Getting cited by AI: the on-page moves that work

There is a cottage industry of AI-search tricks: chunk your content into tiny blocks, stuff in an AI text file, write in a special robotic style. Google looked at the list and said, in effect, skip most of it. Its official guidance calls these tactics things "you can ignore." What earns citations is duller and harder.

Write something only you could write

Google's strongest statement is about content, not markup. It says unique, non-commodity content will shape your presence in AI search "more than any of the other suggestions" it offers. A first-hand review beats a summary of other reviews. A real teardown beats "7 tips for beginners." The model reads many sources; give it a reason to pick yours.

This is also why the citation data favors a few destinations. ChatGPT leans on Wikipedia, Perplexity on Reddit, because those carry information the model judges as primary, not recycled. Your page competes on the same axis.

Make the structure easy to lift

AI answers are assembled from passages. Help the model find the passage that answers the question.

• Lead each section with the answer, then explain. The model, and the reader, both reward it.
• Use clear headings that name the question a section answers.
• Keep one idea per section so a passage can stand alone when quoted.

None of this means chopping your page into fragments. Google is explicit that you do not need to "chunk" content; its systems handle nuance on a normal page. Write for a person who skims.

Earn authority the slow way

AI models pass along signals of expertise and trust. First-hand experience, a named author, real data, and citations to primary sources all read as authority. So does being referenced elsewhere, but only when it is genuine. Google warns against chasing "inauthentic mentions"; manufactured buzz does not move the needle and can trip spam systems.

What about schema and llms.txt

Add structured data, but for the right reason. Google says it is not required for AI search, though it helps you qualify for rich results, which is reason enough. As for llms.txt, be clear-eyed: Google has said plainly it does not use the file. Some other assistants may. Treat it as a low-cost bet, not a growth lever.

The pattern under all of this is consistent. The work that wins AI citations is the work that has always won search: say something true and useful that the next person cannot easily copy. The tools change. The bar did not.

Run your free audit → Back to Academy