How to Get Cited by ChatGPT and Perplexity: A Practical GEO Playbook

Knowing that AI search matters is one thing. Getting ChatGPT or Perplexity to actually pull your page into an answer is a different, much more mechanical problem - and it comes down to a handful of specific, repeatable choices you make on the page itself.

Write for the question, not the keyword

Traditional SEO trained everyone to write headings around keywords - "Digital Marketing Services", "Best SEO Company". AI engines respond much better to headings phrased as actual questions, because that's closer to how the underlying query was asked: "What does a digital marketing agency actually do?" or "How much should local SEO cost?"

Directly under that heading, the first sentence or two should answer the question in full, in 40 to 60 words, before you get into nuance or examples. This is the single highest-leverage change most sites can make. AI engines lift text that reads like a complete, self-contained answer far more often than text buried three paragraphs into a scene-setting introduction.

The schema types that actually matter

Not all structured data carries equal weight for AI citation. Three types consistently show up on pages that get cited:

• Organization schema - tells the engine who you are, unambiguously, so it can attribute an answer to a real, identifiable entity.
• FAQPage schema - if you're already writing question-and-answer content, marking it up formally makes it trivially easy for a machine to lift.
• HowTo schema - for any process-based content (setting up a campaign, fixing a technical issue), this gives the engine a pre-structured sequence of steps rather than making it parse prose.

Most sites we look at have schema only on the homepage, if at all. Every page that's meant to answer a specific question should carry the schema type that matches its content.

Formatting patterns AI engines actually lift

Beyond schema, plain formatting choices make a measurable difference. Numbered steps get lifted whole for "how do I..." queries far more often than the same instructions written as a flowing paragraph. Short definition sentences - "X is..." - get lifted for "what is..." queries. Comparison tables and short bullet lists get lifted for "X vs Y" queries. In every case, the pattern is the same: the AI is choosing the source that requires the least interpretation to turn into an answer.

This doesn't mean writing badly or stripping out nuance. It means putting the extractable version first, and letting the deeper explanation follow after.

A five-minute monthly audit checklist

1. Ask ChatGPT and Perplexity your five most important customer questions and note whether your site is cited.
2. If it isn't, open the page that should have answered it and check whether the answer sits in the first 60 words under a matching heading.
3. Check that page for FAQPage or HowTo schema, and add it if missing.
4. Re-check the same five questions a few weeks later to see if anything shifted.

None of this is a one-time project. AI engines re-crawl and re-evaluate sources continuously, which means a page that isn't cited today can start getting cited a month from now, purely from these structural fixes.