How to Run Your First Customer Interview: A Step-by-Step Guide for First-Timers (2026)

The honest answer up front

If this is your first customer interview, you are going to feel awkward, you are going to ask at least one bad question, and the conversation will still be more valuable than the next ten internal debates about what users want. The goal of a first interview is not to do it perfectly — it is to do it. This guide gets you to "competent enough to start" in 15 minutes of reading.

Here is the full path you will walk in this guide:

1. Define a sharp objective for the interview (1 sentence)
2. Identify and recruit 1 person who fits the segment
3. Write a 30-minute discussion guide using the Mom Test rules
4. Run the interview (with a script for the awkward opening and closing)
5. Capture insights in the next 30 minutes — never later
6. Decide what you learned and whether to keep going

We will walk through each step. By the end you will have a usable kit you can run tomorrow.

Step 1 — Define the objective

A bad objective: "Learn about customers."

A good objective is one sentence with a specific decision attached:

• "Decide whether our team should build the bulk import feature next quarter."
• "Understand why 60% of new signups never invite a second user."
• "Find out how solo bookkeepers price-shop accounting software."

If you cannot finish the sentence "After this interview, I will be able to decide ___", your objective is too vague. Rewrite it.

Pro tip: A first interview can only realistically explore one objective. Resist the urge to cram a pricing study, a UX study, and a positioning study into 30 minutes. Pick one.

For a longer treatment, see writing a research question.

Step 2 — Find one person who fits the segment

For a first interview, you do not need a recruiting pipeline. You need one person who actually fits the user you are designing for.

Where to find them, in rough order of cost and effort:

1. Existing customers. Pull the 5–10 most recent signups. Email 3 of them.
2. Existing pipeline. Sales has a list of prospects who declined — they will often talk for 20 minutes.
3. LinkedIn search + cold outreach. Search for the exact title at the exact company stage. Send a short message (template below).
4. Your investor or advisor network. They almost certainly know someone who fits.
5. Niche communities (Slack groups, subreddits, Discord servers, IndieHackers). Lurk first, then ask.
6. Recruiting platforms (User Interviews, Respondent) — fast, but $50–$150 per participant.

Cold outreach script (proven)

Subject: Quick question, 25 minutes, [your topic]

Hi [Name],

I am the [founder/PM] of [Company], and I am trying to understand how [specific segment — e.g., "solo bookkeepers managing 5–20 clients"] actually handle [specific problem — e.g., "month-end close"]. I am not selling anything — I am trying to learn what to build.

Would you have 25 minutes this week? I would happily share what I am hearing across the other conversations in return.

Thanks, [Your name]

What makes this work:

• "Not selling anything" — explicit
• A specific segment, not a generic ask
• A fair trade (you share findings back)
• Short — under 100 words

Expected response rate: 10–25% from cold LinkedIn, 30–50% from warm intros, 40–70% from existing customers. Sending 10 messages to get 1 interview is a totally normal first ratio.

For more, see our user research recruitment email templates and research participant incentives guide.

Step 3 — Write a 30-minute discussion guide

The single most-cited methodology for customer interviews is Rob Fitzpatrick's The Mom Test — a book taught at Harvard, MIT, and UCL and used as a training manual at companies like SkyScanner and Shopify. The book's premise: if you ask your mom whether your business is a good idea, she will say yes — because she loves you. Most customer conversations have the same flaw. People tell us what we want to hear.

The Mom Test gives you three rules to dodge that trap. They are the foundation of every good first interview:

Rule 1 — Talk about their life, not your idea. Do not describe your product in the first 25 minutes. Ask about how they currently solve the problem. Their answer is data; your idea is noise.

Rule 2 — Ask about specifics in the past, not generics or hypotheticals. "Walk me through the last time you had this problem" beats "Would you use a tool that did X?" every single time. People are unreliable forecasters about themselves. They are accurate historians of their last 7 days.

Rule 3 — Talk less. Listen more. If you are talking more than 30% of the time, the interview is broken. Pause when they pause — silence pulls more out of people than any prompt.

For a deeper treatment, see the Mom Test methodology and the Mom Test user interviews guide.

The 30-minute discussion guide template

Minutes 0–3 — Warm up. Thank them, set expectations, ask permission to record (or note-take). State the topic broadly and the time commitment. Then ask one easy question that gets them talking.

• "Can you tell me a bit about your role and what a typical week looks like?"

Minutes 3–10 — Past behavior. Anchor the conversation in something they actually did.

• "Tell me about the last time you had to [do the task]."
• "Walk me through what happened step by step."
• "When was the last time it went badly? What happened?"

Minutes 10–20 — Current solution & switching cost.

• "How do you handle this today?"
• "What tools or workarounds are you using right now?"
• "What have you tried before that didn't work?"
• "If you tried to switch to a better solution, what would it take? Who else would need to be involved?"

Minutes 20–25 — Outcomes & priorities.

• "If [this problem] disappeared tomorrow, what would that be worth to you? How would your week change?"
• "What does 'good enough' look like for you in this area?"

Minutes 25–28 — A gentle peek at your idea (only if useful). Only mention your product if you have time and the conversation has earned it. Phrase it as "We are exploring ___ — does anything about that resonate or not?" rather than "Here is what we are building."

Minutes 28–30 — Close. Ask "Who else should I be talking to?" — this is the single most underused close in customer research. Thank them. Tell them you will share what you learn.

Step 4 — Run the interview

The first 60 seconds matter more than the rest

The research-backed term for this is rapport building (see our building rapport in interviews guide). The mechanics:

• Use their name. "Hi [Name], thanks so much for the time."
• Acknowledge the asymmetry. "I know you are busy — this is genuinely useful for me, and I will keep us to 25 minutes."
• State the topic and your role honestly. "I am trying to understand how [segment] handles [problem]. I am not selling anything today."
• Get explicit permission to record or take notes.
• Ask one easy throwaway question so they hear their own voice in the conversation before you ask anything hard.

For consent and ethics, see our research ethics guide and research consent form templates.

The 10 questions you can ask in almost any first interview

These 10 questions cover most discovery and validation objectives. Pick 4–6 for any given interview, plus probes.

1. "Can you tell me a bit about your role and what a typical week looks like?"
2. "Tell me about the last time you had to [do the task]. Walk me through what happened."
3. "What was the hardest part about that?"
4. "How do you handle this today? What tools or workarounds do you use?"
5. "What have you tried before that didn't work? Why didn't it stick?"
6. "When was the last time this caused real pain for you?"
7. "Who else is involved when you do this?"
8. "What does 'good enough' look like for you?"
9. "If this whole problem disappeared tomorrow, what would change?"
10. "Who else should I be talking to about this?"

The 5 questions to avoid in a first interview

1. "Would you use it?" — a hypothetical, almost always answered politely. Replace with: "Tell me about the last time you used something like this."
2. "How much would you pay for it?" — guess work, not data. Pricing comes from observed switching costs and current spend, not hypothetical willingness to pay. See pricing research interviews.
3. "Do you like it?" — Compliments are the fool's gold of customer learning: shiny, distracting, and worthless. Replace with: "What about it works? What does not?"
4. "Don't you think [your hypothesis] is a problem?" — leading question. Replace with: "What are the biggest problems in this area?"
5. "Would you recommend us?" — out of place pre-product; if you have a product, use an NPS survey instead.

For more on this, see avoiding bias in interviews and the research bias guide.

Probing — going deeper without leading

When a participant says something interesting, your next move is a non-leading probe:

• "Tell me more about that."
• "What do you mean by [their word]?"
• "Can you give me an example?"
• "What happened next?"
• "How did that make you feel?"
• (Silence — 3 seconds. They will fill it.)

Notice none of these inject your opinion. Each one widens the answer. See probing and follow-up questions for a deeper treatment.

How to take notes without breaking eye contact

Three options, in increasing order of quality:

1. Pen and paper. Write down verbatim phrases, surprises, contradictions, and emotion. Skip everything you can paraphrase later. Good for first-timers because the friction forces you to filter.
2. Record + transcribe later. Pair a recording with very light real-time notes (timestamps + 3–4 quote fragments). Transcribe afterward.
3. AI-moderated or AI-assisted interviews. Modern platforms transcribe, theme, and quote-tag in real time so you can focus entirely on the conversation. See our note-taking in user research and AI transcription for research interviews guides.

Step 5 — Synthesize in the next 30 minutes

The single biggest mistake new interviewers make: leaving the synthesis for "later." Memory fades fast. The richest patterns disappear within 24 hours.

Spend 15–30 minutes immediately after the interview filling in this simple template:

INTERVIEW DEBRIEF — [Participant ID] — [Date]

Objective revisited: [Did this interview move the decision?]

What surprised me (top 3 surprises):
1.
2.
3.

Verbatim quotes worth keeping (3–5):
- "..."
- "..."

Behaviors observed (what they actually do, not what they say they want):
-
-

Pains, ranked by severity in their own life:
1.
2.
3.

Contradictions or doubts (things that did not add up):
-

Open questions for next interview:
-

This 7-field debrief is the difference between a research practice and a notebook full of forgotten conversations. For a deeper version, see research debrief: how to synthesize and share findings after every study.

Step 6 — Decide what you learned

One interview is not enough to make a confident product decision — but it is enough to:

• Sharpen your hypothesis
• Disconfirm at least one assumption
• Find a phrase, a workaround, or a frustration you did not know existed
• Decide what to listen for in the next interview

The right cadence for a first study is 5–8 interviews in the same segment, then a synthesis pass. Most teams find that themes start to repeat by interview 5–7. For the methodology behind this, see how many interviews are enough and data saturation in qualitative research.

The modern shortcut: AI-moderated first interviews

For first-time interviewers, the hardest part of the process is everything around the conversation — recruitment, scheduling, transcription, synthesis, and the social pressure of moderating a stranger live.

AI-moderated platforms like Koji collapse the workflow so a non-researcher can run a study in a few days:

• Share an interview link with your 10 candidates — no scheduling. They answer on their own time.
• Koji's AI moderator asks your discussion guide questions, probes intelligently when answers are shallow, and adapts in real time.
• Voice or text — participants pick what they prefer. Voice interviews capture the same depth as live calls.
• Real-time transcription, theme extraction, and quality scoring — you see patterns emerge as interviews complete.
• AI follow-up probing turns a 1-line survey answer into a 5-minute dialogue. See our AI probing guide.
• Built-in Mom Test guardrails — the AI moderator is trained to ask about past behavior, not hypotheticals; specifics, not generics.

A traditional first study takes 3–4 weeks for a non-researcher (recruit 1 week, schedule 1 week, run 1 week, synthesize 1 week). A Koji-run first study takes 3–5 days. For founders running customer discovery weekly, that difference is the difference between continuous discovery and a one-time project.

See customer discovery interviews at scale and continuous discovery: how to run weekly customer interviews without burning out for how this scales beyond your first interview.

What "good" looks like on your first try

After your first interview, here is the checklist of what you are aiming for. Hitting 6 of 10 is a great first run.

• The participant talked at least 70% of the time
• You captured at least 3 verbatim quotes
• You learned at least one specific past behavior you did not know about
• You found at least one thing that surprised or contradicted your assumption
• You did NOT pitch your product in the first 20 minutes
• You used at least 2 non-leading probes ("Tell me more," "What do you mean by...")
• You closed with "Who else should I talk to?"
• You synthesized within 30 minutes of ending the call
• You identified at least one new question for interview #2
• You can answer: did this move the decision in your objective forward?

If you missed 4+ items, the next interview is your chance to fix them. The improvement curve between interview #1 and interview #5 is steeper than at any other point in your research career — protect that learning by interviewing more, not by reading more about interviewing.

Related Resources

• Structured Questions in AI Interviews — the 6 question types that pair with conversational probing in Koji
• The Mom Test: How to Talk to Customers Without Being Misled — the foundational methodology referenced throughout this guide
• Customer Interview Questions: 60+ Examples — a reusable question bank for discovery, validation, pricing, and churn
• User Interview Guide Template — a more detailed discussion guide framework once you outgrow this primer
• Probing and Follow-Up Questions — going deeper without leading
• Avoiding Bias in Research Interviews — the most common ways first-timers leak their assumptions
• How Many Interviews Are Enough? — when to stop and synthesize
• Customer Discovery Interviews: The Complete Guide — the next-level methodology after your first 5–10 interviews
• Customer Research Habit: A 30-Day Plan for Founders and PMs — how to make customer interviewing a weekly muscle, not a one-time event