How to get mentioned in ChatGPT and Perplexity answers
Guide · AI Visibility · 6 min read · last verified 2026-07-24
The short answer
Getting mentioned in ChatGPT and Perplexity means satisfying two different selection systems. ChatGPT blends training knowledge with a Bing-based web search and pulls from a wide source range — including third-party pages like Wikipedia and Reddit. Perplexity cites every answer by default and favours sources that are authoritative, recent, and cleanly structured. The move is the same in shape but different in emphasis: clear the bot-access gate, then build the evidence each engine rewards — third-party corroboration for ChatGPT, freshness and structure for Perplexity — and measure your presence per platform so you know which to fix.
One honest caveat up front: neither engine publishes its ranking. You optimize the evidence AI sees, not the model itself. What follows is grounded in a public study and reported platform behaviour, labelled by confidence throughout.
The prerequisite you cannot skip: bot access
Before any tactic matters, the engine must be allowed to read you. Block the crawler in robots.txt and that platform cannot cite you — measured, and observable in your own server logs.
- For ChatGPT: allow `GPTBot` (OpenAI's crawler) and `ChatGPT-User` (its live browsing agent).
- For Perplexity: allow `PerplexityBot`.
Check your robots.txt today — many sites block these bots through a blanket disallow or an over-eager security rule. Bot access is a prerequisite, not a growth lever: allowing the crawler does not earn you a mention, but disallowing it guarantees you never get one. Decide access deliberately — see how to design an AI crawler access policy — and confirm it before you spend a day on content.
How each engine picks its sources
ChatGPT casts a wide net. A mention can come from your site or — more often — from somewhere that talks about you. According to industry analyses of AI citations, Wikipedia makes up roughly 7.8% of ChatGPT's citations and Reddit about 1.8%, and that brands are cited more through third-party sources than through their own domain. Owned pages are table stakes; presence across the wider web is the real lever.
Perplexity behaves like a research assistant. It cites every answer with clickable links, reranks results in several passes for authority and quality, and evaluates fresh content fast — so recent, well-structured pages get a real shot, even from smaller publishers.
| Platform | How it picks sources | Your move |
|---|---|---|
| ChatGPT | Blends training + Bing-based search; wide source range; leans on third-party pages (Wikipedia ~7.8%, Reddit ~1.8% of its citations, per industry analyses) | Build accurate third-party presence; keep Wikipedia/Reddit facts correct; allow GPTBot; write in extractable answer form |
| Perplexity | Always cites; reranks for authority, recency, and structure; evaluates fresh content fast | Publish and refresh often; use self-contained paragraphs and FAQ structure; allow PerplexityBot; make each claim citation-worthy |
For the selection logic both engines share, see how AI assistants choose their sources.
Getting mentioned in ChatGPT
Because ChatGPT reaches for third-party sources, your highest-leverage work happens off your own domain, ranked by leverage:
- Build accurate third-party presence. Reviews, roundups, comparison articles, analyst mentions, and community threads are where ChatGPT finds you. Industry analyses of AI citations show comparison articles take the largest single share (~33%), so a place in credible "best X" and "X vs Y" pages beats another post on your own blog — which is also why vendor sites rarely win citations.
- Get Wikipedia and Reddit facts right. Both are disproportionately cited. You cannot spam them, but you can keep any existing entry accurate and well-sourced, and stop relevant community threads carrying stale or wrong claims about you.
- Confirm GPTBot access. The prerequisite from above — verify it before anything else.
- Write the way ChatGPT answers. Lead each page with a direct, self-contained answer, then support it. Content already shaped like an answer is easier to lift into one.
- Add statistics and citations. According to the Princeton GEO study (KDD 2024), citing sources boosts generative-engine visibility by +40%, adding statistics by +37%, and adding quotations by +30%, while keyword stuffing costs about −10%. Those effects were measured on Perplexity-style engines; applying them to ChatGPT is a derived move, not a guarantee. Off-site versus on-page effort is itself a strategy call — third-party corroboration versus own-site AEO.
Getting mentioned in Perplexity
Perplexity rewards the qualities of a good citation — current, structured, worth quoting:
- Allow PerplexityBot. The prerequisite — confirm it first.
- Publish and refresh for recency. Perplexity favours recent sources, so a visible publish/updated date and a real refresh cadence improve your odds (derived from reported time-decay behaviour).
- Structure for clean extraction. Use self-contained paragraphs that answer one thing each, descriptive headings, and FAQ-style Q&A. Perplexity pulls atomic passages, so give it passages that stand alone.
- Make every claim citation-worthy. This is where the Princeton effect sizes land hardest — the study ran on Perplexity itself. Cited sources, statistics, direct quotations, and an authoritative tone all lift visibility; each measured effect is in the table below.
- Put authoritative evidence in public. Public research, data, and reference pages give Perplexity something concrete to cite; evidence gated behind a form the crawler cannot read does not help.
By measured effect and effort:
| Content move | Visibility effect (Princeton GEO, KDD 2024) | Effort |
|---|---|---|
| Cite named sources | +40% | Low |
| Add relevant statistics | +37% | Low–medium |
| Add direct quotations | +30% | Low |
| Authoritative, non-salesy tone | +25% | Low |
| Improve clarity / fluency | +15–30% | Medium |
| Keyword stuffing | −10% (avoid) | — |
Measure per platform, then act, then re-measure
The trap is optimising blind. ChatGPT and Perplexity disagree constantly — present in one, absent in the other for the same question — so a single "AI visibility" number hides where the real gap is. Track presence per engine, per buyer question, on a benchmark you hold still.
This is the loop Magrios runs: measure how each engine answers your priority questions, see the specific gaps (absent from Perplexity on three questions; present but mis-described in ChatGPT on two), fix them, then re-measure on the same locked benchmark so the delta is real movement, not a reworded prompt or a quiet model update. Every claim carries a source link back to its evidence. Which signals to track is covered in AI visibility metrics that matter; the remediation mechanics in how to get cited by AI search engines.
What you can and cannot control
Be honest about the opacity — it changes how you invest:
- Measured: bot access (your logs), whether you are indexed, and the Princeton effect sizes on Perplexity-style engines — observable, or from a published study.
- Derived: that ChatGPT's third-party and freshness preferences respond to the same evidence moves — inference from reported behaviour, not a controlled result.
- Hypothesis: the exact weighting each engine applies, and how long a change takes to surface — neither platform discloses this, and both re-rank without notice.
No one can promise guaranteed placement in ChatGPT or Perplexity. What you can do is clear the access gate, build the evidence each engine rewards, and measure per platform — so your next move is aimed, not hopeful.