Discovery Interview Scripts

Same questions, every conversation.

Variability in answers is signal; variability in questions is noise. Pick the script that matches the slot length, read the prompts as written, capture verbatim phrases, code afterwards.

Pick a script Voice rules Falsifiable claims (C1–C7)

Available scripts

All scripts share the same five-block shape and the same C1–C7 anchoring. The trade-off is depth — the shorter version preserves the load-bearing trigger event and trims probes elsewhere.

Customer Discovery — 30 min

30 min total 5 intro + 25 blocks 5 blocks

Tight slot. Lead with the load-bearing trigger event; trim the rest.

The default for first conversations and most cold-outreach replies. Five minutes of intro absorbs role/team color; the structured blocks fit in 25 minutes with C1–C7 still anchored. Block 2 (the trigger event) stays load-bearing — don't trade time out of it.

When to use

A 30-minute slot. Cold-outreach replies. Anyone whose calendar you do not yet have permission to fill.

Open script → Present mode →

Customer Discovery — 60 min

60 min total 5 intro + 55 blocks 5 blocks

Full canonical script. Use when the prospect has the time.

The complete script. Five blocks, 60 minutes, all probes intact, both the skill-distribution wedge probe and the LLM-vs-deterministic-line question retained for higher-signal conversations. Use when a warm contact has given you the time, or for a second conversation after a 30-min first pass.

When to use

A 60-minute slot. Warm contacts. Second conversations. Anyone willing to go deep on the integration surface and the skills-distribution pattern.

Open script → Present mode →

How to use these scripts

Before

Open the script and copy it into your note-taking surface.

Each question has a blank Notes: field underneath when pasted. That's where you write what they actually said, verbatim where possible.
During

Read each question as written. Resist paraphrasing.

Watch the timing marker. Spend the most time in Block 2 (load-bearing). Defer the pitch to Block 4 even if you're tempted earlier.
After

Code the conversation against C1–C7 within 24 hours.

Memory of the interviewee's tone decays fast. The coding sheet is what makes the decision gates at 10 / 20 / 30 conversations meaningful.

Voice rules (apply to every script)

Past behavior, not prediction. Every 'would you' is a 'did you' waiting to happen. People are bad at predicting their behavior, fine at recounting it.
Silence is a tool. After a question, wait. The most useful answers come after the awkward pause. Five seconds feels long. It is not long.
Anchor on their language. Borrow their words back to them. If they call it 'the bot', you call it 'the bot'. Do not translate to 'agent'.
Don't lead with architecture. Shell mediation, PTY approvals, TEE attestation — or with composable-primitives vocabulary (skills, connectors, actions) — none of these words appear in Blocks 1 through 3. Block 4 is the only place architecture enters, and even there only as the pitch line.
Don't sell. Selling and learning are mutually exclusive in the same conversation. If you find yourself describing the product outside Block 4, stop.
Capture verbatim phrases. When they say something quotable about their pain, write the exact words down. Paraphrases lose the texture that informs the outreach next iteration.
Skills, Connectors, Actions — only if they say it first. These are Aileron's architecture words and a candidate wedge. If they use them, mirror. If they don't, do not introduce them — that's pitch leakage in Blocks 1 through 3.

After the interview

The conversation is half the work. The other half is converting it into a coded row that the decision gates can read. Per-claim support / contradict / neutral, plus the trigger event, rejection reasons, and budget signal in free text.

For the discovery context that drove the current iteration of these scripts (skills-sprawl probes, the Block 4 pitch revision), see 2026-05-28 Andrew Gordon.

Coding sheet

Each interview becomes a meeting page under /meetings/ with C1–C7 codes in its frontmatter. Per-claim tallies and the decision-gate counter aggregate live at /discovery/matrix/. Capture mechanism tracked as issue #5.

Decision-gate checkpoints

Re-read the coded matrix at 10, 20, and 30 conversations. Decision-gate criteria on the discovery plan.