The LLM-friendly content structure (with examples)
A repeatable content template that earns citations across ChatGPT, Perplexity, Gemini and Google AI Overviews. With before/after examples.
The template below is the result of analyzing 4,000 LLM citations across four engines. It is deliberately boring — boring is what gets cited.
1. The TL;DR block
Three bullets, 18–24 words each, written as standalone claims. LLMs lift them into answers because they are pre-summarized.
2. Question-shaped H2s
Turn every H2 into a question users would actually type. The H2 becomes the prompt; the paragraph becomes the response. This is the single highest-leverage move.
3. Self-contained paragraphs
No 'as we said above', no 'see below'. Each paragraph must make sense if pasted into a chat reply on its own.
4. Hard numbers near the top
One concrete statistic in the first 200 words gives the model something to attribute to you specifically. Cite your own source for it.
5. The FAQ block
Four to six questions, 30–60 word answers. Render with FAQPage schema. This block alone earns 30–40% of our LLM citations.
Frequently asked questions
- Does word count matter?
- Less than structure. 1,200 well-structured words beat 3,000 unstructured ones.
- Should I write for LLMs or for humans?
- Both. The structure that helps LLMs (clear questions, short paragraphs, lists) also helps humans skim.
- Can I retrofit old posts?
- Yes, and it is the highest-ROI work you can do this quarter.