How to get cited by ChatGPT and Perplexity
Key takeaways
That is the short version. The longer version matters because ChatGPT and Perplexity do not work the same way, and the playbook that wins one barely moves the other. In Profound's analysis of citations from August 2024 to June 2025, Wikipedia was ChatGPT's most cited source (7.8% of all its citations) while Reddit led Perplexity (6.6%) (Profound). If you optimize for one engine and measure on one engine, you miss most of the picture. This guide treats them as two separate channels, then shows where the work overlaps.
How ChatGPT and Perplexity actually pick sources
Perplexity is a retrieval-first engine. For almost every query it runs a live search, pulls a handful of pages, and synthesizes an answer with inline citations. Publish a well-structured article today and Perplexity can cite it within a day or two. That makes it the fastest, most controllable AI citation channel and the right place to start any program.
ChatGPT is different. It leans heavily on parametric knowledge baked into training data, and only reaches for live web results (via its browsing layer and Bing index) when a query clearly needs fresh data. That is why the citation-chasing tactics that move Perplexity tend to produce less movement on ChatGPT. To get named by ChatGPT you usually have to already be a known entity in its training corpus, which is a slower, mention-driven game.
Their source preferences diverge sharply too. In Profound's data, Wikipedia takes 47.9% of the citations that go to ChatGPT's ten most cited sources, while Reddit is Perplexity's top source (Profound). Semrush's 13-week study of 230,000 prompts on ChatGPT search, Google AI Mode and Perplexity found Reddit and Wikipedia still among the most cited domains across the three in October 2025 (Semrush). The implication is uncomfortable for brands: your own blog is rarely the cited source. The cited source is where people talk about you.
Watch out
The citation graph: mentions beat backlinks
The single biggest mental shift for a link builder is this: in AI search, unlinked brand mentions matter more than links. In Ahrefs' study of 75,000 brands, branded web mentions correlated with visibility in AI Overviews at 0.664, against 0.218 for the number of backlinks (Ahrefs). Muck Rack's analysis of more than 25 million links cited by ChatGPT, Claude and Gemini found 84% of AI citations coming from earned media, and just 0.3% from paid or advertorial content (May 2026).
This is the world of co-citations and mentions. When your brand name appears next to a topic across many independent sources, the model learns an association and grows confident enough to surface you. So the work is not "acquire one DR80 link." It is "appear, consistently and credibly, in the places these engines already trust."
That is exactly the muscle a strong link-building program already has, repointed. If you have run digital PR campaigns or earned editorial placements, you have the raw materials. The difference is you now optimize for breadth of credible mention and entity consistency, not just referring-domain count. For a deeper framework on engineering this, see our guide to generative engine optimization.
| Signal | Weight for AI citations | Where it lives |
|---|---|---|
| Unlinked brand mentions | Highest (0.664 correlation) | Earned media, Reddit, forums, reviews |
| Editorial / earned links | Moderate (0.218 correlation) | News, niche publications, resource pages |
| Structured on-page content | High for retrieval engines | Your own site (FAQ, tables, summaries) |
| Wikipedia / authoritative reference | Very high for ChatGPT | Wikipedia, official docs, .gov/.edu |
| Reddit / community consensus | Very high for Perplexity | Subreddits, Q&A threads, communities |
For benchmarks on what those mentions and placements cost in 2025-2026, our link pricing index and the data in our statistics hub give you a realistic budget before you commit.
Build presence in the sources AI trusts
Since Reddit and Wikipedia stayed among the most cited domains across ChatGPT, AI Mode and Perplexity in Semrush's 2025 tracking, and Forbes rose on all three, these are not optional. Work them deliberately, not spammily.
Reddit and community consensus
Perplexity treats Reddit threads as ground truth for opinions, comparisons, and recommendations. You cannot fake your way in. Earn it: have genuinely useful answers in the subreddits where your buyers ask questions, get your product named organically in recommendation threads, and make sure the framing matches how you describe yourself elsewhere. Consistency across Reddit, your site, and review platforms like G2 is the consensus signal that triggers a citation.
Wikipedia and authoritative reference
For ChatGPT, a Wikipedia entity or a citation on a relevant Wikipedia page is disproportionately powerful. If your brand qualifies for a notable, sourced Wikipedia presence, pursue it through legitimate notability and independent coverage, never paid edits. Where a full article is out of reach, getting cited as a reference on existing topical pages is the next-best foothold.
Tip
Structure content so it gets quoted
For retrieval engines like Perplexity, on-page structure is the lever you fully control. The engines lift well-formed chunks almost verbatim. Make your content easy to extract:
- Answer-first paragraphs. Open each section with a direct, 40 to 60 word answer to the implied question, then expand. This is the format models quote.
- Real FAQ blocks and question-shaped H2/H3s. Headings phrased as the questions people actually ask get pulled into conversational answers.
- Tables with clear headers and row labels. Comparable, structured data is high-signal and gets cited in comparisons and rankings.
- Self-contained sections. Each chunk should make sense without the rest of the page, because the model may retrieve it in isolation.
- Specific, dated, sourced claims. Numbers with attribution read as authoritative; vague marketing copy gets skipped.
Technical hygiene compounds this. Use JSON-LD for Article, FAQPage, and Organization. And confirm the crawlers can reach you: allow GPTBot and OAI-SearchBot for ChatGPT and PerplexityBot for Perplexity in your robots.txt. If you block them, you cannot be cited, full stop.
Strong internal linking also helps retrieval engines understand which of your pages is the canonical answer for an entity, so consolidate rather than scatter your topical authority.
A 90-day plan to earn AI citations
Expect three to six months of consistent publishing and mention accumulation before ChatGPT names you, and faster wins on Perplexity. Sequence it like this:
- Weeks 1-2: Baseline. Run the weekly citation audit, confirm crawler access, and add schema. Identify the 10 queries you most want to win.
- Weeks 3-6: Structure and publish. Rewrite or create the pages targeting those queries in answer-first, table-rich, FAQ format. These are your Perplexity citation candidates.
- Weeks 4-10: Seed mentions. Launch digital PR and earned-media outreach to get your brand named across independent publications. For SaaS specifically, our white-hat SaaS link building playbook maps the credible placement targets.
- Weeks 6-12: Community and reference. Build genuine Reddit presence, pursue legitimate Wikipedia citations, and ensure positioning is consistent everywhere.
- Ongoing: Measure and double down. Track
utm_source=chatgpt.comreferrals (ChatGPT has appended this since June 2025) and watch which sources convert.
Note
chatgpt.com / referral in GA4 because Google does not classify it as a search engine, and visits from apps or links without a referrer can land in Direct instead. Build a custom AI channel group in GA4 rather than trusting default reporting.The payoff justifies the patience. Semrush puts the average AI search visitor at 4.4 times as valuable as a traditional organic visitor, based on conversion rate (Semrush, July 2025), plausibly because the user has already received a synthesized answer that cited you and clicks through with intent and trust. A citation is closer to a referral from a trusted advisor than to a blue link.
Avoid the tactics that backfire
The engines are actively de-biasing toward over-cited sources and getting more resistant to manipulation, which is why Reddit's share crashed in September 2025. Mass-produced AI content, fake reviews, paid Wikipedia edits, and astroturfed Reddit threads all carry downside risk and tend to get washed out on the next model update. If you are buying placements to build the mention graph, do it through real editorial sources and follow our guidance on how to buy backlinks safely. If you are short on outreach time, weigh modern earned-media routes in our roundup of HARO alternatives for 2026.
The durable strategy is unglamorous: be genuinely cited and discussed by credible, independent sources, structure your own content to be quotable, and stay consistent about who you are across the web. That is the same trust the engines are trying to approximate, and there is no shortcut around it. For the broader framework that ties this together, read our pillar on generative engine optimization.