AI search doesn't return results. It returns a decision.
Across ChatGPT, Perplexity, AI Overviews, Claude and Copilot, the user gets a shortlist and acts on it. AI search optimization is the work of being on that shortlist, measured on every surface, every week.
Measure the shortlist, then change it.
Every step is evidenced by a stored answer.
Six surfaces, six behaviours.
They retrieve and cite very differently.
ChatGPT
Browses selectively; rewards clear factual pages and published numbers over marketing language.
Perplexity
Cites visibly and frequently. The fastest surface to win, and the fastest to lose.
AI Overviews
Leans on established entity signals, structured data and conventional search authority.
Claude
Prefers documentation-grade sources and precise technical statements.
AI Mode
Conversational follow-ups mean one weak answer cascades through a whole session.
The top of the funnel moved.
You are evaluated in a conversation you cannot see.
About AI search.
ChatGPT, Perplexity, Google AI Overviews, Google AI Mode, Claude and Microsoft Copilot.
No. Overlap is partial — being named in Perplexity often does not translate to AI Overviews, which weights entity signals differently.
Mention share, citation share and position within the answer, per surface, from stored responses.
No. It runs alongside it, and most of the technical work benefits both.
Retrievability and factual fixes land in days. Competitive shortlists usually take six to ten weeks.
See the shortlist you're missing from.
Free, no account, and it names the competitors who are on it.