How to Talk to Users: The Modern Customer Interview Playbook for Founders (2026)

TL;DR: "How to Talk to Users" is the title of Eric Migicovsky's seminal Y Combinator Startup School lecture — and it is the single most-cited methodology in founder customer interviewing. Migicovsky distills the entire craft into five rules and five questions. The rules: never pitch, never ask hypotheticals, always ask about specific past behavior, talk to real customers (not friends giving polite feedback), and listen 90% of the time. The five questions cover problem severity, current alternatives, last occurrence, frustration, and willingness to act. The bar for the modern founder is no longer 5-10 interviews per month — it is 15-20 conversations per week. Koji is built so a solo founder can hit that bar without becoming a full-time interviewer.

Why Talking to Users Is the Single Highest-Leverage Founder Activity

At Y Combinator, partners tell founders that there are exactly two things they must do to start a company: code (or build the product) and talk to users. Everything else is procrastination.

Eric Migicovsky — founder of Pebble Smartwatch and current YC partner — formalized this in his "How to Talk to Users" lecture for YC Startup School. The lecture has been viewed millions of times, translated into a dozen languages, and quoted in nearly every modern customer-development book.

The premise is simple: the best founders maintain a direct connection to their users throughout the lifespan of the company. This is not a stage you complete and graduate from. It is a forever loop.

Why this is high-leverage: roughly 42% of startups fail because there is no market need (CB Insights), and the only known reliable way to detect that risk before you have burned 18 months of runway is to talk to users early, often, and well.

The Five YC Rules for User Interviews

Rule 1: Don't talk about your idea

This is Migicovsky's rule #1, and it is the rule founders break most often. The instinct to pitch is overwhelming — you have lived inside the idea for weeks, you're proud of it, and you want validation.

But during a user interview, you are not selling. You are extracting information. The moment you start pitching, the interviewee shifts into politeness mode and starts telling you what you want to hear. The data is now poisoned.

What to say instead: "I'm researching how people currently [do the thing]. I'm not going to talk about my product today — I really just want to understand your experience."

Rule 2: Don't talk about hypotheticals

Asking "Would you use a product that does X?" produces wildly optimistic and useless data. Customers are bad at predicting their own future behavior. They overestimate how much they would pay, how often they would use something, and how much they care about a problem.

What to ask instead: "Tell me about the last time you encountered this." Past behavior is a far better predictor of future behavior than stated intention.

This is the same insight that powers The Mom Test by Rob Fitzpatrick — and it is the single most important shift a new interviewer can make.

Rule 3: Listen 90% of the time

A good user interview is 90% interviewee, 10% you. If you find yourself talking more than that, you are either pitching, leading, or explaining — none of which produce useful data.

Embrace silence. After an answer, count to three before you ask the next question. Most of the best insights come in the second sentence the interviewee says after you stop talking.

Rule 4: Talk to real potential customers, not friends

Friends and family give polite feedback. Polite feedback is worse than no feedback because it gives false confidence. Migicovsky's advice: start with friends and coworkers for warm introductions, then use them to reach strangers in your target segment.

Cold outreach feels uncomfortable but is essential. The bar is 15-20 customer interviews per week for an actively-researching founder.

Rule 5: Ask about the past, capture specifics

Vague answers ("It's annoying sometimes") are not insights. Specifics ("Last Tuesday I spent 90 minutes manually copying invoices into QuickBooks because the export was broken") are insights.

Always drill from the general to the specific: "What's the hardest part?" → "Tell me about the last time that happened." → "Walk me through that morning step by step."

Eric Migicovsky's Five Questions for Every User Interview

Migicovsky boiled down the entire founder interview kit into five questions. Memorize them.

Question 1: "What's the hardest part about [accomplishing the goal you're focused on]?"

This is your opener. It is broad enough that the customer takes it in their own direction, which reveals what they actually care about (not what you suspect they care about). Note that you are asking about their goal, not your product.

Question 2: "Tell me about the last time you encountered this problem."

This drives from abstract to specific. You are now collecting concrete behavioral data — a real episode, with a real date, real people involved, and a real outcome.

Question 3: "Why was this hard?"

Drill into the why. The first answer is usually the symptom; the real problem is two or three "whys" deeper. This is borrowed from the 5 Whys technique used in lean manufacturing.

Question 4: "What, if anything, have you done to try to solve this problem?"

This question is gold for two reasons:

1. It tells you what the existing alternatives are (competitors include "doing nothing" and "a manual workaround").
2. It tells you whether the customer cares enough to actually act. A customer who has googled solutions, tried two of them, and built a hacky workaround is a real customer. One who has done nothing is not.

Question 5: "What don't you love about the solutions you've tried?"

This is your bridge to your own product (without pitching it). You are mapping the white space your product can fill. The customer's answer here is also your future positioning copy.

Five questions, in this order, will produce more insight than 30 minutes of unstructured chat.

How Many Interviews Should a Founder Run?

This is the most-asked question in early-stage research, and the consensus has shifted over the last decade.

For a founder in the discovery phase, a practical bar is: 5-10 interviews per customer segment to surface initial patterns, then 12-20 to reach saturation, then continuous interviewing forever. Marketing-research traditions suggest 20-30 interviews to capture 90-95% of customer needs in a segment.

The harder truth: interview quality matters more than interview quantity. A founder who runs 5 well-structured interviews with the five YC questions will learn more than a founder who runs 30 unfocused chats.

How to Source Users to Talk To

The cold-outreach problem is the #1 blocker for new founders. Here is the practical stack, in order of efficiency:

1. Personal network — Start here. Ask every friend and former coworker if they know anyone in your target segment. Aim for warm intros.
2. LinkedIn search — Filter by job title, company size, geography. Send personalized DMs ("I'm researching how X people currently do Y — would you have 20 minutes?"). Expect 5-15% response.
3. Reddit, Twitter/X, niche communities — Find the subreddit or Twitter cluster where your target customer hangs out. Read 100 posts. Comment thoughtfully for a week. Then DM.
4. Customer support / sales channels of your existing product — If you already have any product, every churn ticket and every demo no-show is a user interview opportunity.
5. In-app intercept — A timed prompt on a relevant in-product action: "Got 5 minutes to share why you tried X?" can convert at 5-10%.
6. Participant recruitment platforms — User Interviews, Respondent, Prolific — for when you've exhausted organic channels and need volume. Budget $50-150 per interview.

The Modern Bar: 15-20 Conversations a Week

The old founder norm was "We did 30 interviews before launch." The new bar — set by the YC-backed founders running continuous discovery — is 15-20 customer conversations per week, indefinitely.

This is mechanically impossible for most solo founders to do alone. You would spend 25+ hours/week on interviews alone, plus another 10 on synthesis. That is two-thirds of your week.

The math is exactly why AI-moderated interviewing has become standard at YC-stage startups: it is the only way for a 2-person team to actually hit the bar.

How Koji Modernizes "How to Talk to Users"

Koji is built around the YC interview philosophy, but at machine scale:

• The 5 YC questions are baked into the Customer Discovery and Mom Test methodology frameworks. You don't have to remember to ask about the past — the AI does, and probes for specifics by default. The discussion guide writes itself from your research goal.
• AI-moderated voice and text interviews run 24/7. Send a link by DM, in an email signature, on a landing page, or inside your product — respondents complete a 5-15 minute interview at their convenience. You wake up to 20 interviews instead of 0.
• The AI never pitches and never accepts hypothetical answers. It is trained to redirect ("That's interesting — can you tell me about a specific time that happened?") which a tired human interviewer often forgets to do.
• Probing follow-ups happen automatically. When a respondent says "It's annoying," the AI follows up with "Can you describe the last time it was annoying?" This is the laddering technique that turns vague feedback into specific evidence.
• Six structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no) let you ask the YC five questions alongside quantitative items in the same interview — e.g., a 1-5 severity scale after Question 3, or a force-ranked list of existing alternatives after Question 4.
• Real-time thematic analysis clusters answers across all respondents. Instead of re-reading 20 transcripts, you get a ranked list of pains, alternatives, and unmet gains within an hour of the last interview.
• Verbatim customer quotes are surfaced for each theme. This is what populates your landing page copy, your pitch deck, and your investor update.
• Always-on interview links turn every cold lead, every churn, every signup into a discovery opportunity. The 15-20-per-week bar becomes feasible for a single founder.

Teams using AI-assisted research tools report 60% faster time-to-insight — which for a founder is the difference between updating their understanding of the customer weekly versus quarterly. In a market where competitors are iterating weekly, that gap is fatal.

Three Common Founder Mistakes (and the Fix)

1. Asking "Would you buy this?" → Replace with "Have you ever paid for something like this? How much, and what made it worth it?"
2. Defending the idea when challenged. → If a user pushes back, say "Tell me more about why" and write it all down. Discomfort is the data.
3. Ending interviews without a next step. → Always close with "Can I follow up in a few weeks once we've built more?" — every interviewee becomes a beta tester.

Related Resources

• Structured Questions Guide — How Koji's 6 question types pair quantitative validation with the YC qualitative questions.
• Mom Test Methodology — The deeper version of YC's "ask about the past" rule.
• Customer Discovery Interviews — Step-by-step format for the first 20 founder interviews.
• Customer Interview Questions Examples — 60+ tested questions for discovery, validation, pricing, and churn.
• How Many User Interviews — A data-grounded answer to the universal founder question.
• First Customer Interview Guide — A step-by-step walkthrough for your very first interview.
• Continuous Discovery Handbook — Teresa Torres's methodology for sustaining the YC bar after the discovery sprint.