. Perplexity's crawler documentation explains who fetches your pages and why. It does not explain how Perplexity picks which sources to cite in an answer. Nothing in the evidence describes Perplexity's ranking factors.

So split the problem in two. Access is yours to fix. Selection is Perplexity's. The practical version of this page lives in our Perplexity optimization checklist.

robots.txt rules are per bot. Allowing GPTBot or ClaudeBot does nothing for Perplexity. If you want to appear in Perplexity's index, PerplexityBot needs its own permission. A minimal rule looks like this:

User-agent: PerplexityBot
Allow: /

The most common mistake is a blanket AI block copied from a template. It removes search crawlers along with training crawlers.

robots.txt is only the first gate. A firewall, CDN or bot-management rule can block a crawler that robots.txt allows. The crawler gets a 403 or a challenge page, and no dashboard tells you. The operators document this mechanism. How often it happens has not been measured.

The next move: request a page with the PerplexityBot user agent and look at the status code. The AI Crawler Checker requests a page as each AI crawler and shows which ones your firewall turns away. The same two-gate logic applies to getting cited by ChatGPT and getting cited by Claude, each with its own bot names.

This is the strongest evidence in the whole GEO field. Vercel's crawler study found that ChatGPT and Claude crawlers fetch JavaScript files but "don't execute them. They can't read client-side rendered content." Independent re-testing in June 2026 found PerplexityBot does not render JavaScript either.

What that means: if your pricing, product names or key claims load only after hydration, PerplexityBot may see an empty shell. Fixing this doesn't make Perplexity more likely to choose you. It decides whether your content exists for Perplexity at all.

The next move:

1. Fetch a key page with JavaScript disabled.
2. Compare the visible text with the rendered page.
3. List every fact that appears only after rendering.
4. Move those facts into server-rendered or prerendered HTML.

Check tabs, accordions and "load more" sections too. They often inject content the same way.

Stating a specific fact on a fetchable page gives an engine a sentence it can quote. Examples are a price, a number or a one-line definition. This is the only positive experimental signal for on-page content in the evidence.

Perplexity cites whole sources by integer index, not quoted text spans. A clear, self-contained fact still makes your page easier to use as a source for a specific claim.

Keep the facts consistent. If your pricing page, about page and footer disagree, an engine can repeat the wrong one. Say you run a hypothetical accounting SaaS that lists $39 a month on one page and $49 on another. Any engine that reads both has two answers to choose from.

Topical fit matters more than phrasing. The C-SEO Bench study and a 252,000-trial factorial both found topical relevance and list position were the main drivers of citation. Topical relevance means answering the question the buyer asks. Keyword stuffing is not the same thing. If AI tools misstate your facts, see how to check whether AI quotes your site accurately.

Perplexity's citations concentrate on a few sources. A 2025 study of AI Search Arena data (March to May 2025) measured citation concentration with a Gini coefficient. It found 0.77 for Perplexity, 0.83 for OpenAI models and 0.69 for Google. A higher number means fewer sources take more of the citations. The data is over a year old and covers news sources, so treat it as direction, not a current figure.

Brand prominence is a strong association with no lever attached. Nobody can buy it this quarter.

Position in Perplexity's candidate list also matters a great deal. C-SEO Bench found that moving a document higher in the model's context beat content rewrites. No publisher can set that position.

The route that holds up: when Perplexity cites other sites instead of yours, being named on those sites gets you into the answer. That is an association, not a promised effect. Start by finding the third-party pages AI engines cite, then weigh your own site against third-party mentions.

Most GEO tactics on the market were measured on pages already inside the model's context. The best-designed study, C-SEO Bench, found most conversational SEO methods "largely ineffective" and often harmful to ranking.

Don't count on these for Perplexity:

• Schema or FAQ schema. It is not a lever for AI answers. Existing markup does no harm.
• llms.txt. No engine documents reading it.
• Chunking or an "AI-friendly" style. Google rejects both, and nothing shows Perplexity rewards them.
• The "up to 40% more visibility" claim. It came from a lab engine with sources preloaded into context.

Injected content is a different category. Hidden instructions, prompt injection and cloaking for AI crawlers can change what an agent says once the content is retrieved. Research on search agents shows the effect is real. It is also illegitimate and a reputational risk, so don't use it. Watch for third-party pages that use it against your brand.

Run the checks in order. A later fix is wasted if an earlier gate is shut.

1. robots.txt. Confirm PerplexityBot is allowed, separately from training bots.
2. Firewall and CDN. Request a page as PerplexityBot and confirm a 200, not a 403 or challenge.
3. Raw HTML. Confirm your key facts appear without JavaScript.
4. Suppression. Check for accidental noindex, nosnippet or data-nosnippet from a template.
5. Answers. Ask Perplexity your buyers' questions and note who it cites.

Step five gives a snapshot, not a trend. Perplexity's free app uses an auto-router and paid plans offer a model menu (secondary reports, September 2026), so answers vary by setup. The free checker on this site scores three buyer questions, which is a fair first look before deeper work.