Getting cited by Perplexity: a working playbook

Perplexity is built around citation. Where ChatGPT can answer without sources for many queries, Perplexity almost always returns inline numbered citations users can click. That makes Perplexity the highest-leverage AI surface for traffic referrals: the citation is not just attributional, it is a clickable link.

The product also rewards different signals than ChatGPT. Freshness matters more. Explicit publication dates matter more. Citation-friendly markup (clear sources, structured pages, visible authorship) matters more. The reverse is also true: brand authority on the open web matters somewhat less, because Perplexity's retrieval layer is younger and weights query-time signals more than learned entity authority.

Open robots.txt and confirm PerplexityBot is allowed.

```

User-agent: PerplexityBot

Allow: /

User-agent: Perplexity-User

Allow: /

```

PerplexityBot is the indexing crawler. Perplexity-User is the user-agent for direct browse actions when a Perplexity user explicitly requests a page. Allow both. Some sites have started disallowing PerplexityBot in response to a 2024 dispute about content sourcing; if you do this, you remove yourself from Perplexity citations entirely. Make the call deliberately.

Verify the bot is fetching pages by checking your server logs. Perplexity's crawler is active and recrawls high-value pages frequently.

Perplexity weights recency. The platform's product preference is to surface recent, accurate sources. Two specific things help here.

First, render publication and last-updated dates in plain HTML on every article. Not in JavaScript, not hidden in metadata only. The model wants to see 'Published 2026-04-12' in the rendered page. Add it to the article header.

Second, ship Article schema with datePublished and dateModified properly populated. When you update a page, update dateModified. Perplexity's retrieval layer reads both the visible date and the schema date.

Third, actually update high-value pages on a quarterly cadence at minimum. Stale pages drift down in Perplexity's source weighting even if the content is correct, because the platform cannot distinguish 'still accurate' from 'forgotten.' Refresh them.

Perplexity stitches answers from multiple sources, often citing five to ten in a single response. The unit of citation is smaller than ChatGPT's, sometimes as short as a single sentence with a numbered footnote.

Optimize for this. Every claim that could stand on its own as a citation should be a clean, factual sentence with the entity named. Avoid burying numbers, dates, or definitions in long paragraphs. Surface them in their own short sentences or in well-structured lists.

Lists work especially well in Perplexity. The model often lifts list items directly into its answer. A bulleted list of 'five things to know about X' with each item a complete factual statement gives Perplexity five candidate citations from one section.

Perplexity is the engine where llms.txt has the most observable effect. The platform's source descriptions for cited domains often draw from llms.txt content. We have seen client domains with no llms.txt get described in Perplexity answers using copy that the model improvised, sometimes inaccurately. After shipping a clean llms.txt, the descriptions match the file.

Ship the FPWS llms.txt template at /llms.txt. Pair it with a schema graph: Organization with stable @id and sameAs, Person for every author, Article on every long-form piece. Perplexity's source panel often surfaces the author, which means the Person schema and visible byline both matter.

Perplexity favors structured documentation pages. Tables, definition lists, well-organized FAQs, and explicit comparison content cite well. This is where Perplexity diverges most from ChatGPT, which prefers prose-style passages.

If you have product documentation, API references, comparison pages, or pricing pages, prioritize these for AEO work targeting Perplexity. The structural clarity these pages already have lines up with what Perplexity rewards. Adding clean H2s, FAQPage schema, and visible dates often produces citations within weeks.

Editorial blog content also works on Perplexity, but the tone should be drier and more documentary than promotional. Perplexity's source weighting is biased against marketing language.

Perplexity citation tracking is more accessible than ChatGPT's because the citations are explicit in the UI. Run a monthly panel of 50 to 200 target queries, log the cited URLs, score position within the source list (Perplexity surfaces sources in a visible order), and track sentiment.

DataForSEO has a Perplexity scraper. Manual checks are also straightforward because the citations are visible. We log both. We also track Perplexity-referred traffic in analytics: Perplexity sends actual sessions because users click the inline citations, which makes attribution less ambiguous than ChatGPT.

The leading indicator we watch monthly is citation-share for the queries that drive client revenue. Citation-share is your domain's appearances divided by total citation slots across the panel. A healthy AEO program moves this number from 0 to 5 to 15 percent over six months for the queries that matter.

Three practical differences worth internalizing. First, Perplexity cites more, so there is more citation supply per query. A page that is structurally good often gets a citation slot in Perplexity before it earns one in ChatGPT.

Second, Perplexity sends real traffic. Users click. Citation slots in Perplexity are leads in a way that citation slots in ChatGPT often are not. UTM your high-value cited URLs and watch the sessions.

Third, Perplexity rewards recency more aggressively. A page that was great in 2024 and has not been touched will lose ground to a page published last month, even if the older page is more authoritative. The fix is the quarterly refresh cadence, not abandoning the older page.

If a client wants Perplexity citations specifically (common for technical SaaS, B2B services, and editorial brands), the first 30 days look like this. Week one: robots.txt allow-list, llms.txt, schema spine, visible dates on all articles. Week two: rewrite the top ten pages to clean H2 plus passage structure with visible dates and bulleted facts. Weeks three and four: launch the citation monitoring panel, baseline current state, ship two new pages targeting high-value uncovered queries.

Citations typically begin appearing in Perplexity within four to six weeks of these changes. Compounding to meaningful citation-share takes three to four months of consistent publishing and monitoring.