How to Rank in Perplexity
You cannot rank in Perplexity, because Perplexity has no rankings. It has an index of your whole domain, a trust rating attached to that domain, and a per question shortlist of passages it pulls from that index. Getting cited means being in the index, being rated, and writing passages that answer a question on their own. Nobody can promise you a spot.
How to get cited by Perplexity: how its index and source labels work, how it differs from ChatGPT, and the passage-level writing that survives its shortlist.
On this page
There is no page one here
On Google you have a position. You are third, or eleventh, and you can watch that number move.
Perplexity does not work that way. It runs a search against its own index, pulls back candidate material, reranks it, writes an answer, then attaches a handful of links to the sentences in that answer. There is no list. There is no page two. You are either one of the handful or you are not.
That sounds worse than it is, and I will come back to why.
First, what makes Perplexity worth writing about. It is the only answer engine that has explained its retrieval in public. Perplexity's engineering team says its index covers over 200 billion unique URLs and processes tens of thousands of indexing operations per second. It trains a model to predict whether a URL even needs indexing, based on how important the page is and how often it changes. And it retrieves at both the document and the sub document level, saying the pipeline is built to surface the most atomic units possible.
That last line is the whole post in one sentence. Perplexity is not looking for your page. It is looking for the smallest useful chunk of your page.
Perplexity has already graded your domain
Here is the newest thing, and most people have not caught it.
On 6 August 2026 Perplexity published a help page explaining source labels. Some cited domains now carry a small shield with one of three labels: Government, Academic, or Trusted. Hover a source and you get the label plus a one line description of what that site is known for.
Two details matter. First, Perplexity rates the whole website, not the page, so this is not something you fix per article.
Second, what it checks. Perplexity lists the questions its review process asks, and they are not what you would guess. Does the site correct its mistakes? Does it say who wrote each piece? Does it keep news separate from advertising and opinion? That is editorial hygiene. Not backlinks, not word count, not schema.
Perplexity's developer documentation shows the same thing from the other side. Search results carry a trust object on every hit: a numeric level, a name like credible or trusted, and a plain English sentence describing what the domain is trusted for. One documented example reads that a site "is credible for personally authored software system design case studies and interview preparation content without formal institutional peer review."
Read that again. Perplexity is writing down, in a sentence, what it thinks you are an authority on and where that authority stops. Jon Clark flagged this on LinkedIn on 5 September and made the right point: if that description is wrong, or narrower than what you do, that is not a Perplexity bug. That is how your business reads to a machine. Most domains have no label, which Perplexity says is not a negative judgement.
Perplexity vs ChatGPT: they are not the same job
If you have been optimising for ChatGPT, do not assume any of it carries over.
Suganthan Mohanadasan ran an unusual test in July 2026. Rather than read Perplexity's answers, he hooked into the browser and captured the raw data stream while it worked. For six of his seven queries, Perplexity ran a single search using almost the exact words of the question. It did not expand the query or chase side questions. Only the local query triggered a second round.
That is close to the opposite of ChatGPT, which rewrites and fans out. We wrote up that side in how ChatGPT searches now.
The practical difference is blunt. If Perplexity searches your question nearly word for word, the exact phrasing your customer uses needs to appear on your page. Not a clever variation of it. The phrase.
The gap shows in the numbers too. A Haus Research study published on 2 September 2026 put 310 factual questions about 210 technology companies to Perplexity's two search models. Perplexity cited 9.8 sources per answer. A ChatGPT model given the same questions cited 2.0. Perplexity puts roughly five times as many doors in front of you.
Being read is not being cited
Now the hard part. Suganthan watched brand pages get retrieved and then never cited. On commercial queries, fresh listicles won instead. On product queries, YouTube got 38 citations and Reddit got zero, even though Reddit threads were pulled back. On his local query, every citation came from place data, not the editorial listicles sitting right there in the results.
So Perplexity can crawl you, index you, retrieve you for the right question, and still not quote you. Being in the room is not the same as being called on.
What survives that step is a passage that answers the question by itself. Not a page that builds to an answer over eight paragraphs. A chunk that still makes sense dropped into someone else's sentence.
The Haus study adds a warning. About 16 percent of the URLs Perplexity cited were behind a login, paywall, 403 or bot wall, and a further 2.5 percent only showed their content after JavaScript ran. If your best material sits behind a gate, you are gambling.
The Reddit fight nobody has settled
You will read everywhere that Reddit is the key to Perplexity. Three recent measurements disagree.
Epovest, which sells visibility measurement, archived 923 answers in late July 2026 and found Reddit second on Perplexity and 102nd on ChatGPT. Achtung, which also sells visibility monitoring, ran 948 Perplexity queries from mid May 2026 and saw Reddit cited in none of them, while YouTube turned up in 44.7 percent. Suganthan's capture matches Achtung: YouTube heavily cited, Reddit retrieved and ignored.
Nobody has reconciled these. One likely reason is question mix: most of Epovest's sample was travel questions, where people swap Reddit tips, while Suganthan's were product and local. That is a guess, not a finding.
What all three agree on is that YouTube does unusually well on Perplexity. If you have ever recorded anything useful, that is the cheapest experiment here.
What to actually do
Check your robots.txt first. Perplexity documents two crawlers. PerplexityBot builds the search index. Perplexity-User fetches a page live when someone asks a question, and Perplexity says it generally ignores robots.txt because a person asked for it. Block PerplexityBot and you are not in the index at all. Changes take up to a day to register.
Then, in order: put the customer's exact question as a heading and answer it in the first two sentences underneath. Say who wrote the page and when it was last updated, because Perplexity's index carries a last updated field and its labelling process asks who wrote each piece. Publish a corrections note if you have fixed something. Keep sponsored content visibly separate from editorial. Ungate anything you want quoted.
Then run your own questions in Perplexity and read the source panel. That is your scoreboard.
Why guarantees are impossible
Anyone selling a guaranteed Perplexity placement cannot deliver it. There is no slot to buy. The shortlist is rebuilt per question, from a reranked candidate set, by a model that writes a different answer each time. Perplexity's own docs say labels change as sites are reviewed again. Ask the same question twice and you can get different sources.
What you control is whether you are in the index, whether your domain reads as an accountable publisher, and whether your best paragraph stands alone. That is the job. If you want a second pair of eyes on where your site sits, our audit covers these checks.
Sources
- Perplexity, "Architecting and Evaluating an AI-First Search API", perplexity.ai/hub/blog, 25 September 2025
- Perplexity, "Introducing the Perplexity Search API", perplexity.ai/hub/blog, 25 September 2025
- Perplexity, "Search API: Better Extraction, Dynamic Benchmarks", perplexity.ai/hub/blog, 11 March 2026
- Perplexity Help Center, "Understanding source labels", 6 August 2026
- Perplexity developer documentation, CLI and Search API reference (trust object, date and last_updated fields, recency and domain filters), docs.perplexity.ai
- Perplexity developer documentation, "PerplexityBot and Perplexity-User crawlers", docs.perplexity.ai
- Suganthan Mohanadasan, "How Perplexity Actually Picks Sources (I Read the Stream, Not the Answers)", suganthan.com, 27 July 2026
- Daniel Rossi, Haus Research, "Perplexity Citation Audit", 2 September 2026. 310 questions, 210 companies, 2,511 citation markers. Dataset published under CC BY 4.0. No commercial affiliation disclosed
- Simon Vasconcelos Lee, Epovest, "What AI assistants actually cite", 26 July 2026, updated 8 September 2026. 923 answers, 6,991 citations, 10 to 25 July 2026. Epovest sells AI visibility measurement
- Achtung.app, "Every AI platform has its favourite sources", 6 June 2026. 3,044 queries across four providers from 14 May 2026. Achtung sells AI brand visibility monitoring
- Jon Clark, LinkedIn post on Perplexity source verification labels, 5 September 2026
Frequently asked
Does Perplexity use Google's rankings?
No. Perplexity runs its own crawler and its own index, which it says covers over 200 billion unique URLs, with multi stage reranking ending in cross encoder models. Ranking well on Google does not put you in a Perplexity answer, and practitioners have watched page one Google results go uncited.
How fresh does my content need to be?
Perplexity's index stores both a publication date and a last updated date for every result, and its API lets a developer filter down to the last hour. So freshness is tracked closely. That does not tell you how heavily it is weighted, and no independent study has measured it. Treat monthly refreshing as a bet, not a rule.
Should I add schema markup for Perplexity?
Perplexity's published labelling criteria do not mention schema at all. They ask about corrections, bylines and separating ads from editorial. Schema is worth having for ordinary search, but there is no public evidence it earns a Perplexity citation.
Can I see whether Perplexity trusts my domain?
Partly. Ask Perplexity a question your site should answer, then hover the citation. A shield means you have a label and a description of what you are trusted for. No shield means unrated, which Perplexity says is not a mark against you.
Written by
Lourdes Paul Agilan runs GeoCited, a generative engine optimization agency. He works with B2B companies on how AI assistants describe and recommend them. He writes about that work here, and shares the experiments behind it. The tests that worked, and the tests that did not, written up the same way.