How to Get Cited in Perplexity

Perplexity is the assistant most built around citations: nearly every answer comes with a list of sources. To be one of them, a page first has to be readable by Perplexity's crawler, then be retrieved for the question, then be chosen from what was retrieved. The first step is entirely in your control and is often broken. The second and third are not, and the evidence for specific tactics is thinner than the advice you will read online.

Key Takeaways

• Check access first. PerplexityBot indexes pages for Perplexity search results, and Perplexity says it does not crawl for foundation model training.
• Perplexity-User fetches pages when a person asks, and Perplexity says it generally ignores robots.txt, so a firewall rule is what stops it.
• Perplexity leaves a lot of what it retrieves uncited. In one 2026 study 64 percent of retrieved URLs went uncited, so retrieval is not citation.
• Comparison pages are used heavily on comparison questions, and about three quarters of the ones engines use are third-party.
• Each assistant has its own source mix, so a ChatGPT win does not transfer.

Step 1: Make Sure PerplexityBot Can Read You

Perplexity documents two bots. PerplexityBot is "designed to surface and link websites in search results on Perplexity". Perplexity-User supports a user's action inside Perplexity. The documentation says that if you use a web application firewall you may need to allow its bots explicitly, matching both the user agent and its published IP ranges.

Known Agents measured robots.txt compliance across more than 5,000 sites: a site-wide disallow cut verified PerplexityBot traffic by 99 percent or more. So a Disallow line works, which also means an accidental one works. Around 17 percent of the top 1,000 sites disallow PerplexityBot in that study. Our own site's robots.txt blocked AI crawlers for months without anyone noticing; the account is in we blocked AI crawlers for five months. Run the free agent access check to see what each assistant gets when it asks for your pages. The wider picture is in AI crawler blocking.

Step 2: Be Retrievable for the Question

Perplexity searches the web per question and then writes an answer from what comes back. Profound's April 2026 analysis found Perplexity keeps about 88 percent of the prompt's words in its most different search query, against 13 percent for ChatGPT. In practice this means Perplexity's searches stay close to the wording the user typed, and are more repeatable than ChatGPT's (see how often AI answers change).

That has a simple consequence. Pages that use the language buyers use, in titles and headings, are easier to retrieve for those questions. Controls are missing, so treat this as sensible practice rather than a proven lever. The evidence in other engines is correlational: in one ChatGPT analysis, titles sharing at least half the query's words were cited 20.1 percent of the time against 9.3 percent for titles with under 10 percent overlap.

Step 3: Be Chosen From What Was Retrieved

Perplexity does not cite everything it reads. A 2026 study found it left 64 percent of retrieved URLs uncited. Several patterns show up in vendor data:

• Comparison pages. In Searchable's October 2026 study of 1.8 million answers, comparison pages made up 13.4 percent of Perplexity's sources on comparison prompts and 1.1 percent on other prompts, a ratio of 10.9 times. About three quarters of the comparison pages engines used were third-party. Owning a comparison page did not show a mention lift over owning a guide.
• Review sites. Evertune found review domains made up 2.2 percent of Perplexity's citations, the highest of five engines it checked.
• Community content. Reddit is cited on Perplexity too. See does Reddit affect ChatGPT answers for the figures and why they vary.
• Domain authority. Featured's August 2026 report found 34.5 percent of Perplexity citations went to domains below DA 40. Authority scores do not seem to gate citation.

Perplexity's source mix also moves. Scrunch and Searchable both report that it largely stopped citing YouTube from around 12 August 2026.

What Has No Good Evidence

• Special files for Perplexity. No study shows llms.txt or a markdown copy lifting citations. See llms.txt doesn't work.
• Schema. See does schema markup help AI citations.
• Reusing a ChatGPT strategy. Each assistant has its own source mix, so a ChatGPT result says little about Perplexity. Check each one on its own, as the mention-tracking guide explains.
• Press releases and wire copy. Otterly found press releases and wire copy under 1 percent of publisher citations, and Muck Rack put wire domains at about 1 percent of all citations by December 2025.

A Practical Order of Work

1. Confirm PerplexityBot and Perplexity-User are not blocked by robots.txt or by the firewall.
2. Write down the 15 to 20 questions your buyers ask and run them in Perplexity several times. Note which sources appear and whether you are named.
3. For questions where you are absent, read the pages that are cited. They show the format and the claims Perplexity is using.
4. Where a cited page is third-party, treat getting accurate information onto it as a placement, pitch or listing task. If it is paid, it is labelled commercial content that complements earned coverage, not a replacement for it.
5. Re-run the same questions on a schedule and compare. One check is a coin flip. How often AI answers change explains why.

Lectern includes Perplexity in its checks on every plan, records the sources it cites for each of your buyer questions, and keeps the history. See the pricing page for what each plan covers.