Perplexity citation guide — what gets cited and what doesn't
Perplexity is the most citation-heavy AI search engine. Here is a tactical guide to becoming a numbered source on Perplexity, based on six months of measurement data.
How Perplexity actually works
Perplexity's pipeline is, simplified: query → query rewriting → web search → top-N candidate URLs → fetch and parse → relevance and reliability ranking → answer synthesis with inline citations. Each cited number [1], [2] is a real, clickable source. Perplexity Pro adds a richer search backend and reasoning step, but the citation logic is the same.
What gets cited — concrete patterns
From observation across thousands of prompts, Perplexity disproportionately cites pages that:
- Answer the prompt directly in the first 200 words. Above-the-fold answer extraction. If the answer is buried after 6 paragraphs of intro, the page loses to a competitor that leads with the answer.
- Use H2s framed as questions or claims. „What is X?", „Why does Y matter?", „X vs. Y: 5 differences". Question-framed H2s are gold.
- Include numerical data with dates. „As of 2026, X has 42% market share" beats „X has significant market share" every time.
- Have FAQ schema. Perplexity routinely uses FAQ-marked Q&A pairs as direct answer sources.
- Are recently dateModified. Two pages with similar relevance — Perplexity will cite the fresher one.
What does not get cited
- Pages behind login or paywall.
- Pages that require JS to render the main content (Perplexity's fetcher is HTML-first).
- Pages with thin content under 300 words.
- Pages with stuffed keywords or AI-generated walls of text — Perplexity's relevance ranking penalises low-density signal.
- Pages that block PerplexityBot in robots.txt. (Trivial, but it happens.)
Practical 7-step Perplexity playbook
- Identify your top 30 buyer prompts. Not keywords — full sentences. Run them on Perplexity. Record which competitors are cited.
- For each prompt where you're not cited, identify the cited URL. Read it. What did it do that yours doesn't? Above-the-fold answer? Numbers? FAQ?
- Allow PerplexityBot in robots.txt. Verify with
curl -A PerplexityBot https://yourdomain.com/some-pagethat you get 200 + parseable HTML. - Rewrite the relevant pages with answer-first structure. First 150 words: direct answer. Next sections: substantiation. Add 4–6 FAQs at the bottom with FAQ schema.
- Update the `dateModified` once you publish. Don't fake it — but make sure the meta and visible date reflect today.
- Measure weekly. Re-run the prompt set. Citation gains typically show within 2–4 weeks of a substantive content update.
- Iterate. Add more prompts to the set as you find them in customer calls. Perplexity rewards consistency over time.
An advanced trick: schema-stacked answer pages
For your most strategic landing pages, stack three schemas: Article, FAQPage, and HowTo (where applicable). Perplexity will sometimes cite the same page across multiple sub-questions in a single answer if the page covers them with structured clarity. We've seen pages cited 3–4 times in a single Perplexity response — that's massive surface area.
Tracking your Perplexity citations over time
Manual tracking does not scale past 10 prompts. AI-search visibility tools (we make one) replay your prompt set against Perplexity daily, capture the cited URLs, and produce a longitudinal citation count plus delta alerts. The investment pays back the moment you spot a citation drop and can intervene before it compounds.