Customer discovery interviews: an FDE playbook
Updated
How to run a discovery interview that ends in a measurable problem statement: questions, traps, and the write-back habit.
A discovery interview has one output: a problem statement with a number. Everything below serves that.
The short answer
Ask about today, not about wishes. Watch the work, quantify the pain, write the brief back, get sign-off. Then build.
Before the call
Read what exists: tickets, the spreadsheet that runs the department, last quarter's incident notes. Arrive with three hypotheses and hold them loosely. The FDE Foundations M02 lessons teach the interview spine; every lesson is free.
The question spine
- "Walk me through the last time this broke." (Story, not opinion.)
- "Who else touches this before and after you?" (The real workflow.)
- "What does it cost when it fails: minutes, money, customers?" (The number.)
- "What have you already tried?" (Avoid rebuilding the workaround.)
- "If we fix one thing, what changes in your week?" (Success criterion.)
Worked example: the triage that was really search
A fictional ops team (fictional) asks for "a better triage dashboard." Two interviews reveal operators spend 35 minutes per case finding the original order across three systems. The brief reframes: cut lookup under 3 minutes for 9 of 10 cases. The build is a lookup page, not a dashboard. Same budget, real problem.
Traps
| Trap | Escape |
|---|---|
| Manager summarizes instead of operator shows | "Can you share your screen and do one live?" |
| Wishlist grows | "Which one hurts this week? We start there." |
| Solution dictated | "Help me understand the pain first; the shape follows." |
| No numbers | "Roughly: how many a week, how long each?" |
Checklist: after every interview
- One paragraph of what happens today, with a number.
- Names of two more people to talk to.
- The brief updated and sent back within 24 hours.
- Explicit sign-off on scope and success metric.
Related reading
Straight answers
Frequently asked questions
How long should a discovery interview be?
Twenty to forty minutes with one operator beats two hours with a manager. Short, repeated, with the people who feel the pain.
What is the biggest discovery mistake?
Solution-first questions. Asking 'would you like a dashboard' instead of 'walk me through Tuesday' buries the real problem.
How many interviews are enough?
Until the brief stops changing: usually three to five operators plus one owner. Same pain in different words means you found it.