How to Rank in Perplexity When No One Publishes the Rules

How to rank in Perplexity comes down to getting on the short list of sources it cites for a prompt. Only a few ways to get there have evidence behind them: let its crawler in, rank well in Google, and get mentioned on other sites.

Which of the tips you've read came from Perplexity itself? Which came from a study with a method? Which came from another guide with no source?

We sorted them. We also checked our own data against what we found.

Key takeaways:

• Ranking in Perplexity means being one of a handful of cited sources, not holding the top spot
• Perplexity documents its crawler and some of its steps, but it doesn't publish any weights
• Ranking in Google and getting mentioned on other sites have measured support, while schema and llms.txt have no Perplexity data behind them
• Track a fixed set of prompts every month, and expect small referral traffic

What Does Ranking in Perplexity Mean?

Ranking in Perplexity means getting your page cited as one of the sources behind its written answer. Perplexity writes one answer to a prompt and lists the pages it used.

We tracked a set of prompts through Ahrefs Brand Radar's monthly Perplexity run on 26 to 29 September 2026. Each answer cited a short list of pages, and the lists barely overlapped. Most sites appeared in only one answer, and even the most common source appeared in only a minority of them.

That spread means no site owns the list. Each prompt draws its own set of sources, so you earn a place prompt by prompt.

Other counts point the same way. The SE Ranking study counted about five cited links per Perplexity answer in early 2025. The methods differ, but both say each answer cites only a handful of pages.

So the goal is a spot on a short list of sources.

How Does Perplexity Choose Its Sources?

Perplexity doesn't publish how it chooses sources, but it does document some of the steps. Here's what it says itself.

Its Search API write-up says it searches its own index by exact words and by meaning, then merges the results. It then drops stale and clearly off-topic pages, ranks what's left in stages, and scores whole pages and parts of pages.

That write-up covers the Search API, not the app you chat with, so treat it as a close guide and not a rulebook.

Here's what someone has measured. Four AI assistants answered 15,000 long-tail questions, and an Ahrefs study from August 2025 compared the pages they cited with Google's results. For Perplexity, 28.6% of the cited pages were in Google's top 10 for the same question, against about 8% for the other assistants.

Short head terms behave differently: 65% of Perplexity's results were in Google's top 10 in a later Ahrefs test of 3,311 head terms. The overlap swings with the type of query, and it shows a link, not a cause.

Perplexity runs its own crawler, PerplexityBot, so it doesn't simply copy Google. That same Ahrefs test also reported that Perplexity had been scraping Google results through a third-party provider, so a Google ranking and a Perplexity citation are linked but not the same thing. Engines differ in how they pick sources, and our ChatGPT versus Perplexity comparison shows how for B2B brands.

Here's what nobody has published. Perplexity hasn't shared its weights, so a ranking-factor chart with decimals is fan fiction.

What Did Our Own Perplexity Data Show?

Our own tracking shows a wide spread of sources and a small stream of traffic. In the answers we track, the most common source was LinkedIn, followed by a few agency and software sites.

Outside research points to where Perplexity looks for sources. A 2025 University of Toronto preprint tested Perplexity's API model with August 2025 data.

For well-known consumer brands such as Apple and Nike, about two thirds of the sources were independent third-party sites, about a quarter were social sites, and fewer than one in ten were the brand's own site. In car categories, brand-owned pages made up 29% to 34% of sources.

If that pattern holds for B2B, a brand's own pages may not be enough, and mentions on other sites may be the missing piece. That's our inference, not a proven result for B2B.

Our traffic shows how small the channel is today. Over the 90 days to 11 October 2026, Perplexity sent a fraction of one percent of our sessions. Google sent about a fifth, and direct visits made up more than 70%.

Treat this as a small observation. It covers one set of prompts, one monthly run in late September and one B2B site. Direct visits are a big share of our sessions, so the referral numbers may be low.

Even a small baseline helps, because it makes any change easy to spot.

Which Tactics Have Evidence Behind Them?

Seven tactics come up most, and the evidence behind them runs from firm to thin. Most of it dates from 2025, so Perplexity may have changed since:

None of these guarantees a citation. Our read is that, for most B2B sites, ranking in Perplexity is ordinary SEO plus bot access and third-party signals.

Which Advice Should You Skip?

Skip any advice that comes with a precise number and no study. A precise number with no study is a horoscope with a footnote.

Four claims come up again and again, and none holds up as stated:

• Schema markup: no clear lift in an Ahrefs test on three other engines, and no Perplexity test that we found
• An llms.txt file: we found no Perplexity statement or study showing it helps
• Answer in the first 100 words: we couldn't find the study behind the "90%" figure in guides
• Reddit as nearly half of citations: Profound's 46.7% counts only citations to Perplexity's ten most-cited sites, and across all citations Reddit's share is 6.6%

Unproven claims may turn out true. Until someone shows the data, they don't deserve a week of your time.

How to Rank in Perplexity in Four Steps

Four low-cost steps follow from the evidence above. Some have measured support, and some are our inference.

1. Check bot access. Open your robots.txt and your firewall settings, and make sure PerplexityBot and its published IP ranges can reach your site. A firewall that blocks PerplexityBot hides you for free. Changes can take up to 24 hours to apply, a delay noted in Perplexity's crawler documentation. A separate fetcher, Perplexity-User, generally ignores robots.txt because a person asked for the page.
2. Find the exact prompt, then check Google. Write the question a buyer would ask and search it in Google. Perplexity cites some of the pages that rank there, so fix your own rank first if you're missing. If the top results are other companies' roundups, that's where to earn a mention.
3. Make each section stand alone. Start every section with a full-sentence answer. A section that reads clearly on its own is likely easier to score and to quote, though that's our inference, not a proven lever. Our five content signals cover what makes AI tools trust and cite a page.
4. Track a fixed set of prompts. Pick 20 to 30 prompts, run them once a month and note which sites Perplexity cites. Read the trend, because one run is only a snapshot, and expect small referral traffic.

Mentions on other sites are the slower job. Start them once the four steps are running.

How we do this at Tenpoint Labs

We check bot access and Google rank for the exact prompt before we touch any formatting. Then we track a fixed set of prompts each month, so we can tell a real change from a lucky run.

The data behind it

We pull prompt results from Ahrefs Brand Radar and sessions from GA4, then compare the two. That's how we see which sources Perplexity cites and how much referral traffic it sends.

What you see in your seat

You see the same views we plan from: AI search visibility across five engines, and the acquisition explorer for referral traffic. We connect it for you. See what's in a seat on Tenpoint Analytics.

FAQs

The short list rewards the basics: a crawlable site, pages that rank in Google, and mentions on other sites. This week, check bot access, search your three key buyer prompts in Google, and add a "cited in Perplexity: yes or no" column next to Google position in your keyword sheet. Tenpoint Labs is one team that tracks both for clients in Tenpoint Analytics, and you can find out more if you'd like to see how.