Customer discovery interviews are the foundation of evidence-based product decisions for SaaS teams. The questions you ask determine whether you get polite agreement or genuine insight. This guide provides a practical question bank organized by research goal, with guidance on the adaptive follow-up techniques that separate productive interviews from scripted exercises.
Principles before questions
Before reaching for specific questions, internalize three principles that govern effective discovery.
Ask about behavior, not opinions. “How do you currently handle X?” produces actionable data. “What do you think about X?” produces speculation. Past behavior is the single best predictor of future behavior. Opinions predict almost nothing.
Follow the energy. When a participant’s tone shifts — they become more animated, more frustrated, or more detailed — that is where the real insight lives. Abandon your script momentarily and follow the thread. The most valuable finding in a discovery interview is almost never the answer to a question you prepared.
Pursue specificity. Every generalization hides useful detail. When a participant says “we usually do it this way,” ask about the last specific time. “Tell me about the most recent time that happened. When was it? What exactly did you do?” Specificity prevents participants from constructing an idealized narrative and forces recall of actual events.
Question bank by research goal
Problem discovery
Use these when you are exploring a problem space before committing to a solution direction.
- “Walk me through a typical day when you are working on [relevant task]. What tools do you use and in what order?”
- “What is the most tedious or frustrating part of [workflow area]? Can you give me a specific example from the last week?”
- “When was the last time [problem area] caused you to miss a deadline, waste time, or get frustrated? What happened?”
- “If you could eliminate one step from your current process, which would it be and why?”
- “What have you tried to solve this problem? What worked and what did not?”
The last question is particularly powerful. Users who have tried to solve a problem — building spreadsheets, writing scripts, hiring contractors — have already validated the problem’s intensity through their own investment. Users who have never attempted a solution may not care enough to adopt yours.
Solution validation
Use these after you have validated the problem and need to evaluate a specific solution concept.
- “I want to describe something we have been thinking about. [Describe concept in 2-3 sentences.] Based on what you just told me about your workflow, where would this fit in?”
- “What would need to be true about this for you to try it? What would make you hesitate?”
- “If this existed today, what would you stop doing? What tools or processes would it replace?”
- “Who else on your team would need to be involved for this to work? What would their concerns be?”
- “On a scale of your current workaround to a perfect solution, where does this concept land? What is missing?”
The commitment probes — asking what they would stop doing, who else would need to be involved — separate genuine interest from polite enthusiasm. A user who cannot describe what they would change is expressing a preference, not demonstrating intent.
Competitive landscape
Use these to understand how users evaluate and compare solutions.
- “What tools have you evaluated in this space? What stood out about each one?”
- “If you had to switch away from [current tool] tomorrow, what would you move to and why?”
- “What does [current tool] do well that you would not want to lose? What frustrates you most about it?”
- “When you last evaluated a new tool in this category, what were the top three criteria? How did you weight them?”
Workflow and context
Use these to build a complete picture of the user’s environment before proposing solutions.
- “Walk me through how a [relevant deliverable] gets from initial request to final delivery. Who is involved at each step?”
- “Where do things typically break down or slow down in this process?”
- “How do you currently measure success for [relevant outcome]? What metrics or signals do you track?”
- “What has changed about how you do this in the last 12 months? What drove those changes?”
The art of follow-up
Prepared questions get you to the surface. Follow-up gets you to insight. The 5-7 level laddering technique works by treating each answer as a prompt for deeper exploration.
Level 1 probe: Clarify. “When you say ‘it takes too long,’ what does that mean specifically? How long is too long?”
Level 2 probe: Contextualize. “What are you trying to accomplish during that time? What depends on this being faster?”
Level 3 probe: Quantify. “How often does this happen? Is it every day, every week, or less frequent?”
Level 4 probe: Explore consequences. “What happens when this delay occurs? Who is affected beyond you?”
Level 5 probe: Uncover the root. “If you could redesign this from scratch, knowing what you know now, what would it look like?”
Each level moves from description to diagnosis to prescription. By level 5, you are no longer discussing the original symptom — you are discussing the underlying need that the symptom revealed.
Scaling discovery without losing depth
The traditional constraint on discovery research is interviewer bandwidth. A skilled researcher can conduct 3-4 interviews per day before quality degrades. At that rate, a 20-person study takes a full week just for data collection, not counting recruitment and analysis.
AI-moderated interviews remove this bottleneck. Running dozens of conversations simultaneously, each with adaptive 5-7 level follow-up, produces the same depth as human-moderated interviews at a fraction of the time. A 20-person discovery study that would take 2-3 weeks with traditional methods completes in 48-72 hours, with every conversation recorded, transcribed, and searchable.
For SaaS teams operating in two-week sprints, this timeline means discovery research can inform the current sprint’s decisions rather than the sprint after next. Product managers get evidence while the decision window is still open.
The cost structure supports continuous practice rather than episodic studies. At $20 per interview, running 10 discovery conversations per week costs $800 per month — roughly the cost of a single team lunch. That steady cadence of customer contact keeps the team’s understanding of user needs current as the product and market evolve.
Building an interview practice for UX research
Discovery interviews produce the most value when they are a habit rather than an event. Teams that run 5-10 interviews per week develop an intuitive understanding of their user base that no amount of analytics or survey data can replicate. They recognize patterns faster, spot emerging needs earlier, and make roadmap decisions with confidence.
The key infrastructure is a searchable repository where conversation insights accumulate over time. When a PM asks “what do our users think about reporting?” they should be able to search across six months of discovery conversations and find every relevant mention — traced to specific verbatim quotes, not filtered through someone’s memory or a stale research report. This institutional memory survives team transitions and strategy shifts, compounding in value with every conversation added.