Keyword research for AI search

Start from real data, not guesses

Search volume and competition data grounded in what people actually type is still the foundation, whether the target is a Google ranking or an AI-generated answer. Guessing at topics without checking real demand wastes effort on questions nobody's asking.

Group by question and intent, not just volume

A flat list of a hundred keywords sorted by search volume tells you what's popular, not what's useful. Clustering the same list by the actual question behind each term, and by where that question sits in a buying decision, turns a spreadsheet into a content plan you can act on.

• Definitional questions ("what is X") — top of funnel, high volume, low immediate intent
• Comparison questions ("X vs Y") — mid funnel, the reader is actively deciding
• Evaluation questions ("how to choose X", "is X worth it") — closer to a decision
• Direct commercial questions ("best X for Y", pricing questions) — highest intent

Buyer-intent topics are where AI citation and conversion overlap

Comparison and evaluation questions are exactly the ones people ask ChatGPT and Perplexity directly, and they're also the ones closest to an actual purchase decision. Prioritizing that middle band of intent, rather than chasing the highest-volume definitional terms, is usually the better trade for both AI visibility and pipeline.

Don't skip the low-volume, high-intent terms

A term with modest search volume but obvious buying intent (a specific comparison, a specific use case) is often worth more than a high-volume definitional term that attracts readers who aren't close to a decision yet. Volume alone will consistently steer you toward the wrong priority.