Definition
Also called: ICP-based AI search, ICP Based
Documented
Much AI visibility tracking starts from keywords, or from prompts anyone might type. ICP-Based AI Search starts from the ideal customer profile: who buys, in what situation, and how they phrase their questions. In Obi's version, the agent reads the company's website and what the founder tells it, then proposes up to four segments, each a buyer and the situation they are in. The founder confirms the segments before any question is written.
From there, every question is written the way a buyer in that segment would ask an AI, and each is ranked by its likely impact on a purchase. Visibility on a question nobody in the ICP asks counts for little; a recommendation on the question a qualified buyer asks before a demo counts most.
How it works
Segments first, confirmed by the founder; then questions per segment, in the buyers' words; then a ranking by buyer impact that orders what is measured and what is worked on.
Why it matters
Being mentioned on generic prompts brings generic traffic. A startup with a narrow market needs the AI answers its own buyers read to name it, which means measuring and working on those questions first. Starting from the ICP keeps the effort on the questions closest to a sale.
Example
A medtech startup confirms one segment, clinical operations leads at mid-size hospitals; the questions written are the ones those leads ask ChatGPT about patient intake software, ranked by how close each is to a buying decision.
Sources (1)
- Obi, FAQ: How does Obi find my ICP? and Where do Obi's questions come from?, published 2026-11-06, read 2026-11-08: https://hireobi.com/faq
Checked