Validation SOPs › Customer Discovery

SOP 1 of 100 · Customer Discovery

Customer Interview SOP

In short: Talk to 15 to 20 people who match one narrow customer profile, ask about their past behaviour instead of your idea, and score every call on problem clarity, urgency and money already spent. Ten or more strong scores means the problem is real enough to test a solution.
Difficulty
Beginner
Time
2-4 hours of prep, 1-2 weeks of calls
Cost
$0-100
Best for
B2B and B2C startups before they build anything
Tools
Zoom, Calendly, Notion, Otter.ai
Measure
Problem clarity score, Customer urgency level, Willingness to pay indicator

A customer interview SOP is the cheapest way to find out whether strangers share the problem you want to solve, and it takes about two weeks of calls. You talk to 15 to 20 people from one tightly defined group, ask what they did the last time the problem hit them, and write down facts instead of compliments. Founders who skip this step build for an imaginary customer. Founders who run it well walk away with the exact words their buyers use, a ranked list of problems and the first names for a waitlist.

What is the customer interview SOP and when should you use it?

A customer interview is a 30 to 45 minute one-on-one conversation with someone who fits your target customer profile. The goal is to learn how they experience a problem today: how frequently it happens, what it costs them and what they already tried. You are not pitching. The moment you describe your product, people switch to being polite, and polite answers are useless for validation.

Use this SOP:

  • Before you write code or design screens, when the idea is still a hypothesis.
  • When a survey gives you numbers you can't explain. Interviews tell you the "why" behind the data from an online survey validation.
  • When sales stall. Ten interviews with lost prospects show you which part of the story fails.

Rob Fitzpatrick's book The Mom Test sums up the rule in one line: ask about their life, never about your idea. Our sister site has a summary of The Mom Test in its list of 100 startup books, and this SOP turns its advice into a repeatable process.

What do you need before you start?

  • One customer profile, written down. "Freelance designers in Europe who invoice more than 5 clients a month" works. "Small businesses" doesn't. The persona development SOP or the free user persona generator gets you there in an hour.
  • A problem hypothesis in one sentence: who has the problem, when it happens and what it costs.
  • A booking link (Calendly or similar) and a video tool with recording.
  • A scoring sheet in Notion or a spreadsheet, ready before the first call.

How do you run it, step by step?

1. Define the person you need to hear from

Write the profile with three filters: role or life situation, a trigger event ("hired their first employee in the last 6 months") and a channel where you can reach them. The trigger event matters most, because people who lived through the problem recently remember the details.

2. Write open questions about past behaviour

Prepare 8 to 10 questions that start with "Tell me about the last time…", "What did you do next?" and "What did that cost you?". Ban hypothetical questions such as "Would you use…?" and "Would you pay for…?". People are bad at predicting their own behaviour and very good at describing what already happened.

3. Recruit and schedule 15 to 20 interviews

Post in communities where your profile gathers, ask your network for introductions and send short direct messages. Offer something small in return: a summary of what you learn, a gift card worth $10 to $25, or early access. Expect to contact 5 to 10 people for every interview you book.

4. Record with permission and listen more than you talk

Ask for consent to record at the start of the call. Aim for an 80/20 split: they talk, you listen. When they mention a strong emotion or a workaround, dig in with "Why was that hard?" or "Can you show me?". Stop each call by asking who else you should talk to.

5. Score every call within an hour

Fill in your scoring sheet while the call is fresh: problem clarity (1 to 5), urgency (1 to 5) and money or time already spent on a fix. Copy their exact phrases into a "quotes" column. These quotes become your landing page headlines later.

6. Look for patterns after every five calls

Group the notes by problem, trigger and workaround. After five calls you will see the first repeat themes. Adjust the questions if you keep hearing about a different, sharper problem than the one you expected.

7. Write the decision memo

After 15 to 20 interviews, write one page: the top three problems in the customers' words, how many people scored 4 or 5 on urgency, what they pay today and your decision (go, change the problem, or change the customer).

Template you can copy

CUSTOMER INTERVIEW SCRIPT (30–40 min)

Intro (2 min)
"Thanks for your time. I'm researching how [profile] handle [problem area].
I'm not selling anything. Is it OK if I record so I can take fewer notes?"

Context (5 min)
1. Tell me about your role and a typical week.
2. Where does [problem area] show up in that week?

Last occurrence (15 min)
3. Tell me about the last time [problem] happened. What triggered it?
4. What did you do next? Walk me through it step by step.
5. What did that cost you in time, money or stress?
6. What have you tried to fix it? Why did you stop or keep using it?

Priorities (8 min)
7. Where does this rank among the problems you deal with every month?
8. If it disappeared tomorrow, what would change for you?

Close (5 min)
9. Who else should I talk to about this?
10. Can I follow up when I have something to show you?

SCORING (fill in within 1 hour)
Problem clarity 1–5 | Urgency 1–5 | Spends money/time today Y/N |
Best quote | Follow-up OK Y/N | Referral names

Worked example: a booking tool for physiotherapists

Here is an illustrative case. A founder wants to build appointment software for independent physiotherapists. Her hypothesis: "Solo physios lose income from no-shows." She interviews 16 physios recruited through two professional Facebook groups and LinkedIn.

The scores tell a different story than she expected. Only 4 of 16 rate no-shows as urgent, because most already charge a cancellation fee. Eleven of 16 rate insurance paperwork a 4 or 5, and nine of them pay an assistant or spend Sunday evenings on it. One quote repeats in different forms: "I didn't train for six years to fill in forms."

Her decision memo changes the problem, keeps the customer, and the next step becomes a landing page test for paperwork automation. Two weeks of calls saved her months of building the wrong product.

What should you measure, and when do you stop?

MetricPass signalWarning signal
Problem clarity score10+ of 15–20 people describe the same problem without promptingEvery person describes a different problem
Customer urgency level50% or more score urgency 4 or 5Most say "it's annoying, but fine"
Willingness to pay indicator5+ people already pay for a workaround (tool, freelancer, extra hours)Nobody has tried to fix it
Referrals1 or more introductions per 3 callsPeople decline to refer anyone
Follow-up permission70%+ agree to see a prototypePolite refusals

Stop when new calls stop adding new themes. Griffin and Hauser's Voice of the Customer research in Marketing Science found that 20 to 30 interviews uncover roughly 90 percent of customer needs in a category, so a narrow profile reaches that point at 15 to 20 calls.

Which mistakes should you avoid?

  • Pitching during the call. Save the demo for a separate meeting with the people who scored high.
  • Interviewing friends and family. They protect your feelings. Recruit strangers who match the profile.
  • Asking "Would you pay for this?" Ask what they paid for the last fix instead.
  • Counting compliments as validation. "Cool idea" is noise. Money, time and repeated workarounds are signal.
  • Talking to everyone. A mixed group of freelancers, agencies and enterprises gives you three half-answers instead of one clear one.

How can AI speed this up?

Transcription tools such as Otter.ai turn each call into searchable text. Paste five transcripts into an AI assistant and ask it to list every problem mentioned, with exact quotes and the number of people who said it. Check the output against your own notes, because AI summaries smooth over the hesitation and emotion that tell you how urgent a problem feels. You can also ask an AI to rewrite your questions so none of them are leading.

If you want a co-founder that runs this kind of analysis with you, the F/MS AI co-founder works through validation tasks step by step.

Where does this SOP fit in your validation plan?

Customer interviews come first in the Customer Discovery stage. Next steps that build on them:

  1. Turn the quotes into a sharp problem statement with the problem statement validation SOP.
  2. Map what customers think, feel and do with customer empathy mapping.
  3. Fill knowledge gaps with an expert interview.
  4. Check how many people share the problem with an online survey.

See all 15 Customer Discovery SOPs. To practise the whole process with deadlines and feedback, the Fe/male Switch startup game walks first-time founders through interviews, validation and launch, and the free idea validator gives a quick first read of your concept.

FAQ

How many customer interviews do I need?

Plan for 15 to 20 per customer profile. Stop earlier if five calls in a row bring no new themes, and run more if your customer group is broad.

How long should a customer interview take?

30 to 45 minutes. Shorter calls stay on the surface. Longer calls tire both sides and drift into pitching.

Should I pay people to talk to me?

A small thank-you of $10 to $25 or early access is fine for consumer interviews. For B2B, offer the summary of your findings; busy professionals value that more than a gift card.

Can I run customer interviews by email or chat?

You can, but you lose tone of voice and the follow-up questions that uncover the real story. Use written answers only as a backup for people who refuse calls.

Go further on our sister sites