Perplexity SEO: how to appear in Perplexity sources

What Perplexity does, and how it differs from Google

Perplexity's Help Center describes the process plainly. It searches the web in real time, pulls together what it finds, and writes an answer with numbered citations that link back to the original sources. Those numbered links are what you're after. Access comes first, because Perplexity can't quote the text of a page it can't read. Beyond that, neither the Help Center article nor Perplexity's crawler documentation says how it picks one page over another, so access is the part you can check for yourself.

Perplexity's crawler documentation names two user agents and treats them differently. PerplexityBot crawls sites so Perplexity can surface and link them in its search results, and Perplexity recommends allowing it. Perplexity-User visits a page when a person's question needs it, and Perplexity says it generally ignores robots.txt because a person asked for the fetch. Its Help Center article on robots.txt adds that Perplexity also partners with third-party crawlers to help build its search index, and has updated its agreements so those providers respect robots.txt too, particularly on news publisher sites. The article doesn't name them, so the log checks below can only cover the two named agents.

So what is Perplexity SEO? It's ordinary SEO plus three checks specific to Perplexity: its crawler can get in, its visits are real, and you track its answers over time. The table compares what each company publishes for website owners. Google's AI features guidance says a page needs no special markup or AI text file to appear in AI Overviews or AI Mode: it only has to be indexed and eligible to show in Google Search with a snippet. Perplexity and OpenAI each run their own crawler for search, so a robots.txt rule written for Googlebot does nothing for them. For the other two products in depth, see how to rank in ChatGPT and AI Overviews and AI Mode for local businesses.

Step one: confirm PerplexityBot can get in and read the page

A block isn't always deliberate. It can come from an old robots.txt rule, a security plugin, a hosting default or a firewall setting nobody remembers switching on. And a page that lets the crawler in can still hand it an empty shell, if the words you want quoted only appear after JavaScript runs in a browser. Work through these checks in order.

If your site runs through Cloudflare, check its AI bot settings as well. Cloudflare's changelog of 1 July 2026 says every customer, Free plan included, can block AI traffic by behaviour (Search, Agent or Training) on all pages or only on pages that show ads. New domains added from 15 September 2026 block Training and Agent traffic on ad pages by default. Cloudflare's AI Crawl Control bot reference lists both PerplexityBot and Perplexity-User, and its Crawlers tab shows the AI crawlers requesting your pages, with allowed and unsuccessful requests, robots.txt violations and the action applied. On the Free plan, Cloudflare says it spots AI crawlers by their user-agent name only.

If this is more than you want to do yourself, robots.txt, firewall and rendering settings sit squarely in technical SEO work, and a developer or host can make the change once they know which rule to add.

1. Open yourdomain.com.au/robots.txt in a browser. Look for a group that names PerplexityBot, and for the catch-all group that starts User-agent: *.
2. Work out which group applies. Under the robots.txt standard, RFC 9309, a crawler that finds a group naming it follows that group and skips the catch-all. So Disallow: / under PerplexityBot shuts it out even if everything else is open. With no PerplexityBot group, a blanket Disallow: / under the catch-all blocks it too. According to Perplexity's Help Center, a block stops PerplexityBot indexing the page's text, but Perplexity may still index the domain, headline and a brief factual summary.
3. Make sure robots.txt itself loads. The same standard says a crawler must treat a server error on that file as a block on the whole site.
4. Check the words are in the HTML your server sends. Right-click the page, choose View page source, and search for a sentence from your main answer. If it isn't there, the text is built by JavaScript. Perplexity's documentation doesn't say how much JavaScript its crawler runs, so the safer setup is to have the answer, the service area and the contact details in the HTML from the start.
5. Test the response a crawler gets. A developer can request the page with curl, sending the full PerplexityBot user-agent string from Perplexity's crawler page, and check for a 200 status and the same text. A firewall that matches IP ranges may still treat that test differently from the real crawler, because the test comes from your own address.
6. Check your firewall, CDN and security plugins. Perplexity's documentation recommends allow rules that match both the user-agent name and its published IP ranges, with steps for Cloudflare and AWS WAF.
7. While the file is open, review the other AI crawlers in it. OpenAI lists OAI-SearchBot for search and GPTBot for training, Google lists Google-Extended for Gemini training and grounding, and Anthropic lists ClaudeBot, Claude-User and Claude-SearchBot. Each needs its own group, and a rule for one doesn't touch the others.
8. Give it a day. Perplexity says robots.txt changes can take up to 24 hours to reach its systems, and firewall changes can take time to spread.

Step two: prove the visits are real

An open robots.txt only gives permission. The proof sits in your server access logs, or your host's log viewer, where each request shows a date, an IP address, a user-agent name and the status code your site sent back. Logs and IP checks can only confirm visits that identify themselves. A request sent under an ordinary browser name from an unlisted IP address won't look like Perplexity at all.

A verified crawl shows Perplexity can read the page. It doesn't show Perplexity used the page in an answer, and a human visit is a third thing again. The guide on why a crawl does not prove an AI citation covers that gap.

1. Search the logs for PerplexityBot and Perplexity-User.
2. Check each IP address against the matching JSON list on perplexity.com: perplexitybot.json for the crawler and perplexity-user.json for the fetcher. Anyone can type a user-agent name into a script. Google's guidance on verifying its own crawler warns about the same trick, with fake visitors claiming to be Googlebot.
3. Note the status code. A 200 means the page was served. A 403 or 503 means something turned the visitor away, whatever robots.txt says.
4. Note which pages it requested. Visits to your home page tell you little about the service page you want quoted.
5. Write down the date you checked and the creationTime value at the top of the JSON file you compared against. Perplexity says its IP lists are updated regularly, so an old copy can make a real visit look fake.

Step three: shape each page so a passage can be quoted

Once Perplexity can read the page, make the answer easy to lift out. Perplexity doesn't publish how it chooses which sentence to quote, so treat this as sound writing practice rather than a documented Perplexity factor: put the customer's question in a heading, answer it in the first sentence underneath, then add the detail that only you can give. Then test each answer: read the paragraph under a heading without the one above it, and if it only makes sense with that context, add the missing service name, place or fact.

Say a pest control business in Maroochydore has a termite page that opens with a paragraph about the company's history, and only mentions inspections halfway down. The table shows a rewrite. Nothing new is claimed. The same facts move to where a question lands.

Keep the page current, and only say it changed when it did. Google's sitemap documentation says Google uses the lastmod value if it's consistently and verifiably accurate, and counts changes to the main content, structured data or links as significant, but not a new copyright date. That's Google's rule, not Perplexity's, but an honest date is worth having everywhere. The guide on keeping AI visibility content current sets out a review routine.

• Show an updated date on the page and move it only when the content changes; a date that moves while the text stays the same tells readers nothing.
• If the page carries Article structured data, keep its dateModified the same as the visible date.
• Keep each URL's lastmod in your XML sitemap accurate, so a crawler that reads it isn't misled.

Step four: keep dated source observations

Then watch the answers. Perplexity writes each answer fresh, so the sources it cites can shift from one month to the next. One screenshot proves very little. A dated log of the same questions, asked the same way each time, shows a pattern you can act on.

Set each check up so it can be repeated. Perplexity's own documentation says follow-up questions carry the context of earlier ones, memory can personalise answers, and without a set location, answers are based on your network location. So start a fresh thread for every question, test signed out or in incognito mode, where memory is switched off, and note where you tested from. A search from your office, signed in, isn't a neutral test, which is often why owners see their business in an answer their customers never get.

Add one column most logs leave out: for the same question, note whether your page ranks in Google and Bing. Perplexity doesn't say whether it draws on either engine, so don't assume a link. Your own log will show whether the pages Perplexity cites for your questions are the ones that rank there, or different ones.

Summarise the log in a few plain measures: how many checks cited your site out of the total, which of your pages were cited, and which other domains were cited when you weren't. For clicks, open GA4's Traffic acquisition report and look for a session source containing perplexity. GA4 Help defines a source as the publisher or inventory source the traffic came from, and a visit only shows there if the referrer reaches your site. The guide on measuring ChatGPT referrals walks through the same report.

What your log is telling you

Read several dated checks together before you change anything. Then match what you see to a next step.

Pay close attention to the other domains in the citation list. For a local service question, those are often directories, review sites, local news, councils or industry bodies. You can't control whether Perplexity cites them, but you can make sure your business details there are correct and current, so an answer drawn from them still gets your facts right. Start with citations and business listing accuracy.

When you compare your page with the ones that were cited, look at what Perplexity itself says it values in a website. Its Help Center article on source labels says the review behind those labels asks questions such as whether a site corrects its mistakes, says who wrote each piece, and keeps news separate from advertising and opinion. The article doesn't say labels decide which pages get cited, and most sites carry no label at all. Even so, a named author, a visible updated date, clear contact details and an easy way to report errors are sound standards for any page you want quoted.

Wrong details in an answer are a separate problem with its own fix. Start with AI answers and accurate company facts, then AI source correction and retraction.

Worked example: a Maroochydore pest control business

This is a hypothetical business, and every count below is illustrative only. It isn't a benchmark or a typical result. It shows the method end to end.

Say the owner of a pest control business in Maroochydore asks Perplexity which local companies do termite inspections, and the business isn't in the sources. Here's how the guide's steps play out.

You control access, the clarity of your pages, the accuracy of your business facts and the quality of your log. Perplexity controls the answer, and no one can guarantee that it will cite a particular page, agencies included. Repeat the log monthly and after any change to your site, hosting, robots.txt or firewall. For the wider picture of AI search work, see AI SEO, and if you'd like a second opinion on your own setup, you can request a strategy review.

1. Access. The robots.txt file has Disallow: / under User-agent: *, left over from when the site was built on a staging server. There's no PerplexityBot group, so the catch-all applies and PerplexityBot is shut out. The developer removes the blanket rule and checks the Cloudflare AI bot settings, where Search traffic is allowed.
2. Reading. View page source on the termite page shows the opening text, but the questions and answers lower down load through a JavaScript widget. The developer moves them into the HTML the server sends.
3. Verification. Four days later the access log shows PerplexityBot requests to the termite page with a 200 status. The IP address matches perplexitybot.json, and the owner notes the date and the file's creationTime.
4. Passage. The page gets a question heading about inspection frequency, a direct first sentence, the suburbs the business covers, the name of the licensed technician who checked the page and a visible updated date.
5. Observation. Once a month, signed out, in a new thread each time and from the same location, the owner asks the same six customer questions word for word and logs every numbered source, plus the Google and Bing results for each.
6. Reading the log. Say that by the third month the business is cited for two of the six questions. For the other four, a directory listing is cited, and it shows an old phone number. The owner corrects the listing, which matters whether or not Perplexity ever cites the business's own page.
7. Review. The owner keeps the log going, reruns the access checks after any hosting or firewall change, and judges progress on the trend across months.