How to get cited by Perplexity

The short answer

• Check search access first. Review PerplexityBot rules and verified requests at the CDN or firewall.
• Publish the answer with its evidence. Put the specific fact, procedure or comparison on a stable public URL, with dates and conditions beside the claim.
• Distinguish retrieval from citation. A page in a returned search result list has been found; it has not necessarily been used as a source in the final answer.
• Retest consistently. Save the same questions and settings before changing a page, then repeat them over time.

This guide is about making a page useful as a source. For a recurring measurement workflow, use how to track brand mentions in Perplexity.

1. Allow PerplexityBot and decide on Perplexity-User

Perplexity documents PerplexityBot as its search crawler and recommends allowing both the robots token and published IP ranges. Perplexity-User handles requests made for a user and generally ignores robots.txt. Its access therefore needs a separate decision.

For public pages you want eligible for search crawling, a simple rule is:

User-agent: PerplexityBot
Allow: /

Review the whole robots.txt file before adding it. A crawler-specific group can change how broader rules apply; preserve exclusions for paths that should remain private. Robots.txt is a crawler policy, not authentication or a security boundary.

Then check a real public URL and the layers before your application:

1. Confirm that the canonical page returns useful HTML without a login or challenge.
2. Inspect firewall and CDN rules for crawler blocks, rate limits and challenge responses.
3. Compare observed requests with Perplexity's published identity information. Do not grant a bypass merely because an arbitrary client says it is PerplexityBot.
4. Recheck after a deliberate policy change. Existing answers and cached references need not change immediately.

If you use Cloudflare, follow the AI Crawl Control guide. The AI crawler list helps separate search agents from training controls. Allowing one agent does not require allowing every AI crawler.

2. Give the answer a stable, useful source page

Choose one question that the page can answer better with evidence you actually have. A product comparison needs current scope and conditions; an instructional page needs steps someone can reproduce. Put that answer near the matching heading, rather than burying it under a company introduction.

For an illustrative documentation page about importing a file, include the supported format, the tested version, an example, failure cases and a review date. Keep important limitations with the answer so an extracted passage is still accurate. If a table compares limits, state the plan and billing period beside each value.

Link to primary evidence for external claims. Identify the author or reviewer and explain a test method where you report results. Distinguish your own observations from a provider's published behavior. These practices make a page easier for readers to verify; they are editorial recommendations, not a measured Perplexity ranking formula.

Check that the main text appears in the initial HTML and the canonical URL points to the intended page. Avoid maintaining several near-identical URLs with conflicting facts. The GEO audit checklist and free AI visibility checker help surface technical problems, while verified crawler logs answer a different access question.

3. Inspect retrieval through the API

This is an optional paid diagnostic. Manual app checks are enough to start, and no API call is required to use this checklist.

Perplexity's Agent API supports a web_search tool. When search runs, a search_results output item can contain the queries and results with IDs, URLs, titles, snippets and date fields. Keep the structured response alongside the answer instead of scraping every URL from generated prose.

Record the requested model, search settings and location with the prompt. A model may decide whether to search, so absence of a search output is not proof that your site is missing from an index. API output and the Perplexity app can also differ.

Use an unfiltered search for the baseline. A domain-restricted diagnostic can help investigate whether your page is returned, but it changes the candidate sources. Label that run separately; it cannot establish that you would win an ordinary competitive query. Do not silently combine different API products or modes into one trend.

4. Compare cited pages with retrieved pages

Match a reference in the answer to the structured source ID or URL where that mapping is available. Store retrieved results separately. A result list alone does not prove that the final answer used every page in it. If the response does not expose a reliable citation mapping, mark the citation status unknown.

For a retrieved page, look for a concrete missing fact before rewriting everything. Does the competing page name the exact version? Does it give a current price with conditions? Does your own answer require opening a widget? A more direct passage is a testable editorial change, but any later increase can still reflect other changes in retrieval.

5. Measure visits and repeat the checks

Use the same question set, mode and market across repeated checks. Keep the full answer, source list, cited URLs and run status. Report citations as a count with a denominator, such as an illustrative 3 of 12 completed answers, and report failed checks separately.

In analytics, inspect referrals containing perplexity and the pages people reach. Some app visits have no referrer, so those referrals are an observed subset of traffic. A Perplexity-User request in server logs is a fetch, not proof of a human click. Neither referral totals nor crawler requests count every citation.

Make one documented change at a time when practical, then compare several runs. Keep the previous result so you can inspect what changed. The cross-engine tracking guide provides spreadsheet fields and cautions about small samples.

Related guides

• How to get cited in AI search and how to get cited by ChatGPT.
• Track Perplexity mentions, ChatGPT mentions and Gemini mentions separately.
• Compare Ahrefs Brand Radar alternatives for ongoing research, or run the free AI visibility checker before buying a monitoring plan.

Sources

• Perplexity: Crawlers and firewall guidance
• Perplexity: Agent API web search and response shape

Reviewed on 2026-10-26. The access and response fields come from Perplexity's documentation. Content and measurement recommendations are practical guidance; no citation, ranking or traffic outcome is guaranteed.