Perplexity ranking factors: what Perplexity has said, what studies guess, and what nobody knows

Perplexity has not published a list of ranking factors.

I checked on 13 October 2026, three ways. The Perplexity Crawlers page says how PerplexityBot behaves and nothing about ranking. The help center index lists every article, and no English title or summary mentions ranking, SEO or an algorithm. The Publishers' Program post talks about money, not order.

So the first honest answer to "perplexity ranking factors" is short. Nobody outside Perplexity knows them.

That is a problem for this post, and for every post ranking for this phrase. Most of them print a confident list anyway.

This one splits the subject into three piles: what Perplexity has said, what only third parties claim, and what is still unknown. Then the few steps the first pile supports.

Step 1: What Perplexity says it does with a question

Start with Perplexity's own description.

Its help article How does Perplexity work?, last modified 3 September 2026 and read on 13 October 2026, gives three stages: understand the question, search the web, summarise the most relevant findings.

It says the search gathers information from "authoritative sources like articles, websites, and journals." It says each answer carries numbered citations to the original sources.

A second article, What is an answer engine, read the same day, adds one phrase: Perplexity works by "identifying trusted sources."

That is the whole public model. Search, pick, summarise, cite.

Notice what is missing. No word on how "authoritative" or "trusted" is scored, how many results are read, or how freshness, links or page structure are weighed.

Step 2: How a page gets into the pool at all

This part is documented, and it is the only part you fully control.

The crawler docs, read on 13 October 2026, say PerplexityBot is "designed to surface and link websites in search results on Perplexity." They say: "To ensure your site appears in search results, we recommend allowing PerplexityBot." Changes can take up to 24 hours to reflect, by the same page.

The help article How does Perplexity follow robots.txt?, read the same day, adds a detail most guides skip. If a page is blocked, Perplexity "may still index the domain, headline, and a brief factual summary."

It also says Perplexity "partners with third-party crawlers to help build our search index." The article does not name them.

So your pages can reach Perplexity's index through a crawler you never see in your logs. That is the first complication. Blocking PerplexityBot does not make you invisible, and allowing it is not the only door in.

The difference between PerplexityBot and Perplexity-User, and Cloudflare's 2025 stealth-crawling report, are covered in our PerplexityBot and Perplexity-User breakdown. I will not repeat them here.

Step 3: The one quality signal Perplexity has described

There is one.

Perplexity's help article Understanding source labels, read on 13 October 2026, says some cited domains carry a shield icon with one of three labels: Government, Academic or Trusted.

It rates whole websites, not single pages. Its review process asks questions such as whether the site corrects its mistakes, says who wrote each piece, and keeps news apart from advertising and opinion.

The Trusted label goes to sites "that appear often enough in results to receive a label" and publish within their own area of expertise.

Two lines in that article matter most for a marketer. Partnerships, payments and business arrangements "do not affect a site's label." And having no label "is not a negative judgment."

Now the concession. The article never says a label changes how often a site is cited or where it sits in an answer. It describes a badge shown to readers.

So a label is evidence of what Perplexity values. It is not evidence of a ranking boost. Anyone telling you to chase the Trusted badge for rankings is guessing.

The review questions are still worth answering on your own site. Bylines, a corrections line and clear separation of ads cost almost nothing.

Step 4: The money question

People ask whether Perplexity sells placement.

The Publishers' Program post, dated 30 July 2024 and read on 13 October 2026, describes revenue sharing: when Perplexity earns from an interaction "where a publisher's content is referenced," that publisher earns a share.

That is payment after a citation happens. The post does not say partners are cited more.

It does say something else, in passing: "Recently, we updated how our systems index and cite sources." No detail followed. It is a reminder that whatever Perplexity does, it changes without notice.

Step 5: What third-party studies add, and where they stop

Here is my contrarian line. Most Perplexity ranking lists are fiction.

Not because the authors lie. Because the method cannot produce a ranking factor. A third-party study runs a set of prompts, records which domains Perplexity cites, and infers why. That gives you a pattern, not a cause.

I did not open a Perplexity-specific citation study for this post, so you will not find their numbers here. Our rule is that a number appears only next to a source we read.

What NovaTechRay has read elsewhere points one way. Across engines, citation follows ordinary search ranking and brand mentions far more than any on-page trick, which is the argument in how much of an AI citation is just your Google ranking.

Platform matters too. VALORAE Media's post on which platform gets you into AI answers lays out a third-party count of how rarely Instagram is cited next to YouTube and Reddit, which is useful if your proof lives in video.

And any study's result moves with its prompt list. Our post on why AI visibility tools disagree covers why one run is noise.

Before trusting a Perplexity study, ask for three things: a date, a prompt count, and a method you could repeat. Without them, skip it.

Step 6: What is still unknown

Plenty. As of 13 October 2026, Perplexity's public pages do not say:

• which search index or partner crawlers supply results, by name
• how many sources are read before an answer is written
• whether freshness, links, schema or page length change selection
• whether a source label changes citation frequency
• how much the chosen model changes which sources get cited

That last one is real. The how it works article says Pro users can pick models such as GPT-5 and Claude 4.6 Sonnet. Whether the model changes source choice, or only the summary, is not stated.

We do not know. Neither does anyone selling a checklist.

Step 7: The steps the evidence supports

Short list, on purpose.

1. Allow PerplexityBot. Perplexity's own docs recommend it if you want to appear in results.
2. Check your CDN, not only robots.txt. A bot rule at the edge can block Perplexity before the file is read. Our Cloudflare decision tree covers the switches.
3. Put the answer in the HTML. Perplexity has not said whether its crawler runs JavaScript. Do not bet on it. Our JavaScript menu test takes two minutes.
4. Answer the source-label questions. Bylines, a corrections note, ads kept apart from content. Perplexity says these are what its reviewers check.
5. Rank for the question. Perplexity says it searches the web. Being in that search is ordinary SEO.
6. Measure with repeated prompts. Run your customer questions in Perplexity on a schedule, as in our AI visibility tracking guide.

None of these is a ranking factor. They are the parts of the pipeline Perplexity has described in public, plus the plumbing that lets it read you.

The short version

• Accept that Perplexity publishes no ranking-factor list.
• Allow PerplexityBot and check your CDN rules.
• Serve the answer text in the HTML, not behind a script.
• Add bylines, corrections and clear ad separation.
• Treat every third-party Perplexity ranking list as a pattern, not a cause.
• Ask any study for its date, prompt count and method.

NovaTechRay does this audit at novatechray.com, and we will update this post when Perplexity says more.