How to get cited by ChatGPT, Perplexity and AI Overviews
Generative engines cite pages that are easy to quote and easy to trust. A practical structure for articles that get picked up, based on what the research actually measured.
Quick answer
To be cited by generative engines, publish pages that answer a specific question directly, back each claim with named sources, statistics or quotations, show a real author with credentials, keep the content crawlable by AI bots, and expose the structure through headings, FAQ markup and a machine-readable site summary.
When someone asks an AI assistant “which agency should we hire for local SEO in Portland?”, the assistant does not return ten blue links. It reads a handful of pages, composes an answer and cites two or three of them. Being one of those citations is the new first page. The good news is that the behaviour is predictable: generative engines behave like diligent readers, and diligent readers reward the same things good editors do.
Start with one question per page
A page that tries to answer everything is hard to quote. Pick the question your reader actually types or says, put it in the title, and answer it in the first paragraph in 40–60 words. The rest of the article earns that answer: context, evidence, exceptions, next steps. This is the same discipline that wins featured snippets, which is not a coincidence; the engines share the same retrieval instincts.
Make every section quotable
Use headings that state the sub-question, then open the section with the claim before the explanation. Avoid pronouns that depend on the previous paragraph. A sentence like “Local service businesses should refresh their Google Business Profile at least monthly” can be lifted into an answer; “You should do this too” cannot.
Add the three signals the research measured
The 2024 GEO paper tested nine content changes against a benchmark of generated answers. Three of them stood out consistently:
- Citations. Link the claim to a credible source, inline or in a references list at the end of the article.
- Quotations. Quote a named authority, with attribution, rather than paraphrasing anonymously.
- Statistics. Replace “many businesses” with a number and its source.
Each of these raised a source’s visibility in generated answers by up to about 40% in the benchmark. Adding keywords, the old reflex, reduced it. The engines are looking for evidence, not repetition.
Show who wrote it
Generative engines inherit the search engines’ trust signals. A named author, a job title, a short bio and a link to a professional profile tell both humans and models that the page is accountable. Pair it with Organization and Person structured data so the relationship is machine-readable, and keep a visible “last updated” date so the engine knows the page is maintained.
Let the bots in
Many sites block AI crawlers by default without deciding to. Check your robots.txt for GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended, and decide deliberately. If you want citations, allow them. A short llms.txt file at the site root, listing your most important pages with one-line summaries, gives those crawlers a map instead of a maze.
Measure citations, not just rankings
Keep a fixed list of 20–50 prompts that matter to your business and check monthly whether your pages are cited in ChatGPT, Perplexity and Google AI Overviews. Record which page was cited and for which claim. Over time this tells you which sections are quotable and which are being ignored, which is far more actionable than a keyword ranking.
Sources
- Aggarwal, P., Murahari, V., Rajpurohit, T., Kalyan, A., Narasimhan, K., & Deshpande, A. (2024). GEO: Generative Engine Optimization. KDD '24.
- OpenAI. Overview of OpenAI crawlers.
- Perplexity. PerplexityBot and Perplexity-User.
- llmstxt.org. The /llms.txt file.
- Google Search Central. AI features and your website.
Good to know
FAQ
Do AI engines only cite big brands?
No. They cite the clearest, best-supported passage they retrieve for a prompt. Smaller sites with specific, well-sourced pages are cited regularly, especially for local and niche questions where large publishers are vague.
Should we write differently for each engine?
Not at the content level. The same structure (direct answer, evidence, authorship, crawlability) works across ChatGPT, Perplexity, Claude and Google. Differences show up in which crawlers you allow and in how each engine presents citations.
Does structured data help with AI citations?
Indirectly. Structured data helps search engines understand the page, and generative engines lean on search indexes for retrieval. FAQ, Article and Person markup also make the question–answer structure explicit, which makes passages easier to extract.
Cite this article
Authoric (2026, October 19). How to get cited by ChatGPT, Perplexity and AI Overviews. Authoric. https://authoric.net/articles/how-to-get-cited-by-chatgpt-perplexity-and-ai-overviews/