How to Write a Discussion Guide for User Interviews

Learn how to write a user interview discussion guide that keeps sessions focused without over-scripting. Covers structure, question types, probing hierarchy

What Is a Discussion Guide (and What It Isn’t)
Before You Write a Single Question: Ground the Guide in Objectives
The Five-Part Structure of an Effective Discussion Guide
Writing Questions That Open Doors: Question Types and Phrasing
Building a Probing Hierarchy So You Never Lose a Thread
Formatting the Guide So It Works in the Room
Piloting the Guide Before Fieldwork Begins
Common Mistakes and How to Avoid Them
Frequently Asked Questions

What Is a Discussion Guide (and What It Isn’t)

A discussion guide is a structured facilitator reference — a scaffold for conversation, not a script to be read aloud. If you’ve ever watched an interviewer lose rapport because they were squinting at a screen trying to find their next question, you’ve seen what happens when a guide becomes a crutch instead of a tool.

The distinction matters. A screener filters participants. A survey captures responses at scale. A discussion guide does something different: it holds the shape of a conversation so that every session covers comparable ground while still leaving room for the unexpected. The facilitator stays present and curious; the guide handles the architecture.

Why bother with the artefact at all? Three reasons. Consistency — when multiple researchers run sessions, a shared guide means you’re studying the same thing. Alignment — mapping guide sections to research objectives keeps you honest about what you’re actually trying to learn. And reduced cognitive load — a facilitator who doesn’t have to remember what comes next can spend that mental energy listening.

The trap is over-engineering it. A rigid script produces interview data that looks tidy and tells you almost nothing, because participants sense they’re being processed rather than heard. A good guide preserves spontaneity while keeping sessions comparable.

On sequence: the guide comes after your user research plan template is agreed and before fieldwork begins. It translates research objectives into live conversation. Read our overview of how to conduct user interviews alongside this if you’re building your process from scratch, and if you’re unsure whether interviews are the right method at all, start with usability testing vs user interviews.

Before You Write a Single Question: Ground the Guide in Objectives

The most common guide-writing mistake is starting with questions. Start instead with your research questions — the ones documented in your research plan — and work forwards from there.

Each section of your guide should map to a specific learning objective. If you can’t name which objective a question serves, it probably doesn’t belong. This single discipline prevents the scope creep that turns a 60-minute interview into a breathless sprint through 25 questions, none of them explored properly.

Before drafting, make two lists. The first: assumptions you’re testing, where you have a hypothesis and need evidence to confirm or challenge it. The second: open territory, where you genuinely don’t know what participants will say and you want to follow their lead. These require different question types and different amounts of probe depth.

Then have an explicit conversation with your stakeholders about must-cover topics versus nice-to-have topics. Stakeholders will always want more than a 60-minute session can hold. Making the trade-off explicit before writing means you’re not discovering it mid-session when you realise you’ve spent 40 minutes on the first two topics.

Time-box the session and work backwards. A 60-minute interview realistically accommodates 3–4 substantive topic sections once you account for opening, warm-up, and closing. If your must-cover list has eight topics, you need either a longer session or a narrower scope — not a faster facilitator.

The Five-Part Structure of an Effective Discussion Guide

A reliable guide for a 60-minute session has five parts. Each has a job to do.

1. Opening (5 minutes). Welcome the participant, explain the purpose in plain terms, confirm consent, request recording permission, and set expectations explicitly: there are no right or wrong answers; you’re here to understand their experience, not to test them. This section is often written out in more detail than the rest — not to be read verbatim, but because facilitators, particularly less experienced ones, benefit from having the exact framing available under pressure.

2. Warm-up questions (5–10 minutes). Broad, easy, about recent behaviour. “Tell me a bit about your role and how your day typically starts” does more than break the ice — it gives you context that will shape how you interpret everything that follows. Warm-up questions should never feel like the interview has begun in earnest. Participants need to arrive at the conversation before you take them anywhere difficult.

3. Core topic sections (35–40 minutes). This is where the substantive learning happens. Organise topics from general to specific — the funnel principle. Begin with the broader context of the problem space before narrowing to the specific behaviour or decision you care about. Jumping straight to your most pointed question before the participant has oriented themselves is a reliable way to get shallow answers.

4. Concept or stimulus exploration (if applicable, 5–10 minutes). If you’re showing a prototype, a concept, or any stimulus material, keep it near the end. Showing product material early anchors participants to your framing and contaminates the unprompted responses you need from earlier sections.

5. Closing (5 minutes). Ask whether there’s anything the participant wanted to say that you didn’t get to. This frequently surfaces the thing they consider most important — which isn’t always what you thought you were studying. Explain what happens next, thank them, and confirm any incentive logistics.

The order is not arbitrary. Every structural decision serves the goal of getting uncontaminated, reflective responses from someone who feels safe enough to be honest.

Writing Questions That Open Doors: Question Types and Phrasing

Open, behavioural questions are the backbone of a useful interview. “Tell me about the last time you had to [do X]” consistently outperforms hypothetical questions because it anchors the response in actual experience rather than imagined behaviour. People are reliably poor at predicting what they would do; they’re much better at recounting what they did.

The past-behaviour question works for a specific reason: memory of concrete events carries detail — the workaround someone improvised, the moment they gave up, the person they called for help. Hypothetical questions invite the participant to describe their ideal self, which is rarely the person you need to understand.

Two question types to eliminate from your guide:

Leading questions embed an assumption. “How frustrating did you find the checkout process?” tells the participant you expect frustration. Replace it with: “Walk me through what happened when you got to checkout.”

Double-barrelled questions ask two things at once. “Did you find it easy to use, and did you complete the task?” forces the participant to choose which part to answer, or to conflate two separate assessments. Ask one thing at a time.

A rewrite example:

Weak: “Would you say the onboarding process was confusing and off-putting?”
Strong: “Tell me about the first time you set up your account — what did you do, and what happened?”

Jargon is a quieter version of the same problem. When your question uses product-team vocabulary that participants don’t share, you’re asking them to translate before they can answer. Match your language to theirs, not to the product roadmap.

Primary questions are the guide’s backbone. Beneath each one, you’ll write follow-up prompts — pre-written nudges for when a response needs more depth. Below that, in the room, you’ll use silent probes: a pause, a nod, a minimal encourager like “mm-hmm.” The guide covers the first two tiers; the third is facilitator skill.

Building a Probing Hierarchy So You Never Lose a Thread

Every primary question in your guide should have 2–4 pre-written probes beneath it. This is the part most researchers underinvest in, and it shows: sessions where the facilitator has no pre-written probes tend to stall at the surface.

Four probe types worth knowing:

Elaboration: “Can you say more about that?”
Example: “Can you give me a recent example?”
Clarification: “When you say [X], what do you mean by that?”
Consequence: “What happened next?”

Format them with clear visual indentation beneath their parent question. The facilitator should be able to glance at the probe without losing eye contact with the participant. A guide that requires careful reading mid-conversation is a guide that damages rapport.

The distinction between planned probes and spontaneous probes matters. Planned probes live in the guide; you write them during preparation. Spontaneous probes are in-session judgements — the follow-up you invent because something unexpected just emerged. Good facilitation needs both, but the guide can only train the former.

One failure mode to watch for: over-probing on comfortable, expected answers while moving quickly past surprising or uncomfortable ones. The response that makes you think “that’s odd” is usually the one most worth exploring. Pre-written probes won’t catch that on their own — but having a probe habit means you’re more likely to pause and dig when something unexpected surfaces.

We’ve seen this play out in practice. On a study exploring how people managed complex information-tracking tasks, a primary question about daily workflow was yielding predictable answers about tool preferences. A pre-written probe asking about workarounds — specifically, “Is there anything you do outside the official process to make this work?” — surfaced a pattern that hadn’t appeared in the primary responses at all. Participants had built parallel systems in spreadsheets because the official workflow didn’t account for a common exception case. That finding reshaped the product direction. The primary question alone wouldn’t have reached it.

Formatting the Guide So It Works in the Room

A guide that’s hard to scan mid-session is a guide that gets abandoned. Format for live use, not for the document review.

Use a two-column or clearly indented layout. Primary questions sit prominently on the left or at the top level; probes are indented or in a subordinate column. The facilitator’s eye should be able to move between participant and guide without effort.

Add time cues next to each section — not to enforce a rigid schedule, but to give the facilitator a reference point for self-regulation. “If I’m here at 35 minutes, I’m on track” is more useful than a vague sense that things are running long.

Bold or highlight the single most important question in each section. When a session runs short on time, the facilitator needs to know instantly what cannot be skipped.

Leave a margin or column for in-session annotations. Noting where a participant paused, where something unexpected came up, or where a probe worked particularly well takes seconds and pays back considerably during analysis. Read more about that stage in our post on how to analyze user interview data.

Keep the guide to one or two pages of dense content. A longer guide is usually a planning problem — too many topics, not enough prioritisation. Researchers who carry a six-page guide into a session typically use the first four pages and rush the last two.

Version-control the document. Date it, number it. When multiple researchers are running sessions in parallel, a version mismatch means you’re running different studies.

Piloting the Guide Before Fieldwork Begins

Run at least one internal pilot before the first real session. Sit a colleague down, ask them to play a participant, and run the guide as you would live. This catches problems that aren’t visible on the page.

Check three things specifically. Does the warm-up actually feel warm, or does it feel like the interview starting abruptly? Are any questions confusing when spoken aloud — things that read fine but land oddly? Does the session fit the allocated time when you account for natural conversation pace?

Questions that consistently produce one-word answers during the pilot need either a built-in probe or a rewrite. One-word answers are the guide telling you something.

Amend after the pilot. If you had to skip a section to stay on time, that section either needs to be cut or the session needs to be longer — not faster.

Once fieldwork has started, a light-touch update after sessions one or two is acceptable. You might clarify a question that confused the first participant, or add a probe for something that came up unexpectedly. Freeze the guide once you’re past the first couple of sessions. Mid-study changes compromise comparability and make synthesis harder.

Common Mistakes and How to Avoid Them

Too many questions. Five rich topics explored properly will teach you more than fifteen questions skimmed. Depth over breadth, always.

Asking “why” directly. “Why did you do that?” triggers rationalisation — participants construct a plausible-sounding explanation rather than reporting what actually happened. Replace with “What led you to that?” or “Walk me through how you got there.”

Jumping to solutions before understanding the problem. Questions about what participants want from a product, or what features they’d find useful, belong near the end if they appear at all. The first half of an interview should be entirely about their experience and context, not your product.

Reading the guide verbatim. A facilitator who reads is not listening. The guide is a reference; the conversation is the job.

Omitting consent and recording logistics. These belong in the opening section without exception. Discovering mid-session that you forgot to request recording permission is a problem that a well-structured guide prevents.

Skipping the closing. The final two minutes — asking whether there’s anything the participant wants to add — consistently surfaces the thing they consider most important. It’s the section most often cut when sessions run long, and often the section that costs the most to lose.

Good analysis depends on good data, and good data depends on a guide that was written with care. Once you’ve run your sessions, our posts on writing a UX research report that drives action and thematic analysis in qualitative research cover what comes next.

Frequently Asked Questions

How long should a discussion guide be?

For a 60-minute interview, aim for 5–7 primary questions across 3–4 topic sections, plus opening and closing. In page terms, that’s typically one to two pages. A longer guide is a planning problem, not a sign of thoroughness. If your guide runs to four pages, the real issue is that you haven’t prioritised your objectives clearly enough.

What’s the difference between a discussion guide and an interview script?

A script is read verbatim, constraining the conversation to pre-set wording and a fixed order. A discussion guide provides topic areas, primary questions, and probes that the facilitator adapts in real time. The goal is cross-session consistency on the topics covered, not word-for-word replication of how questions were asked. A guide preserves the flexibility to follow an unexpected thread; a script closes it off.

About Glasgow Research — Glasgow Research helps B2B SaaS teams turn customer and market research into product decisions. Work with us.