Platforms

Every answer engine retrieves from different indexes and trusts different sources, so a tactic that wins Perplexity can be invisible in AI Overviews. This pillar is the per-platform layer: how each engine actually assembles answers, and the specific levers that move each one. The universal craft from Pillars 3–5 still applies underneath, but which surfaces it points at, and which signals dominate, changes engine by engine.

The number that makes per-platform playbooks mandatory rather than optional: only about 11% of cited domains overlap between ChatGPT and Perplexity for the same query, and 71% of cited sources appear on only one engine. You cannot "optimize for AI" in the singular; you optimize for ChatGPT's Wikipedia lean, Perplexity's Reddit-and-freshness bias, AI Overviews' Google index, and so on, each measured separately. The nodes below are the field guide to each surface, plus the emerging ones worth a cheap hedge before they matter.

Don't read all eight and optimize everything at once. Prioritize the engines your audience actually uses, learn each one's bias, and measure them separately; the 11% overlap means a win on one tells you little about another.