Writing an Interview Guide You Can Actually Follow

A good interview guide is a map, not a script: a short list of goals and open prompts organized around story arcs, with an escape hatch for tangents worth following. It keeps you anchored to what you need to learn without forcing you to read questions off a page while a customer's real story tries to happen around you.

Quick Answer: Write goals, not questions. Structure the guide as 3-4 "arcs" (warm-up, core story, specifics, close), each with one goal and 2-3 open prompts. Leave room to abandon the plan when the customer hands you something better.

Why a Verbatim Script Wrecks the Interview

A word-for-word script optimizes for the interviewer's comfort, not the customer's candor. The moment you're reading questions instead of listening to answers, you signal that this is an interrogation, not a conversation — and people self-censor accordingly.

Scripts also fail structurally. Real conversations don't move in the order you wrote your questions. A customer answers question 4 while responding to question 1, and a rigid script leaves you either re-asking (which feels robotic) or awkwardly skipping ahead (which breaks flow). Teresa Torres, creator of the opportunity solution tree method, has long argued that discovery interviews should elicit specific past stories rather than opinions or hypotheticals — a script optimized for coverage instead of story-elicitation tends to produce exactly the abstract, unreliable answers you're trying to avoid.

The opposite failure — no plan at all — is just as costly. Steve Portigal, author of Interviewing Users, describes unstructured "friendly chat" interviews as ones that feel great in the room and yield almost nothing analyzable afterward, because there's no consistent frame across conversations to compare against. You end up with 45 minutes of rapport and no evidence.

The Guide as a Compass, Not a Script

A guide built around goals and prompts solves both problems. Each section states what you need to learn, not the exact words to say — so you can rephrase on the fly, follow a tangent, and still land where you needed to. If you're new to the distinction between question types that unlock this kind of flexibility, see this breakdown of open, leading, and closed interview questions — the guide's prompts should almost all be open.

The Four-Part Structure: Warm-Up, Story Arcs, Specifics, Escape Hatch

A followable guide has four parts: a warm-up that lowers the stakes, 2-3 core story arcs that carry the real signal, a specifics pass for details you can't extrapolate, and an escape hatch for time or curiosity overruns. Each part has a single job — don't blend them.

1. The Warm-Up (2-3 minutes)

The warm-up isn't small talk for its own sake — it's calibration. You're establishing that you want their actual experience, not a performance for your benefit, and you're getting a read on how much detail this particular person naturally gives.

• Goal: Build rapport and set expectations without asking anything analyzable yet.
• Sample prompt: "Before we get into specifics, tell me a bit about your role and how [the general problem area] shows up in a typical week."
• What to avoid: Don't ask about your product or feature yet. It primes the rest of the conversation toward your assumptions.

2. Core Story Arcs (20-30 minutes)

This is the heart of the guide — 2-3 arcs, each anchored to a real past event, not a hypothetical. Erika Hall, author of Just Enough Research, frames this as chasing specificity: a vague "how do you usually handle X" invites a generalized, unreliable answer, while "tell me about the last time X happened" forces recall of an actual event.

Pick 2-3 arcs relevant to your research question, not all four every time. Each arc should get roughly 8-10 minutes and end with a summarizing question like "did I get that right?" — which does double duty as a comprehension check and a trust-building signal.

3. Specifics Pass (5-10 minutes)

Once the story is on the table, fill gaps with pointed, closed-ended clarifiers: dates, tool names, frequency, who else was involved. This is the one place closed questions are appropriate — you're confirming facts, not exploring perspective.

• "How often does this happen — weekly, monthly?"
• "Who else touches this before it gets to you?"
• "What tool were you using at that point?"

4. The Escape Hatch

Always end your prompt list with an explicit permission-giving question: "Is there anything about this that we haven't touched on that you think I should know?" This single line consistently surfaces things no structured guide anticipates, because the customer — not you — decides what's relevant.

Build in a time-box override too: if an arc is producing rich material at minute 25 of a 30-minute call, let it run and cut the specifics pass short rather than the reverse. The guide should have a written note like "if arc 2 is gold, skip specifics questions 2-3." That's what makes it a guide and not a script — it explicitly tells you when to abandon it.

A Real Example: Interview Guide for Expense-Reporting Friction

Here's a worked guide for a PM investigating why employees delay submitting expense reports — concrete enough to adapt directly. Notice each section states a goal first, prompts second.

GOAL: Understand where expense reporting breaks down for individual contributors.
DURATION: 35 minutes.

WARM-UP (3 min)
Goal: Establish role context, lower stakes.
- "Tell me about your role and how often travel or purchases come up for you."

ARC 1 — Recent trigger (10 min)
Goal: Get a specific, recent instance of delay or friction.
- "Walk me through the last expense report you submitted — from purchase to submission."
- "What made you submit it when you did, rather than earlier or later?"
- Follow-up as needed: "What was going through your mind at that point?"

ARC 2 — Workaround (10 min)
Goal: Learn what they do to cope with the current process.
- "Is there anything you do to make this easier on yourself — a spreadsheet,
  a note, a reminder?"
- "What made you start doing that?"

SPECIFICS (7 min)
- "About how many reports do you submit a month?"
- "Do you use the mobile app, desktop, or both?"
- "Has anyone ever rejected or sent back one of your reports? What happened?"

ESCAPE HATCH (5 min)
- "Is there anything about expense reporting we haven't covered that you'd
  want the team building this to know?"

OVERRIDE NOTE: If Arc 1 surfaces a strong emotional reaction, spend the
remaining time there and skip specifics questions 2-3.

The 80/20 Rule: Why You Should Barely Talk

The single best predictor of interview quality is your own talk-time ratio: aim for the interviewer speaking 20% or less of the conversation, with the customer carrying the remaining 80%. Most first-time interviewers run closer to 50/50 without realizing it — every clarifying comment, every "that makes sense," every unsolicited example from your own experience eats into the customer's airtime and subtly steers their story toward yours.

Why the Ratio Matters More Than the Questions

Michael Margolis, who led Google Ventures' research practice and popularized the "5-second rule" for interview pacing, has described silence as the interviewer's most underused tool — count to five after an answer before speaking, because customers frequently fill that gap with the most valuable, unprompted detail of the whole conversation. Cutting in early to fill silence is the single most common way PMs unintentionally suppress signal.

How to Enforce the Ratio in Practice

1. Ask, then stop. Deliver the prompt and hold silence — resist the urge to add examples or clarify unless genuinely asked.
2. Echo instead of editorialize. Reflect back their last phrase ("...delayed a week?") rather than summarizing in your own words.
3. Save opinions for after the call. Any urge to validate, correct, or agree belongs in your notes, not the conversation.
4. Track it. Record a call once and estimate your own talk-time afterward — most PMs are shocked by the gap between perceived and actual ratio.

If you're building a discovery cadence around these interviews, pairing this ratio discipline with a weekly discovery habit of two interviews makes the skill compound instead of resetting every time you go a month between calls.

Grading Your Questions Before the Call, Not After

The most common guide-writing mistake is discovering, mid-interview, that a prompt you wrote as "open" is actually leading — a customer nods along to "don't you find X frustrating?" whether or not it's true. Catching this before the call is far cheaper than catching it in a transcript afterward.

How This Connects to Broader Discovery Practice

A single interview guide is only as useful as the discovery system it feeds into. If you're mapping how individual interviews should roll up into an ongoing practice, the complete guide to product discovery covers the cadence and synthesis layer this article's guide format feeds into. And once you have several interviews' worth of stories, running them through an opportunity solution tree for shipping teams is the natural next step for turning raw stories into prioritized bets.

Two adjacent frameworks are worth layering in as you refine your arcs. The jobs-to-be-done framework sharpens your "recent trigger" arc into a causal chain (push, pull, anxiety, habit) rather than a loose narrative. And mapping responses against a customer journey helps you spot which stage of the journey your arcs are systematically under-sampling.

Common Guide-Writing Mistakes to Avoid

Most weak guides fail in one of a handful of predictable ways — recognizing the pattern is faster than re-deriving the fix each time.

Key Takeaways

• Write goals and open prompts, not verbatim questions — a guide organized by intent survives a conversation that doesn't follow your order.
• Structure around 3-4 arcs: warm-up, 2-3 core story arcs anchored to real past events, a specifics pass, and an escape hatch.
• Always include a time-box override that explicitly permits skipping the specifics section when a story arc is producing strong signal.
• Hold your own talk-time to roughly 20% — silence after a question is often when the most valuable detail surfaces.
• Grade every question as open, leading, or closed before the call, not after, since leading phrasing is far cheaper to fix on your screen than in a transcript.
• Treat the guide as one input to a larger discovery system — individual interviews compound only when they roll into a consistent cadence and synthesis process.