{"site":{"name":"Koji","description":"AI-native customer research platform that helps teams conduct, analyze, and synthesize customer interviews at scale.","url":"https://www.koji.so","contentTypes":["blog","documentation"],"lastUpdated":"2026-05-25T13:02:10.731Z"},"content":[{"type":"documentation","id":"52428ba9-0d65-417c-8dd9-9d3310416073","slug":"discussion-guide-template","title":"Discussion Guide Template: How to Structure Your Research Sessions","url":"https://www.koji.so/docs/discussion-guide-template","summary":"A discussion guide is a structured document outlining objectives, question flow, probing strategies, and ground rules for research interviews. Effective guides follow a five-part structure: setup, introduction/warm-up, core open-ended questions with probing follow-ups, structured quantitative questions (scale, choice, ranking, yes/no), and a closing section. Key principles: ask about past behavior not hypotheticals, one question at a time, neutral language, and built-in buffer time. AI platforms like Koji replace static guides with adaptive systems — the AI probes on unexpected threads, extracts structured values automatically, and runs consistent sessions at scale without scheduling. Koji supports 6 question types: open_ended, scale, single_choice, multiple_choice, ranking, and yes_no.","content":"# Discussion Guide Template: How to Structure Your Research Sessions\n\nA discussion guide is the backbone of any user research session. It's the document that gives your interview structure, keeps you focused on research objectives, and ensures you cover the territory that matters — without turning the conversation into a rigid interrogation.\n\nThis guide covers everything you need to create effective discussion guides, from structure and question types to how modern AI-powered research tools like Koji replace static scripts with adaptive, intelligent interviews that probe more deeply and consistently than any human moderator.\n\n## What Is a Discussion Guide?\n\nA discussion guide (also called an interview guide or moderator guide) is a structured document that outlines:\n- The objectives of a research session\n- The flow and timing of the conversation\n- The specific questions and topics to explore\n- Probing strategies for going deeper\n- Ground rules for the interviewer\n\nIt's distinct from a survey or questionnaire — a discussion guide is designed for conversation, not form completion. Questions are open-ended by default, with structure to guide rather than constrain.\n\n**The key tension in every discussion guide:** structure enough to cover your research goals, flexible enough to follow interesting threads. The best guides feel like a thoughtful conversation, not a checklist.\n\n## Discussion Guide vs. Interview Guide vs. Moderator Guide\n\nThese terms are often used interchangeably, with minor distinctions:\n\n- **Discussion guide** — most common term; implies conversational flexibility\n- **Interview guide** — slightly more formal; common in academic and B2B research\n- **Moderator guide** — emphasizes the facilitator role; common in focus group research\n- **Screener + guide** — combined document that includes participant qualification criteria\n\nFor the purposes of this article, \"discussion guide\" refers to the document used by a researcher (human or AI) to conduct one-on-one research sessions.\n\n## The Standard Discussion Guide Structure\n\nMost effective discussion guides follow this five-part structure:\n\n### 1. Header / Session Setup (1–2 minutes)\nBefore the recording starts or the interview begins, this section covers logistics:\n\n```\nStudy: [Study Name]\nDate: [Date]\nParticipant ID: [ID]\nInterviewer: [Name]\nSession Duration: [e.g., 45 minutes]\nRecording consent: [ ] Obtained\n```\n\n### 2. Introduction and Warm-Up (3–5 minutes)\nPurpose: build rapport, set expectations, get consent.\n\nTemplate:\n- Welcome the participant and thank them for their time\n- Explain the purpose: \"We're trying to understand [general topic] — there are no right or wrong answers\"\n- Get recording consent (if not already obtained)\n- Warm-up question: \"Can you tell me a bit about your role and how you [relevant activity]?\"\n\n**Why warm-up questions matter:** Participants who feel comfortable early provide more candid, detailed answers. Don't skip this section.\n\n### 3. Core Questions (25–35 minutes)\nThe main body of the discussion guide. This is where your research objectives get addressed. Structure this as 3–5 topic areas, each with a primary question and follow-up probes.\n\n**Example format:**\n\n```\nTOPIC 1: [Area of Focus]\nPrimary question: [Open-ended question about the topic]\nProbes:\n  - Tell me more about that...\n  - What happened next?\n  - Why was that important to you?\n  - How often does this happen?\n  - What did you do to solve it?\n```\n\n### 4. Structured Questions (5–10 minutes)\nMany research sessions benefit from a mix of open-ended exploration and structured quantitative questions. These produce data that can be aggregated across participants for reporting:\n\n- **Scale questions** — \"On a scale of 1–10, how [satisfied/frustrated/confident] are you with...?\"\n- **Single choice** — \"Which of these best describes your experience: A, B, or C?\"\n- **Multiple choice** — \"Which of these challenges have you encountered? Select all that apply\"\n- **Yes/No** — \"Have you ever [done X]?\"\n- **Ranking** — \"Rank these features by importance to you\"\n\nIn traditional research, structured questions are awkward to ask mid-conversation. AI interview platforms like Koji handle this seamlessly — in text mode, the AI automatically presents interactive widgets (sliders, radio buttons, checkboxes) for quantitative questions while continuing the qualitative conversation. In voice mode, the AI asks the question conversationally and extracts the structured value automatically.\n\nKoji supports 6 structured question types: **open_ended**, **scale**, **single_choice**, **multiple_choice**, **ranking**, and **yes_no** — each producing clean, aggregatable data across all your participants. Learn more in our [Structured Questions in AI Interviews](/docs/structured-questions-guide) guide.\n\n### 5. Closing (3–5 minutes)\nPurpose: capture any final insights and leave the participant feeling valued.\n\nTemplate:\n- \"Is there anything we haven't talked about that you think is important?\"\n- \"If you could change one thing about [topic], what would it be?\"\n- \"Who else do you think we should be talking to?\"\n- Thank them, explain next steps, confirm incentive delivery\n\n## Full Discussion Guide Template\n\nHere's a complete template you can adapt for your own research:\n\n---\n\n**[Study Name] — Discussion Guide**  \n*Research objective: [What decision will this research inform?]*\n\n**SECTION 1: SETUP**  \nDuration: 45 minutes | Participant: [ID] | Date: [Date]\n\n**SECTION 2: INTRODUCTION** (5 min)  \n\"Thanks for joining me today. I'm [name] from [company]. We're doing research on [topic] — this isn't a test, and there are no right or wrong answers. I'd love to just hear about your experience. Is it okay if I record this session?\"\n\nWarm-up: \"Can you start by telling me about your role and how you typically [activity related to research topic]?\"\n\n**SECTION 3: CORE QUESTIONS** (30 min)\n\n*Topic 1: [Current Experience]*  \nQ: \"Walk me through how you currently handle [task/situation].\"  \nProbes: \"What does that look like in practice?\" / \"What's the most frustrating part?\" / \"How often do you run into this?\"\n\n*Topic 2: [Pain Points / Challenges]*  \nQ: \"What's the hardest part about [area]?\"  \nProbes: \"Tell me about a specific time when...\" / \"What have you tried to solve it?\" / \"What didn't work?\"\n\n*Topic 3: [Goals / Motivations]*  \nQ: \"What does success look like for you when it comes to [topic]?\"  \nProbes: \"How do you measure that?\" / \"How would you know if it improved?\"\n\n*Topic 4: [Current Solutions]*  \nQ: \"What tools or approaches are you using now?\"  \nProbes: \"Why did you choose that?\" / \"What do you wish it did differently?\"\n\n**SECTION 4: STRUCTURED QUESTIONS** (5 min)  \n- \"On a scale of 1–5, how [satisfied] are you with [current approach]?\" [scale]\n- \"Which of these describes your biggest challenge: A / B / C / D?\" [single_choice]\n- \"How often do you [activity]?\" [single_choice: daily / weekly / monthly / rarely]\n\n**SECTION 5: CLOSING** (5 min)  \n\"Is there anything we haven't covered that you think is relevant?\"  \n\"If you could wave a magic wand and change one thing about [topic], what would it be?\"  \n\"Who else would be valuable for us to talk to?\"  \n\"Thank you so much — [explain incentive delivery and timeline]\"\n\n---\n\n## How to Write Great Discussion Guide Questions\n\n### The 5 Principles of Good Discussion Guide Questions\n\n**1. Ask about the past, not hypotheticals**  \nBad: \"Would you use a feature that...?\"  \nGood: \"Tell me about the last time you tried to [do this thing].\"\n\nPast behavior is far more reliable than future predictions. This is the core insight of the Mom Test approach — see our [Mom Test Methodology](/docs/mom-test-methodology) guide.\n\n**2. Open-ended by default**  \nBad: \"Do you find [X] frustrating?\"  \nGood: \"What's your experience been with [X]?\"\n\nYes/no questions shut down conversation. Open-ended questions invite stories.\n\n**3. One question at a time**  \nBad: \"What do you use today, and how long have you been using it, and what do you like about it?\"  \nGood: \"What do you use today for [task]?\" [then probe from there]\n\nMulti-part questions confuse participants and produce incomplete answers.\n\n**4. Neutral language**  \nBad: \"How much does [problem] frustrate you?\"  \nGood: \"How does [problem] affect you?\"\n\nLeading questions prime participants to confirm your hypothesis. Neutral language surfaces genuine experiences.\n\n**5. Specific, not general**  \nBad: \"Tell me about your [area] experience.\"  \nGood: \"Walk me through the last time you [specific task].\"\n\nSpecificity produces stories. Stories produce insights.\n\n## Timing Your Discussion Guide\n\nA common mistake is creating a guide with 45 minutes of questions for a 30-minute session. Build in buffer:\n\n| Session Length | Core Question Time | Number of Topics |\n|---|---|---|\n| 20 minutes | 10 minutes | 2 topics |\n| 30 minutes | 18 minutes | 3 topics |\n| 45 minutes | 28 minutes | 4 topics |\n| 60 minutes | 38 minutes | 5 topics |\n\nIf you're consistently running out of time, cut topics — not probing depth. Fewer topics explored deeply beats many topics touched superficially.\n\n## How AI Replaces the Static Discussion Guide\n\nA well-written discussion guide is essential — but human moderators using static guides have limitations:\n\n- **Inconsistency** — different interviewers ask questions differently, making cross-participant analysis harder\n- **Missed probes** — tired moderators forget to probe on interesting threads\n- **Limited scale** — a single moderator can run 5–8 sessions per week at most\n- **Scheduling friction** — 72+ hours of logistics per session\n\nKoji's AI interviewer replaces the static discussion guide with a dynamic, adaptive system. You define your research brief (objectives, methodology, key questions, guardrails) and the AI conducts every interview:\n\n- Consistently asks every question and follows up with intelligent probes\n- Adapts the conversation based on participant answers — if someone mentions an unexpected pain point, the AI follows the thread\n- Runs voice or text sessions 24/7 without scheduling\n- Extracts structured values from quantitative questions automatically\n- Generates individual analysis and aggregate themes as sessions complete\n\nFor researchers who still want to run live sessions, Koji's brief structure mirrors a well-organized discussion guide: problem context, target participant, methodology, key questions (typed as open_ended, scale, single_choice, multiple_choice, ranking, or yes_no), topics to explore, and guardrails. You can export the brief as your human-moderated guide and use the same study structure for AI-moderated sessions with the rest of your participants.\n\n## Discussion Guide Checklist\n\nBefore your first session, verify:\n\n- [ ] Objectives are clearly stated at the top\n- [ ] Introduction includes rapport-building and consent\n- [ ] Core questions are open-ended and about past behavior\n- [ ] Each topic has 3–5 probing questions ready\n- [ ] Structured questions are placed logically (not as the opening)\n- [ ] Closing section captures \"anything else\" and leaves on a positive note\n- [ ] Total question count fits within session time (with 20% buffer)\n- [ ] Language is neutral — no leading questions\n- [ ] Recording consent process is documented\n\n## Common Discussion Guide Mistakes\n\n**Treating it as a rigid script** — The guide is a map, not a script. If a participant mentions something unexpectedly valuable, follow it. You can always return to your guide.\n\n**Too many questions** — Researchers often add questions \"just in case.\" This creates rushed sessions where nothing gets explored deeply. 5 well-probed topics beats 15 surface-level questions.\n\n**No probing questions** — Primary questions open the conversation; probes are where the insights live. For every primary question, have 3–5 probing follow-ups ready.\n\n**Starting with sensitive topics** — Build rapport first. Don't jump into the hardest questions before participants are comfortable.\n\n**Forgetting the close** — The closing \"is there anything else?\" question frequently surfaces the most candid insights. Participants often save their real opinion for the end.\n\n## Templates for Specific Research Types\n\n### Usability Research Discussion Guide\nFocus on task completion and friction points. Open with \"walk me through how you'd [task]\" and observe before asking questions.\n\n### Customer Discovery Discussion Guide\nFocus on current behavior, workarounds, and alternatives. The Mom Test question patterns are ideal here.\n\n### Feature Validation Discussion Guide\nFocus on whether the proposed feature maps to real behavior. Show concept, ask \"walk me through how you'd use this,\" and probe on gaps.\n\n### Exit / Churn Discussion Guide\nFocus on the moment of decision, alternatives considered, and what would have changed the outcome. Sensitive — build maximum rapport before going deep.\n\n### NPS Follow-Up Discussion Guide\nStart with the score as the anchor: \"You gave us a [score] — can you walk me through what's behind that?\" Then probe for specifics.\n\n## Related Resources\n\n- [Structured Questions in AI Interviews](/docs/structured-questions-guide)\n- [User Interview Guide Template: How to Plan, Run, and Analyze Interviews](/docs/user-interview-guide-template)\n- [Open-Ended Interview Questions: 100+ Examples and How to Use Them](/docs/open-ended-interview-questions)\n- [Probing and Follow-Up Questions: Going Deeper in Research Interviews](/docs/probing-and-follow-up-questions)\n- [The Mom Test: How to Talk to Customers Without Being Misled](/docs/mom-test-methodology)\n- [Semi-Structured Interviews: The Complete Guide](/docs/semi-structured-interview-guide)\n\n\n## Further reading on the blog\n\n- [How to Run AI-Powered Customer Interviews at Scale](/blog/how-to-run-ai-powered-customer-interviews-at-scale) — Learn how to conduct effective customer interviews at scale using AI. This comprehensive guide covers everything from planning and question \n\n<!-- further-reading:blog -->\n","category":"Interview Techniques","lastModified":"2026-05-16T03:27:38.00446+00:00","metaTitle":"Discussion Guide Template for User Research | Koji Docs","metaDescription":"Create a research discussion guide that keeps interviews focused and uncovers real insights. Includes full template, question types, timing guide, and how AI-powered tools adapt the guide dynamically.","keywords":["discussion guide template","interview guide","moderator guide","research questions","user interview structure"],"aiSummary":"A discussion guide is a structured document outlining objectives, question flow, probing strategies, and ground rules for research interviews. Effective guides follow a five-part structure: setup, introduction/warm-up, core open-ended questions with probing follow-ups, structured quantitative questions (scale, choice, ranking, yes/no), and a closing section. Key principles: ask about past behavior not hypotheticals, one question at a time, neutral language, and built-in buffer time. AI platforms like Koji replace static guides with adaptive systems — the AI probes on unexpected threads, extracts structured values automatically, and runs consistent sessions at scale without scheduling. Koji supports 6 question types: open_ended, scale, single_choice, multiple_choice, ranking, and yes_no.","aiDifficulty":"beginner","aiEstimatedTime":"12 minutes"}],"pagination":{"total":1,"returned":1,"offset":0}}