audience-research

Audience Research

Research that produces topics is useless. "They care about scalability" is a topic; it cannot be written from, argued with, or headlined. "We rewrote it twice and I'm not doing that again" is a sentence a real person said, and it is the piece.

So this skill collects sentences, verbatim, with a source and a date. A paraphrase is the founder's words wearing the audience's clothes — and the founder's words are the problem, since they are the reason prospects say "I didn't really understand what you do".

When to use

Before content-plan, and whenever the plan starts filling with topics. Also when the founder is about to open a keyword tool, which is a distant fourth source and the only one that never argues back.

Inputs

Read first, in order — house rule 1. The ranking is the method:

1. pipeline.md — what prospects actually asked on calls, and the objection the founder answers every single time. This is the best data in the company and it is already in the workspace, unread.
2. clients/ — what buyers said before they bought, and what they complained about after
3. Public writing by ICP-fitting people — their posts, their forum threads, and their job ads. Job ads are the most underrated source in existence: the ICP describes their own pain, publicly, in their own words, with a budget attached, in a document written to be read.
4. Keyword tools — last, and only to check volume on a phrase you already heard a human say.

• offer.md — the ICP. A quote from someone outside it is noise, and it is attractive noise, because there is much more of it.
• ingestion-gate — every line this skill writes is a sentence a person said, so the gate is not an extra step here, it is the ranking above restated as a rule. Source 3 is where it bites: a public post is the ICP describing their own situation, which is FACT, right up until it is a vendor describing a market, which is paid speech and does not enter.