How to rank in Perplexity: what it cites, and how to become one of its sources

The short answer

Your own pages can be Perplexity’s source: in TIN’s measurement, its largest source category was company sites, at 38.9% of its 1,099 citations, and it cited 634 different sites for 60 everyday questions, more than any of the seven engines measured source. Let PerplexityBot in, put every fact in the HTML, answer the exact question with a current date, earn links and mentions on sites that already rank, and skip schema and llms.txt as citation levers.

Contents

1. What Perplexity cites today
2. How Perplexity finds and picks a page
3. What to do, in order of the evidence
4. What not to pay for
5. How to check whether Perplexity cites you
6. What this guide cannot tell you

What Perplexity cites today

TIN asked ChatGPT, Perplexity, Gemini, Copilot, Claude, Google AI Overviews and Google AI Mode the same 60 everyday questions in 10 countries, recorded every site their answers cited, and coded all 1,395 cited sites into 10 categories source. The engines were run through Ahrefs Brand Radar as custom prompts and read on 2026-10-02 source. The study was pre-registered, and its later changes are published as dated amendments source.

Perplexity returned 1,099 citations, the most of the seven engines source. Its most-cited sites were forbes.com (13), wikihow.com (13), shopify.com (11), britannica.com (10) and nerdwallet.com (10) source.

Source for every row: source.

The shape is the point. One category, company sites, takes the largest share, and the other nine sit between 2.5% and 9.8% source. The study’s own reading: “For Gemini and Perplexity, your own site can be the source. Company sites are their largest category (58% and 39%), so a company’s own pages are the kind of source they quote most” source.

Perplexity also spreads its citations wider than any other engine. It cited 634 different sites, and its 10 most-cited sites hold 8.6% of its citations, the lowest top 10 share of the seven source.

Source for every row: source.

A wide spread means more room for a site that is not already a household name. That is TIN’s inference, not a measured finding: the study counts where citations went, not who could have won one.

How it differs from ChatGPT and Google’s AI

Perplexity is close to the opposite of ChatGPT. ChatGPT’s largest category is government, at 50.5% of its 299 citations, and it cited 174 different sites source.

Source for every row: source.

The engines barely share sources. Of Perplexity’s 25 most-cited sites, 1 is also in ChatGPT’s top 25, 4 in Gemini’s, 4 in Copilot’s, 2 in Claude’s, 4 in AI Overviews’ and 7 in AI Mode’s source. Across all seven engines, “1,091 of 1,395 cited sites (78%) appear in only one engine’s answers” source.

Video is where Perplexity and Google part most clearly. youtube.com is the most-cited site in both AI Overviews (17) and AI Mode (28) source. It has 0 Perplexity citations in TIN’s data source.

The practical rule is the study’s own: “Do not optimise for ‘AI’ as one channel” source.

The Reddit question

On TIN’s everyday questions, social and video platforms, the category that includes Reddit, take 2.5% of Perplexity’s 1,099 citations source.

The conclusion is narrow: on these questions, Reddit was not how Perplexity’s answers were built. That does not settle Reddit’s weight on other questions; in a startup preprint, community presence predicted discovery once the Reddit data was cleaned source.

How Perplexity finds and picks a page

A citation is the last of several stages. Perplexity’s crawler has to reach the page, read its content, and then a passage from it has to beat other passages. TIN’s analysis of passage retrieval describes the last step: “An answer engine fetches your page, splits it into blocks, scores each block on its own, and shows the model a handful of them” source.

Perplexity does not publish how it ranks sources. Its documentation says only that PerplexityBot is “designed to surface and link websites in search results on Perplexity” source.

Let PerplexityBot in

Perplexity runs two fetchers, and they behave differently.

Three things to check, in this order:

1. robots.txt. Make sure no rule blocks PerplexityBot, including a broad Disallow aimed at every bot. Per Perplexity’s documentation, the crawler that surfaces and links sites in its search results is PerplexityBot source.
2. Firewall and CDN. A web application firewall can block a bot that robots.txt allows. Perplexity’s documentation gives allow-list guidance for WAFs and publishes its IP ranges source.
3. Wait before you test. For its robots.txt tags, Perplexity says “it may take up to 24 hours for our systems to reflect changes”; for firewall changes, only that they “may take some time to propagate” source.

Blocking in robots.txt does not stop every Perplexity fetch, because Perplexity-User generally ignores it source. If a page must stay out of Perplexity entirely, robots.txt alone will not do it.

Put every fact in the HTML

PerplexityBot does not run your JavaScript. The crawler-log study by Vercel and MERJ found: “None of the major AI crawlers currently render JavaScript,” and it names Perplexity’s crawler, PerplexityBot, in that list source.

The fix is mechanical. Every price, date, figure and answer the page wants cited goes in the server-rendered HTML. Open the page with JavaScript switched off; what you see is what the crawler gets. Content loaded by a script, a tab that fetches its text on click, or a figure drawn only inside a chart image is not there for it.

What to do, in order of the evidence

Make your own page the best answer to the exact question

This is the step the Perplexity data supports most directly: company sites are its largest source category source.

What makes one page beat another comes from a different test. In a controlled experiment of 252,000 trials across six language models, “topical relevance and list position are the biggest drivers of being cited first”, and “including explicit price information and a recent timestamp also helps consistently” source. Perplexity is not one of the six models tested source. Read it as evidence about how language models pick between sources, not as a Perplexity result.

In practice: one page per question your buyers actually ask, the answer in its opening lines, a visible and current date, and a stated price if the question is commercial.

Earn links and mentions on sites that already rank

The one study that looks at what made Perplexity surface a product found that the answer was outside the page. A January 2026 preprint tested 112 startups from Product Hunt’s top 500 across 2,240 queries; each startup’s site was scored on GEO practices, and the score “showed no correlation with actual discovery rates” source. For Perplexity, referring domains and Product Hunt ranking predicted visibility, and “after cleaning the Reddit data for false positives, community presence also emerged as significant” source.

It is one author and one preprint source. It still points the same way as the SEO basics: links from sites that are already trusted in your field are the off-page work that the evidence supports.

Be present on the sites Perplexity already leans on

After company sites, Perplexity’s citations go to single-topic sites (9.8%), news (9.6%), government (8.3%), reference (7.5%) and comparison sites (7.5%) source.

For a company, that means being accurately described on the reference and comparison sites that cover its category, and being covered by the news sites its buyers read. The data shows where Perplexity’s citations go; it does not test any tactic for getting onto those sites, so this guide names none.

Write passages that stand alone

The retrieval mechanism above works on blocks, not whole pages; TIN’s analysis describes answer engines in general, not Perplexity in particular source. The working rule from TIN’s retrieval analysis: “One complete claim per block, 150 to 300 words, the range inside the one bundle that was tested under control, and not a measured optimum” source.

Put the entity name and the figure, with its date, inside the block that makes the claim. A block that says “as shown above” means nothing once it is lifted out of the page.

Localise outside the US

Perplexity’s answers are far more local in some countries than in others. The local share is the share of its citations going to sites that mainly serve the country the question was asked from source.

Source for every row: source.

In Japan, Sweden and Italy most of Perplexity’s citations go to local sites. In the UK, Australia and India, an English-language question is mostly answered from sites serving other countries source. The study’s advice: “Outside English, translate and localise” source.

What not to pay for

The evidence does not show schema or llms.txt raising citations.

The full evidence table behind these rows is in TIN’s review of generative engine optimization.

None of these rows proves a tactic does nothing on Perplexity.

How to check whether Perplexity cites you

Ask the question your buyer would ask, not your company’s name. In the preprint above, Perplexity recognised a product asked about by name 94.3% of the time, while the same products appeared 8.29% of the time for the question a buyer would ask without the name source. Asking by name measures recognition, which is not the number that brings a buyer.

Run each prompt more than once. A 2026 survey of 45 studies reports that Schulte et al. (2026), across four engines and 45 days, observed daily source-level Jaccard scores of approximately 0.34 to 0.42 and proposed seven to eight repetitions per prompt as a starting point; the survey adds that this “derives from a small universe of Swiss queries” source. Record the exact prompt, the date and the country for every run, because Perplexity’s local share differs by country in TIN’s data source.

What this guide cannot tell you

• Everyday questions only. The Perplexity figures come from 60 everyday questions; “Business-to-business questions are not covered” source. TIN’s agency study covers a B2B population, but it counts Perplexity only inside all platforms and prints no Perplexity figure, so the B2B picture for Perplexity is unknown.
• One snapshot. Edition 1 is one weekly run, read 2026-10-02 source. Perplexity’s mix may look different next month.
• Associations, not causes. The study says of its own findings: “These are associations in one snapshot, not causes” source. A large company-site share does not prove that improving your own pages wins a citation.
• Perplexity’s ranking is not public. Everything here is measured from outside.
• TIN is not yet a proof case. TIN’s own AI visibility baseline is 0% source. This guide does not claim Perplexity cites TIN.

Read next

• Which websites AI engines cite, TIN’s study: the full source mix for all seven engines, with the dataset behind every Perplexity figure here.
• Generative engine optimization: the controlled evidence: the crawler, schema, llms.txt and formatting evidence behind “what not to pay for”.
• How ChatGPT ranks passages: how passages get fetched, scored and dropped.
• Which agencies AI engines cite: the B2B population the everyday sample does not cover.

Get cited. If you want Perplexity’s sources measured for your own market and your buyers’ questions, tell us what you sell.

Sources

1. TIN, “AI citation sources” study, edition 2026-10-02. https://theinternetninja.com/recon/ai-citation-sources-2026/
2. TIN, “Generative engine optimization: what the controlled evidence shows”. https://theinternetninja.com/blog/generative-engine-optimization-what-the-controlled-evidence-shows-2026/
3. TIN, “ChatGPT ranks passages, not pages”. https://theinternetninja.com/blog/ai-search-optimization-what-the-retrieval-record-shows-2026/
4. TIN, “Agency AI citations” study. https://theinternetninja.com/recon/agency-ai-citations-2026/
5. TIN, “ChatGPT stopped citing Reddit”. https://theinternetninja.com/blog/chatgpt-stopped-citing-reddit-substitute-platforms-2026/
6. Perplexity, “Perplexity crawlers”, read 2026-10-16. https://docs.perplexity.ai/guides/bots
7. Vercel with MERJ, “The rise of the AI crawler”. https://vercel.com/blog/the-rise-of-the-ai-crawler
8. Vishwakarma, Kumar and Jamidar, SIGIR 2026, controlled experiment on citation factors. https://arxiv.org/abs/2605.25517
9. Amit Prakash Sharma, “The Discovery Gap: How Product Hunt Startups Vanish in LLM Organic Discovery Queries”, preprint, January 2026. https://arxiv.org/abs/2601.00912
10. Ahrefs, schema and AI citations. https://ahrefs.com/blog/schema-ai-citations/
11. Ahrefs, “We Analyzed 137K Sites: 97% of llms.txt Files Never Get Read”, 15 June 2026. https://ahrefs.com/blog/llmstxt-study/
12. Survey of 45 studies on generative engine optimization, 2026. https://arxiv.org/html/2607.14035