TL;DR
- Perplexity documents two agents: PerplexityBot builds its search index and follows robots.txt, while Perplexity-User fetches pages on a user's request and generally ignores robots.txt.
- Blocking PerplexityBot keeps full or partial page text out of the index, but Perplexity says the domain, headline and a brief factual summary may still be indexed.
- Perplexity publishes IP ranges for both agents as JSON files and says robots.txt changes may take up to 24 hours to take effect.
- Perplexity does not publish how it ranks sources or name the third-party crawlers it works with, so no documented rule ties a Google or Bing position to a citation.
- In GA4, citation clicks appear with source perplexity.ai. The default channel group has an AI Assistant channel, but Google's help page does not name Perplexity among its examples, so a custom channel placed above AI Assistant and Referral gives Perplexity its own line.
Perplexity SEO is the work of making your pages easy for Perplexity to crawl, read and cite when it answers a question. Perplexity runs a live web search for each query, writes a summary and attaches numbered citations that link to the source pages, so the practical goal is to become one of those cited sources. Perplexity does not publish how it ranks sources, which means honest perplexity seo rests on what you control: crawler access and pages that answer questions directly, plus measurement of the visits that follow.
This guide covers what Perplexity documents about its own crawlers and citations, what it leaves undocumented, and how to set up GA4 so that referrals from Perplexity are visible. The topic is one part of generative engine optimisation, and it sits alongside the work described on the AI SEO page.
How Perplexity picks and cites sources
Perplexity's help centre describes how the product works: it interprets the question, searches the web in real time, summarises what it finds, and cites its sources. Each answer carries numbered citations that link to the original pages so the reader can check a claim or read further. The help centre describes the sources only in general terms, as authoritative articles, websites and journals. It gives no list of ranking factors and no weighting.
Two consequences follow from that description. The first is that retrieval happens at the time of the question, so a page has to be reachable and readable by Perplexity's systems before it can be considered at all. The second is that the thing being cited is a page that supports a specific statement in the answer. A page that states a fact plainly, near a heading that matches the question, gives the system something it can attribute. A page that circles the topic without committing to an answer gives it nothing to point at.
Perplexity also offers a Deep Research mode. Its help centre says this mode runs dozens of searches and reads hundreds of sources before producing a report.
What cannot be said honestly is which signals decide the order of sources, how much a domain's reputation counts, or how freshness is scored. None of that is publicly documented. Any checklist that claims to reproduce Perplexity's ranking algorithm is guesswork presented as fact.
Does ranking on Google or Bing help you get cited?
There is no documented rule that a page ranking well on Google or Bing will be cited by Perplexity. Perplexity's own documentation describes PerplexityBot as the crawler that surfaces and links websites in Perplexity search results, and its help centre adds that Perplexity also works with third-party crawlers for its search index. It does not name those providers, and it does not describe how results from its own crawling and from third parties are combined.
That leaves a narrower and safer statement. The conditions that let a conventional search engine crawl, index and understand a page are conditions an answer engine needs too: the page returns a 200 status, the main content is in the HTML, and the title and headings describe the content. Work on those foundations serves both. What you should not do is assume the two move together query by query. Perplexity documents no link between the two, so a strong organic position does not guarantee a citation, and the only way to know is to test the questions that matter to you.
The test is simple. Write down ten to twenty questions a buyer would ask, in the words they would use, run each one in Perplexity, and record which domains are cited. Repeat it monthly. That log tells you more about your own category than a general theory does.
PerplexityBot crawling and robots.txt
Perplexity documents two user agents, and they behave differently.
- PerplexityBot is the crawler that surfaces and links websites in Perplexity search results. Perplexity states that it is not used to crawl content for AI foundation models. Its user-agent string contains "PerplexityBot/1.0" and a reference to perplexity.ai/perplexitybot. Perplexity says it respects robots.txt directives.
- Perplexity-User handles user-initiated actions. When a person asks a question, this agent may visit a page to help answer it and may link to that page in the response. Its user-agent string contains "Perplexity-User/1.0". Because the fetch is requested by a user, Perplexity's documentation says this agent generally ignores robots.txt rules. Perplexity states it is not used for web crawling or to collect content for training AI foundation models.
To allow the crawler, the robots.txt file needs a group for it that does not disallow the pages you want cited. The minimal form is a line reading "User-agent: PerplexityBot" followed by a line reading "Allow: /". If the file contains a blanket "Disallow: /" under "User-agent: *" and no specific group for PerplexityBot, the crawler is blocked along with everything else. Perplexity's documentation says changes to robots.txt may take up to 24 hours to take effect.
Blocking is not a clean off switch. Perplexity's help centre says that when a site disallows PerplexityBot, the full or partial text of the page is not indexed, but the domain, the headline and a brief factual summary may still be indexed. A blocked site can therefore still be mentioned, with less of its own wording behind the mention.
Robots.txt is only half of access. A block can also sit at the firewall or CDN, where a bot-protection rule challenges automated visitors before they reach the page. Perplexity publishes the IP ranges of both agents as JSON files on perplexity.ai (perplexitybot.json and perplexity-user.json) and recommends that firewall rules match the user agent together with those IP ranges, since a user-agent string on its own can be faked. Its documentation includes setup notes for Cloudflare and AWS WAF and advises refreshing the rules from the JSON files regularly.
A quick access check takes four steps:
- Open your robots.txt and look for any group that names PerplexityBot or disallows all agents.
- Check the bot settings in your CDN or security plugin for a rule that blocks or challenges AI crawlers by default.
- Search your server logs for the two user-agent strings and note the status codes returned. A run of 403 or 503 responses means the request never reached your content.
- Compare the requesting IP addresses with the published ranges before trusting that a hit is real.
Content formats Perplexity cites most
Perplexity has not published data on which content formats it cites most, so no format can be called the winner on evidence. What can be reasoned from the documented behaviour is this: the system searches, reads pages and attaches citations to statements in a summary. Formats that make a statement easy to find and attribute fit that process better than formats that bury it.
- Answer-first sections. Put the direct answer in the first one or two sentences under a heading that states the question. Supporting detail follows.
- Self-contained passages. Name the subject in the sentence instead of relying on "it" or "this" from three paragraphs earlier, so the passage still makes sense when read alone.
- HTML tables for comparisons. Prices, specifications, plan differences and checklists with outcomes are clearer as rows and columns than as prose.
- Facts you are the origin of. Your own prices, service areas, opening hours, policies and product specifications are things only you can state with authority. Publish them as text.
- Visible dates. A published date and an updated date let a reader, and any system reading the page, judge whether the information is current.
- Text in the HTML. Perplexity's documentation does not say whether its agents execute JavaScript. The safe assumption is that content rendered on the server is readable and content that appears only after scripts run may not be.
Perplexity's crawler documentation covers robots.txt and firewall rules. It does not list any special file, tag or markup as a requirement for being cited, so a claim that one particular file format earns citations has no support in the vendor's own pages.
How to track Perplexity referrals in GA4
When someone clicks a citation in Perplexity and the browser passes the referrer, GA4 records the session with the source perplexity.ai. No extra tagging is needed for that to work. The medium is less certain. GA4's default channel group has an AI Assistant channel, and Google Analytics Help says the medium is set to ai-assistant when the referrer matches Google's list of AI assistants. The examples that page names are ChatGPT, Gemini, Deepseek, Copilot and Grok, and Perplexity is not among them (as read on 16 November 2026), so the session may show the medium ai-assistant or the medium referral. To see which one your property records:
- Open Reports, then Acquisition, then Traffic acquisition.
- Change the primary dimension to Session source / medium.
- Type "perplexity" in the search box above the table.
- Add Landing page + query string as a secondary dimension to see which pages received the visits.
In the default channel group these sessions sit either in AI Assistant with other assistants or in Referral with every other referring website, depending on that medium. To give Perplexity its own line, create a custom channel group:
- Go to Admin, then Data display, then Channel groups, and create a new channel group.
- Add a channel and give it a name such as "Perplexity".
- Set the condition to Source matches regex and enter perplexity\.ai.
- Move the new channel above AI Assistant and Referral in the list. Google Analytics Help says traffic is included in the first channel whose definition it matches, given the current order of channels in the group.
- Save, then choose the new channel group in the Traffic acquisition report.
This data has limits. A visit that arrives without a referrer, for example from an app that does not pass one or from a copied link, is counted as Direct. A person who reads the answer and never clicks produces no session at all. GA4 also shows the landing page and not the question that was asked. GA4 therefore measures clicks from Perplexity, which is a smaller number than citations in Perplexity. Server logs fill part of the gap: requests from Perplexity-User show that a page was fetched in response to a user's question, whether or not a visit followed.
If channel definitions were never reviewed after the property was created, a GA4 setup review is the place to fix that before reading AI referral numbers.
Perplexity vs ChatGPT vs AI Overviews optimisation
The three surfaces share most of the content work and differ in access controls. The table lists what each vendor documents.
| Platform | Agents the vendor documents | Documented control |
|---|---|---|
| Perplexity | PerplexityBot (search index) and Perplexity-User (fetch triggered by a user) | robots.txt for PerplexityBot. Perplexity-User generally ignores robots.txt. |
| ChatGPT search | OAI-SearchBot (search), ChatGPT-User (user-triggered) and GPTBot (model training) | robots.txt for OAI-SearchBot and GPTBot. OpenAI says robots.txt rules may not apply to ChatGPT-User, because its visits are user-initiated. Disallowing GPTBot is the training opt-out and is separate from search. |
| Google AI Overviews | Googlebot | Standard Search controls: the page must be indexed and eligible for a snippet. nosnippet, max-snippet and noindex apply. |
For Perplexity, access is managed per agent, and the user-triggered agent sits outside robots.txt control. For ChatGPT search, OpenAI separates the search crawler from the training crawler, so a site can allow OAI-SearchBot and disallow GPTBot. That subject has its own page on ChatGPT SEO. For AI Overviews, Google states there are no additional requirements beyond being indexed and eligible to appear in Search with a snippet, and that the Google-Extended token does not govern Search. Traffic from AI Overviews is reported inside the Search Console Performance report under the Web search type, so it cannot be separated the way a perplexity.ai referral can in GA4.
The shared work is the larger part: crawlable pages, direct answers and facts stated as text. The platform-specific work is small and mostly technical. It covers which agents are allowed, which firewall rules apply, and where the resulting traffic is reported.
What this means in the Thai market
For a business in Thailand the same rules apply, with a few local points. Bilingual sites should check that both the Thai and the English versions are reachable by PerplexityBot. Prices, branch addresses, opening hours and promotion terms that exist only inside an image or a social media post cannot be read as text, so publish them on the page in Thai as well. Sites that sit behind a CDN or a managed security service should confirm that the default bot rules are not challenging Perplexity's agents.
Rather than guessing how many people in Thailand use Perplexity, size the effort from your own data. Set up the GA4 channel described above, let it collect for a few weeks, and compare sessions and conversions from perplexity.ai with other referral sources. Run your question log in both Thai and English and compare the cited sources.
Frequently asked questions about Perplexity SEO
Is perplexity seo different from regular SEO?
Perplexity SEO overlaps with regular SEO on crawlability and content quality, and differs in access controls and measurement. You manage two Perplexity agents instead of one search crawler, and you measure referral sessions and citations instead of ranking positions.
Should I block PerplexityBot in robots.txt?
Block PerplexityBot only if you do not want your pages surfaced and linked in Perplexity search results. Perplexity's documentation recommends allowing it for sites that want to appear, and its help centre notes that a blocked site's domain, headline and a brief factual summary may still be indexed.
Does Perplexity use my content to train AI models?
Perplexity's documentation states that neither PerplexityBot nor Perplexity-User is used to crawl content for training AI foundation models. Its help centre also says Perplexity does not build foundation models, so site content is not used for model pre-training.
How long do robots.txt changes take to apply to Perplexity?
Perplexity's documentation says robots.txt changes may take up to 24 hours to take effect. Check the server logs after a day before concluding that a change failed.
Can GA4 show which Perplexity question led to a visit?
No, GA4 shows the source, the medium and the landing page of a Perplexity referral, and it does not receive the question the person asked. To connect questions to citations you need a manual log of test prompts and the sources each answer cites.
Getting cited in Perplexity mostly comes down to removing access problems and publishing pages that answer real questions in plain text. If you would like help checking crawler access, content structure and AI referral tracking, the GEO service page explains how Relevant Audience approaches generative engine optimisation.







