left arrow icon

Run 15 problem interviews before you build

Problem interviews ask people about the last time they hit a problem, what they tried and what it cost, without pitching. In a review of empirical tests, interview studies reached saturation within 9 to 17 interviews for a narrow, homogeneous group. Recruiting is the hidden cost: one team sent over 400 cold emails in four weeks to book its calls. Interest in an interview is not a purchase.

Evidence C, limited evidenceStart here

Stage
Stage 1. No proven demandIs the problem real, and does anyone want this solved?
Category
Customer research
Time
1 to 2 weeksAwiser's estimate of the active work
Cash
No cash
Live calls
Needs live calls
First result
Within weeks
Risk
LowThe cost is time, mostly spent recruiting. The risk is reading polite interest as demand.
Needs
A list of people who visibly have the problemA short script of questions about past behaviorA way to take detailed notes, or to record with permission

When does it fit?

  • You have not built yet and want to learn what people do today: YC suggests asking what, if anything, they have done to solve the problem.[1]
  • Your target is one narrow group: a systematic review in Social Science & Medicine found interview studies of fairly homogeneous groups reaching saturation within 9 to 17 interviews.[5]
  • You can reach people through a warm network: in a Jeff Gothelf case study, a founder found cold LinkedIn outreach had a low success rate and turned to personal and investor networks.[6]

When should you skip it?

  • You want the idea confirmed rather than tested: The Mom Test starts from the premise that people lie to you about your idea, and YC counts compliments among the worst data an interview produces.[1][2]
  • You can only talk in writing: no source found compares written problem interviews with live ones, and Steve Blank calls in-person interviews the gold standard.[3]

How do you run it?

  1. Pick one narrow group

    One role, one context, one problem. A broad target needs many more interviews before answers start to repeat.

  2. Recruit people who have the problem now

    Start with warm introductions from friends, coworkers and communities, then strangers. Expect recruiting to take longer than the calls.

  3. Ask about the past, not the future

    Ask about the last time the problem happened, what they did, what they tried, and what it cost in time or money.

  4. Keep the product out of it

    Do not pitch. Write compliments down as noise, not as evidence.

  5. Take detailed notes

    Capture exact words and numbers, or record with permission. You cannot know in advance which details will matter.

  6. Stop when answers repeat

    Track the new themes each interview adds. When several in a row add nothing, you have enough for this group.

What usually happens?

  • Recruiting takes effort: Ankit, a Recontact founder, reports sending over 400 cold emails as a team, for 4 weeks, to book discovery calls.[7]
  • Discovery programs ask for far more: the U.S. National Science Foundation requires at least 100 potential customer interviews in its seven-week I-Corps program, and Steve Blank expects more than 100.[3][4]
  • Some founders go further still: in a case study by Jeff Gothelf, one founder spoke to more than 300 companies before building anything.[6]

Who has done it, and what happened?

Each case is one team's own report: what happened to them, not a typical result.

  • Ankit and the other Recontact founders, then pre-revenue2024, Mixed result

    Booked discovery calls through cold email for four weeks and talked to dozens of people in their target domain before deciding what to build.

    Ankit reports that a chance conversation led to a first customer 2 days later, paying $25.[7]

    The team sells a tool that analyses user interviews, and a later week of outreach produced no sale.

What does the evidence say against it?

  • Interest is not buying: after talking to over 60 people, Ankit reports a week of more than 300 outreach requests with not a single sale, and buyers who came from another persona.[7]
  • No dataset found links the number of interviews before building to how a startup does.

How do you read the result?

What to count
Interviews done, and how many gave a recent example of the problem, what they tried and what it costs
When to read it
After each block of five interviews
How to decide
Build for this group only if people name a recent example and a real cost. YC asks for a hard number on what the problem costs; if nobody can give one, change the group or the problem.[1]

What goes wrong?

  • Pitching during the interview: YC says an interview is not the time to pitch the product.[1]
  • Asking about the future: YC recommends asking about the last time the problem happened instead of hypotheticals.[1]
  • Skipping notes: YC advises detailed notes because you cannot know which facts will matter later.[1]
  • Waiting for perfect strangers: YC lists friends and coworkers as an early step, to talk to and for warm introductions; the safeguard is how you ask.[1]

How was it graded?

Evidence
C. Limited evidenceA widely taught method and a review of saturation studies support how to run the interviews, but no dataset measures what they change, and the one recent case with numbers is mixed and comes from a team that sells an interview tool.
Verdict
Start hereBefore building, asking people about their last brush with the problem is the cheapest check there is. Treat it as learning what to build, not as proof that anyone will pay.

How the grades and verdicts are decided

Sources

  1. 1
    Startup School Week 1 Recap: Kevin Hale and Eric Migicovsky (opens in a new tab)Y Combinator, Eric Migicovsky. September 4, 2019. Method.
    Fourth question is, what, if anything, have you done to try to solve this problem?
    Some of the kind of worst bad data that you may encounter is compliments.
    I like to get a hard number, either in terms of how much revenue do they stand to earn if they solve this problem or how much expense do they currently spend trying to solve this problem, how much money is wasted today as they try to solve this problem?
    We're founders, we love to pitch our idea, we love to talk about the product that we're working on, but during a user interview, that is not the time to be pitching the product.
    The second question, to the point that I was making earlier about trying to get to specifics rather than hypotheticals is to ask the question, tell me about the last time that you encountered this problem.
    Some tips for this stage, take notes, take detailed notes, because like I said before, you'll never know until later which key facts of these user interviews may be useful.
    The next step after that is to talk to friends, is to talk to coworkers to get warm introductions.
    Read on October 20, 2026.
  2. 2
    The Mom Test (opens in a new tab)momtestbook.com, Rob Fitzpatrick. Undated. Method.
    How to talk to customers & learn if your business is a good idea when everyone is lying to you.
    Read on October 20, 2026.
  3. 3
    Customer Discovery In the Time Of the Covid-19 Virus (opens in a new tab)steveblank.com, Steve Blank. April 7, 2020. Method.
    The gold standard are in-person interviews where you can not only do all of that, and get to see what's on their desks, the awards on their walls, the books on their shelves, and other ephemera that may give you clues about their interests and behavior.
    Given you're not traveling to customer interviews, you ought to be able to do more than 100 interviews.
    Read on October 20, 2026.
  4. 4
    I-Corps National Teams Applicants (opens in a new tab)U.S. National Science Foundation. Undated. Method.
    Complete a minimum of 100 potential customer interviews during the seven-week training program.
    Read on October 20, 2026.
  5. 5
    Sample sizes for saturation in qualitative research: A systematic review of empirical tests (opens in a new tab)Social Science & Medicine, Monique Hennink, Bonnie N. Kaiser. November 2, 2021. Study.
    Studies using empirical data reached saturation within a narrow range of interviews (9-17) or focus group discussions (4-8), particularly those with relatively homogenous study populations and narrowly defined objectives.
    Read on October 20, 2026.
  6. 6
    Case Study: How 3 startups built their customer discovery practice (opens in a new tab)jeffgothelf.com, Jeff Gothelf. September 11, 2023. Analysis, from a seller of what it measures.
    Initially Cohn did cold outreach on LinkedIn. With a relatively low success rate, he then began reaching out to his personal network as well as his investor's networks.
    Before building any product Cohn, on his own, spoke to more than 300 different companies.
    Read on October 20, 2026.
  7. 7
    How we saved our company with 100+ user interviews in a month. (opens in a new tab)Recontact, Ankit. April 7, 2024. Founder report, from a seller of what it measures.
    We collectively, as a team, atleast 400+ cold emails, straight for 4 weeks and got our calls booked.
    Once we'd talked to over 60+ people and developed an acumen, is when the board was reset.
    The past week, we just had 300+ outreach requests and not a single sale. And then, we ended up selling to a totally different persona that came through inbound - "early stage founders".
    2 days later he became our 1st customer and we made our 1st $25 with Recontact.
    Read on October 20, 2026.

Sources last checked on October 20, 2026. First published on October 20, 2026.

Awiser is the builder network where founders show their skills, find collaborators, and grow their projects with an AI copilot.