--- name: customer-interview description: Plans and scripts customer discovery interviews using The Mom Test. Use when a user wants to plan a customer interview, write discovery interview questions, build an interview script, generate Mom Test questions, or talk to users about a problem. Produces a non-leading question script grounded in past behavior, plus a note-taking template. --- # Customer Interview Most interviews fail before they start: the interviewer pitches their idea, asks people to predict the future, and walks away with polite lies that feel like validation. The Mom Test (Rob Fitzpatrick) fixes this — talk about the customer's life and what they've actually done, not your idea or what they think they'd do. This skill plans a discovery interview and writes a script that can't be answered with empty compliments. It's the BEFORE; once you've run the interviews, use the `synthesize-research` skill to analyze what you heard. **Grounded in:** *The Mom Test* — Rob Fitzpatrick: ask about specific past behavior, never pitch or ask hypotheticals. **Go deeper (The Product Channel):** [How to Discover Customer Needs](https://sidsaladi.substack.com/p/how-to-discover-customer-needs) ## When to use this - You're about to talk to users and want a plan + question script, not just a vibe - You're testing a risky assumption (is this problem real? painful enough? who has it?) - You keep getting "yeah I'd totally use that" and want truth instead of flattery - You're doing problem discovery, not a demo or a sales call ## The Mom Test rules (the spine of a good interview) - **Talk about THEIR life and past behavior, not your idea.** Their world is fact; your idea is a hypothetical they'll be nice about. - **Ask about specifics in the past, not generics or the future.** "Tell me about the last time…" beats "Do you usually…" beats "Would you…". - **Talk less, listen more.** Aim for them talking ~80% of the time. Silence is a tool — let them fill it. - **Dig:** "tell me about the last time that happened" · "what did you do about it?" · "how much does it cost you today (time/money/sanity)?" · "what have you already tried?" · "why is that hard?" ## Retention / churn / win-back interviews (invert the frame) The standard problem-story frame assumes the pain is external to your product — you're hunting for a problem worth solving. For retention, churn, and win-back interviews the product is already in the story, so **invert it**: anchor on the last time they actually USED the product and the specific moment or week engagement trailed off, not on a generic problem. - **Start from real usage, not the problem.** "Walk me through the last time you used " → "When was the last time before that?" Find the moment the rhythm broke. - **Pinpoint the drop-off.** "Which week did you stop reaching for it, and what changed that week?" Dated specifics beat "I just got busy." - **Find who disengaged first.** In a team or household, one person usually drifts first — "who stopped using it before you did?" surfaces the real failure mode. - **Ask what filled the gap.** "What do you do now instead?" The replacement (a competitor, a spreadsheet, nothing) tells you what job you stopped doing well. - **Stay out of pitch mode.** It's tempting to defend or re-sell the product when they describe leaving it. Don't — the goal is the honest story of why it stopped earning their time, not a save attempt. ## Before you start (gather these) - **Who** you're interviewing and which segment they represent (role, context, why them) - **The riskiest assumption** you're trying to kill — the one belief that, if wrong, sinks the idea - **The learning goal**: the single most important thing you must learn from this conversation - **Logistics**: length (20–40 min is plenty), recording consent, who takes notes If any of these are unclear, ask the user before generating the script — a script with no learning goal produces unfocused interviews. ## Process 1. **Define the learning goal + riskiest assumption.** Write one sentence: "I believe [X], and if it's false, [consequence]." That's what the interview must test. 2. **Write past-behavior questions mapped to it.** For each assumption, draft questions that surface real, dated behavior ("the last time…", "walk me through how you…"). Every question should trace back to the learning goal — cut the rest. **Budget the time:** cap the script to the handful of questions that map to the riskiest assumption, and treat the probe follow-ups as optional depth you reach for when an answer is rich — not a checklist to grind through. That keeps a session to ~30 min and leaves room for them to talk. 3. **Sequence the conversation:** warm-up (rapport, context on their role) → context (how their day/workflow actually works) → problem story (the last time the pain hit) → current workarounds (what they do today, what they've already tried, what they've paid for) → wrap (anyone else I should talk to? anything I should've asked?). 4. **Prep follow-up probes.** For each main question, stage 2–3 digging probes so you can go deeper on the spot instead of moving on too soon. 5. **Set up note capture.** Have the note template open so you log verbatim quotes and signals live, not from memory afterward. ## Output template Produce three copy-pasteable blocks. **1. Interview plan** ``` LEARNING GOAL: RISKIEST ASSUMPTION: I believe ; if false, . SEGMENT: — chosen because WHO (repeat per interviewee; same plan reused across the segment): - — - — … LOGISTICS: · recording consent? · notetaker: SUCCESS LOOKS LIKE: ``` **2. Question script** (grouped, non-leading, with probe follow-ups) ``` WARM-UP - Tell me a bit about your role and what a typical week looks like. CONTEXT - Walk me through how you currently handle . ↳ probe: What tools or people are involved? Where does it break down? PROBLEM STORY - Tell me about the last time came up. ↳ probe: What did you do about it? How long did it take? ↳ probe: How much did that cost you — in time, money, or stress? ↳ probe: When did you first notice this was a problem? ⤿ RETENTION / CHURN VARIANT — swap this whole block for the inverted frame: - Walk me through the last time you actually used . ↳ probe: When was the time before that? (find where the cadence thinned) ↳ probe: Which week did it stop being part of your routine — what changed that week? ↳ probe: Who stopped first, and what filled the gap? CURRENT WORKAROUNDS - What have you already tried to fix this? ↳ probe: Why did/didn't that work? Have you paid for anything to solve it? ↳ probe: What do you do today when it happens? WRAP - What's the hardest part of this for you, and why? - Who else deals with this that I should talk to? - Is there anything I should have asked but didn't? ``` **3. Note template** ``` | Quote (verbatim) | Observation | Pain (1–5) | Workaround | Surprise | |------------------|-------------|-----------|------------|----------| | | | | | | ``` After the interviews, hand the filled-in notes to the `synthesize-research` skill to cluster themes, rank findings by frequency and impact, and turn them into recommendations. ## Avoid (anti-patterns) - **Leading questions / pitching your solution** — "Don't you hate when…?" or "Would this feature help?" plants the answer. - **Hypothetical "would you use…" questions** — people are terrible at predicting their own future behavior; ask what they did, not what they'd do. - **Asking for opinions instead of past behavior** — "Do you think this is a good idea?" gets you flattery, not facts. - **Fishing for compliments** — if you leave feeling great, you probably learned nothing. Useful interviews surface friction, not praise. ## Tips - Bad data is worse than no data — a false "yes" sends you building the wrong thing. - Anchor every claim to a specific moment: "when exactly?" turns a generic into evidence. - Watch for emotion and curse words — that's where the real pain lives. - If they say a problem matters, ask what they've spent (time/money) on it. Commitment > compliments. - Embrace awkward silence after a question; the second thing they say is usually the honest one. - Take notes in their words, not your paraphrase — verbatim quotes are gold in synthesis.