The best customer discovery interviews are not conversations about your product — they are conversations about the customer's life, their problems, and their current solutions, conducted without mentioning your product at all until you have a clear picture of the problem landscape.
Customer discovery is the most important and most commonly misperformed research activity in product management. The failure mode is not that teams do not talk to customers — most do. The failure mode is that they talk to customers about their product instead of about the customer's problem, producing feedback that confirms existing assumptions rather than challenging them.
Why Customer Discovery Differs from Customer Validation
Customer discovery and customer validation are sequential, distinct activities:
| Activity | Goal | Timing | What You Bring | |----------|------|--------|----------------| | Discovery | Understand the problem space | Before building | Nothing — no product, no mockup | | Validation | Validate a proposed solution | After problem definition | Mockup, prototype, or concept |
Running validation interviews before completing discovery is the most expensive research mistake a PM can make. You will get feedback on your solution before confirming the problem is real, severe, and worth solving.
Best Practice 1: Recruit for Problem Experience, Not Demographics
The most common recruiting mistake is targeting people who match your demographic ICP rather than people who have experienced the specific problem you are investigating.
Wrong recruiting criteria: VP of Marketing at a 100-500 person SaaS company.
Right recruiting criteria: VP of Marketing at a 100-500 person SaaS company who has personally tried to improve their attribution reporting in the last 6 months.
The second criteria ensures you are talking to people who have actively engaged with the problem — not people who theoretically might encounter it.
H3: The Screener Survey
Use a 3–5 question screener before scheduling interviews:
Example screener for an attribution tool:
- How often does your team produce channel attribution reports? (Filter: at least monthly)
- Who is primarily responsible for attribution reporting at your company? (Filter: the person you want — not a subordinate)
- In the last 6 months, have you experienced a situation where you were unsatisfied with your attribution data? (Filter: yes)
- Would you be willing to share your screen during a 45-minute call? (Filter: yes — required for context observation)
This screener takes 2 minutes to complete and eliminates 60–70% of unqualified respondents before you invest 45 minutes in an interview.
Best Practice 2: Open with Context, Not Questions
The first 5 minutes of a discovery interview are the most important. Do not start with a prepared question list. Start by establishing shared context:
Hello, thanks for joining. I want to be upfront about what we're doing today — I'm not here to pitch you anything or show you a product. I'm here to understand how [problem area] works at your company. I'm going to ask you to walk me through some specific situations. There are no right or wrong answers — I'm trying to learn, not test you.
This framing does three things: sets expectations (no pitch coming), establishes safety (no right answer), and signals respect for their expertise.
Best Practice 3: Use the TEDW Question Framework
Discovery questions should follow the TEDW framework: Tell, Explain, Describe, Walk-me-through. These question types elicit narrative rather than opinion.
Tell: Tell me about the last time you needed to [do the job]. Explain: Explain what your current process looks like for [job]. Describe: Describe what a good week looks like vs. a bad week for [job area]. Walk-me-through: Walk me through what happened step by step.
These questions produce specific, past-behavior evidence — the highest-quality signal in discovery research.
Avoid:
- Would you (hypothetical preference — unreliable)
- Do you think (opinion — less useful than behavior)
- How important is X to you (leads to everything being important)
- Have you ever (binary — follow with Tell me about a specific time)
Best Practice 4: Listen for the Three Signal Types
During discovery interviews, filter what you hear into three categories:
| Signal Type | Example | What It Tells You | |------------|---------|-------------------| | Pain signal | I spend 3 hours every Monday rebuilding that report | Problem is real and frequent | | Workaround signal | We export to Excel and do it manually | Current solution is inadequate | | Gravity signal | My CMO asks me for this every week | Problem has organizational importance |
Workaround signals are the most reliable indicators that a problem is worth solving: if someone is doing something painful enough to warrant a workaround, they are motivated to pay for a better solution.
According to Lenny Rachitsky's writing on customer discovery, the best discovery signal is not a customer expressing pain — it is a customer describing a workaround they built themselves. That workaround is a product waiting to be built.
Best Practice 5: Probe Assumptions Aggressively
Every statement a customer makes contains an assumption. Probe it:
Customer: We really need better reporting.
PM: When you say better reporting — what's not working about what you have today? Tell me about a specific time you felt limited by your current reports.
Customer: The attribution model is too simple.
PM: Walk me through what happened when the attribution model gave you information that didn't match your intuition. What did you do next?
This technique — asking for a specific instance after every generalization — converts opinion into evidence.
Best Practice 6: Conduct Shadow Sessions
For B2B products, combine interviews with screen-sharing observation sessions where the customer performs their actual workflow live.
Ask: Would you be willing to share your screen and show me how you currently [do the job] while you walk me through what you're doing?
Watching the actual workflow reveals:
- Tools they use that they did not mention in the interview
- Workarounds invisible to the user because they are so habitual
- Pain points they accept as normal that they would not have thought to mention
Shadow sessions produce 3–5x more actionable insights per hour than pure question-and-answer interviews.
Synthesis: From Interviews to Insights
After 8–12 discovery interviews, synthesize using a structured approach:
H3: Opportunity Statement Framework
For each significant pain signal, write an opportunity statement:
[User type] needs a way to [do job] because [current situation fails them] which leads to [negative outcome].
Example: Marketing teams at 200-500 person SaaS companies need a way to understand channel contribution to pipeline without rebuilding attribution models weekly because their current tools only provide last-touch attribution, which leads to misallocated budget and channel-level conflict between marketing and sales.
This statement is your product opportunity — not a solution, a problem worth solving.
FAQ
Q: What are customer discovery interviews? A: Structured qualitative research conversations where a PM explores a specific problem domain with target customers before building any solution — the goal is to understand whether a problem is real, severe, and worth solving, not to validate a specific product concept.
Q: How many customer discovery interviews should you conduct? A: 8–12 for initial discovery. You typically reach saturation — hearing the same themes repeatedly — by interview 7–9. Run additional rounds if you expand to new customer segments.
Q: What questions should you avoid in customer discovery interviews? A: Hypothetical preference questions (would you use X?), leading questions, binary yes/no questions, and questions about future behavior. Focus on past behavior and specific instances.
Q: What is the difference between customer discovery and user testing? A: Discovery explores whether a problem is worth solving and how customers currently experience it. User testing evaluates whether a proposed solution is usable and effective. Discovery always precedes user testing.
Q: What are the most reliable signals in customer discovery interviews? A: Workaround signals — when customers describe a manual or imperfect process they built themselves to solve a problem. If they invested time building a workaround, they are motivated to pay for a better solution.
HowTo: Conduct Customer Discovery Interviews for Product Validation
- Write recruiting criteria that screen for problem experience, not demographics — use a 3 to 5 question screener to qualify participants before scheduling
- Open each interview by establishing that there is no pitch coming and no right answers — create psychological safety for honest responses
- Use TEDW question framework (Tell, Explain, Describe, Walk-me-through) to elicit specific past-behavior narratives rather than opinions
- Filter what you hear into three signal types: pain signals, workaround signals, and gravity signals — workaround signals are the most reliable indicators of a problem worth solving
- Probe every generalization with a specific instance request: tell me about a specific time this happened
- Synthesize findings using opportunity statements that describe user type, job to be done, current failure mode, and negative outcome without prescribing a solution