Perplexity does not cite the same way ChatGPT does. The differences are not dramatic but they matter, and most of the AI search advice circulating in 2026 conflates the two engines into a single set of rules. The most consistent finding across our tests: Perplexity favors clear answer formatting, dated content, and named-source statistics more aggressively than ChatGPT does.
This article covers what Perplexity is actually scoring, how its retrieval differs from ChatGPT, and the four specific moves we have seen move Perplexity citation candidacy on real B2B pages.
ChatGPT browse mode uses Bing’s retrieval layer, then re-ranks the candidates by the three signals we covered in How ChatGPT actually decides what to cite: answer density, named-source density, entity clarity.
Perplexity runs a hybrid: it queries Google search, its own crawler, and select third-party APIs in parallel, then applies its own re-ranking. The candidate pool is larger and more varied than ChatGPT’s. Perplexity also displays citation footnotes more prominently in its output, which means the user often clicks through to verify, and Perplexity’s ranking signals seem partly tuned to favor pages where verification will not embarrass the engine.
The practical implication: Perplexity is more aggressive about three things ChatGPT cares about more weakly.
1. Clear answer formatting. Perplexity preferentially cites pages where the answer to the prompt is in a clearly delimited section. H2 questions, bulleted lists, numbered steps, and answer paragraphs that are visually separated from surrounding context all help. Walls of running prose are cited less.
2. Dated content. Perplexity weights freshness more heavily than ChatGPT in our tests. A page dated 2024 on the same topic as a page dated 2026 is cited noticeably less often, even when the 2024 page has stronger schema and better answer-first structure. Pages with dateModified updated within the last 90 days perform better.
3. Named-source statistics. Perplexity is more aggressive than ChatGPT about preferring pages that cite named sources for any non-obvious claim. The pattern: Perplexity uses your citation chain as a verifiability score. If your page cites Pew, Forrester, Ahrefs, or any named third-party for a statistic, Perplexity is more likely to lift the statistic and credit your page.
In tests across nine B2B pages over 30 days, four changes moved Perplexity citation candidacy more reliably than other changes.
Move one: refresh dates aggressively on pages you care about.
If a page’s dateModified is older than 90 days and the content is still accurate, update the date and ship one or two minor content improvements. This is not a hack. It is signaling to Perplexity (and to Google) that the page is being maintained. We saw 1.2 to 1.5 times more Perplexity citations within 14 days of date updates on pages where the content remained substantively the same.
The rule we use: every commercial page gets a dateModified refresh quarterly, with at least one substantive update (new section, new example, new statistic, or removed obsolete content). Pages that are simply re-dated without a substantive change do not perform as well in the second cycle. Perplexity’s freshness signal degrades quickly if the content has not actually changed.
Move two: restructure your H2s as questions.
Pages that use H2 headings phrased as direct questions (“How does X work?” “What is Y?”) are cited by Perplexity at higher rates than pages that use declarative H2s (“How X Works” or “About Y”). The mechanic: Perplexity scans H2s as semantic landmarks. Question-formatted H2s map cleanly to user prompts. Declarative H2s require interpretation.
This is a 30-minute refactor on most pages. Walk through your H2s. For each one, ask: “Could this be a question a buyer would type?” If yes, rewrite as the question.
Move three: add a named-source citation in the first 200 words.
Perplexity preferentially cites pages that themselves cite. Adding one named-source statistic in the first paragraph or two of the page meaningfully changes citation behavior. The source does not need to be a tier-one academic publisher. Industry reports (Forrester, Gartner, Ahrefs, SparkToro), peer-reviewed publications, government data, and named first-party studies all work.
The hand-waving version (“research suggests…”, “industry data shows…”) does not work. The link to the actual source matters. Even if the link is to a press release rather than a primary source, the explicit citation chain is what Perplexity is scoring.
Move four: add a footnote-style references section.
Pages that include a “References” or “Sources” section at the bottom, with named third-party sources linked, are cited by Perplexity at higher rates than pages that hide their sources inline. The mechanic: Perplexity uses the references section as a confidence boost, because it can verify the page’s claims by walking the citation chain.
This is also a 15-minute addition. Add a short “Sources” or “References” block at the end of any page that makes substantive claims. Link 3-7 named sources. Make sure each link is to a stable URL that will not 404.
A few changes we tested that did not produce reliable lift on Perplexity specifically.
More schema types stacked on the same page. Adding HowTo and DefinedTerm on top of FAQPage on the same page did not produce additional citation lift in our tests. One well-formed FAQPage schema was sufficient. The marginal benefit of stacking more types was below measurement noise.
More internal links. Adding 10-15 internal links to a page did not move Perplexity citation candidacy. Internal linking matters for traditional SEO but Perplexity’s re-ranking does not appear to weight it heavily for citation candidacy.
Longer content. Padding a 1,500-word page to 3,500 words by adding more sections did not improve citation candidacy. In two tests it made citation worse, presumably because the answer became harder to extract from the longer page.
Author Person schema alone. Adding rich Person schema for the author was a measurable plus for ChatGPT but did not produce a clean signal for Perplexity in our 30-day window. It may matter on longer timeframes. We did not see it in our test.
If you only ship one Perplexity-specific change this month: refresh the dateModified on your top 10 commercial pages, with one substantive update on each. The freshness signal is the cheapest, fastest way to move citation candidacy on this engine specifically. The other three moves (question-formatted H2s, in-paragraph source citations, references section) are higher-effort but compound over time.
This is a different rule than the ChatGPT-specific rule (answer-first writing in the first 60 words). The two engines respond to different signals. Pages optimized for both end up well-rounded.
Does Perplexity follow robots.txt?
Yes. PerplexityBot respects robots.txt. If you have not explicitly allowed PerplexityBot (or have a wildcard Allow), you are fine. Some sites accidentally block AI crawlers via plugin defaults. Audit your robots.txt.
Does Perplexity respect nofollow links?
The public documentation is light here. In our tests, Perplexity’s crawler reaches pages even when reachable only via nofollow links, but ranking signals from those pages appear weaker. Practical implication: do not rely on nofollow-only inbound paths for pages you want cited.
Will Perplexity cite my paid content?
Yes if the page is fully crawlable. Perplexity does not penalize paid content as a category. Some publishers paywall content that Perplexity then cannot read, which obviously prevents citation.
Can I see what Perplexity has cited me for?
Not directly via Perplexity’s own interface. Tools like Otterly, Profound, Peec, and AthenaHQ track this across multiple AI engines including Perplexity. Manual testing (typing your target queries into Perplexity and logging the results) is the alternative.
If your top commercial pages do not have recent dateModified values, that is the first Perplexity-specific move. If they do, the question-formatted H2 rewrite is the next.
If you want us to ship all four Perplexity moves on your top 10 pages this month, book the fit call. The audit ships in week one. Productized SEO and AI search at $497 a month, no contract.
Related reading:
– How ChatGPT actually decides what to cite (three signals tested)
– Answer-first writing: before-and-after on real B2B pages
– GEO 101: Generative Engine Optimization Explained
– NetPageTwo vs Otterly: tracking tool vs productized service
