Stage 2 · Prompt Research

Stage 2 maps the conversational demand: what your buyers actually ask engines, how engines decompose those questions into follow-ups, and where you're absent from the answers. It's the antidote to the most expensive production mistake there is: writing content for questions nobody asks. Keyword lists tell you what people type into a search box; prompt research tells you what they say to an assistant, which is longer, more conversational, and chained into follow-ups that are the real buyer journey.

The method is harvest → expand → cluster → score. You gather real prompts from where they live (community threads, sales calls, support tickets, autocomplete, competitor comparisons), follow each seed down its conversational follow-up chain, cluster them so that one cluster equals one page that can fully answer it, and then rank by business value × AI-answer presence × your current absence. The output is a prioritized prompt-to-page map, the production queue Stage 3 works from. This node carries no impact score; its leverage is realized when the pages it prioritizes get produced and cited.

Gather real prompts from where buyers actually ask, follow the conversational chains, cluster them into page-sized units, and rank by the gap; the output is a production queue Stage 3 can build straight from.