Appearing in LLM answers requires both structured, extractable content on your own site and mention density across third-party sources the model trusts.
[01]/ the data
What we found
Every engine we track weighs these two factors differently: Gemini leans on your own schema, ChatGPT/Claude lean on external mention density.
[02]/ how to do it
Three steps
Fix your own schema/structure first (fastest lever).
Build external mention density in parallel (slower, higher ceiling).
Track citation by engine, since the two levers pay off differently per engine.