How-to guides are the workhorse of LLM citations. When a user asks "how do I [task]?", the AI searches for a source that contains structured, numbered steps for completing that task. Guides that provide clearly numbered, self-contained steps with an explanation of why each step matters are extracted and reproduced at far higher rates than prose descriptions of the same process.
The Optimal How-To Structure for LLMO
Title: "How to [Specific Task] in [N] Steps" — the title should match the user's likely query verbatim. Opening paragraph: One sentence defining what this guide achieves and who it is for. Steps: Numbered, bold step labels followed by 2–3 sentences explaining the step and why it matters. Each step must be independently understandable — the LLM may extract one step without the context of others. Caveats: A "What if" or "Common mistakes" section adds citable context for edge cases. FAQ: 3 questions about the most common points of confusion in the process. Add HowTo schema markup to maximise structured data extraction.