interviews

/interviews owns the discovery-interview lifecycle end to end — drafting a script around your actual hypotheses, helping you find people to talk to, then turning what they said into linked evidence that moves your bets.

​

When to reach for it

• “Help me prep for discovery calls” / drafting a first script
• “I don’t know who to talk to”
• Refining questions after a few interviews, or drafting a script for a second segment
• “I just talked to someone — here’s what they said” (file, paste, or memory)
• “How did I do as an interviewer?”

​

Drafting your first script

A guided conversation, one question at a time: confirm the target persona (one script = one persona — a second segment gets its own script later), pick a length (15/30/45/60 minutes — which sizes the number of topics, biased toward fewer-and-deeper), and say how much interviewing you’ve done (which tunes how much supporting scaffolding each topic gets). The advisor then maps your hypotheses into conversation topics and drafts section by section — opening, topics, closing — showing exact text and revising as you react. The result is deliberately not a questionnaire. Scripts are organized around topics to explore — learning themes with a “why it matters” line linking the hypotheses at stake, and a few starting questions framed as prompts. The point is a free-flowing conversation that goes deep where it gets interesting, not a checklist read off a teleprompter.

​

Finding people to talk to

Can’t name anyone to interview? The skill treats that as signal worth acting on, and walks you through building a prospect list of 10–20 people you can actually reach — people you’ve worked with, follow, or have seen complain publicly — then writing short, specific, honest emails with a low ask (a 20-minute call). Goal: 3–5 conversations scheduled. Non-responses are information, not something to chase.

​

Analyzing an interview

Bring what you have — a transcript file, pasted text, or just your recollection of the call — and the skill runs the loop-closer:

1. Your raw material is saved as a transcript (with an intake question about who you actually talked to — divergence from who you targeted is itself signal).
2. A bias-isolated analysis extracts the statements that matter — factual, behavioral, or belief-bearing quotes, each tagged and linked to the hypotheses it bears on — and reviews your interviewing technique against Mom’s Test principles.
3. A second independent pass reassesses your hypotheses across all interviews to date.
4. You get one summary: what changed → next action per touched hypothesis, the sharpest next move, any candidate new hypotheses that emerged, and 1–3 concrete technique notes (“you jumped past her ‘overkill’ comment without probing”).

Status changes and new hypotheses are confirmed per item before anything is written.

​

What it writes

​

Good to know

• Prerequisites are soft: scripts work best with an Audience defined and ≥3 hypotheses to build around — the skill suggests both, blocks on neither.
• Recollection-based analyses are fully supported but yield weaker technique feedback — your own questions weren’t captured.
• Statements that don’t fit any hypothesis stay unlinked deliberately; when the same theme recurs across interviews, it comes back as a proposed new hypothesis.
• Topic coverage is weighted honestly: going deep on one high-signal topic is a win, not a failure to cover all five.
• Older rigid question-list scripts get a one-time offer to convert to the topic format — opt-in, never forced.

​

Related