Talking to Humans: The Founder's Customer Discovery Field Guide

Customer discovery is the disciplined practice of learning directly from real people before you commit to building. In Talking to Humans, Giff Constable frames it as a repeatable loop: decide who to learn from, recruit them, run interviews for stories instead of speculation, and turn what you hear into product decisions.

Quick Answer: Talking to Humans by Giff Constable teaches founders to validate ideas by talking to real customers instead of guessing. Choose a specific target segment, recruit roughly 15 to 20 people per persona, interview for past behavior and real stories, listen for genuine signal over politeness, then decide whether to persevere or pivot.

Why founders must get out of the building

Founders need to get out of the building because the facts required to validate an idea live with customers, not inside a conference room. Steve Blank's foreword to Talking to Humans states the premise bluntly: there are no facts inside your building, so you have to go outside to find them.

On day one, your startup is a stack of untested assumptions — about who has the problem, how badly it hurts, whether anyone will change their behavior, and whether they will pay to make the pain go away. Internal debate cannot turn a guess into a fact. Only contact with real people can.

The book's reframing is subtle but powerful: early on, your job is not to defend your idea but to pursue the truth. That means being genuinely willing to be wrong. Constable treats a startup as a search for a viable business model rather than the execution of a known one, which makes the scientific method — hypothesis, test, evidence — the right mental model.

In practice, that means writing down your riskiest assumptions as explicit hypotheses, then treating each conversation as a test designed to prove yourself wrong rather than right. If you want to see where discovery interviews sit within the broader research picture, our complete guide to customer research for founders maps how the pieces connect.

There is a simple economic argument underneath all of this. A few weeks of conversations is cheap; a few months of engineering aimed at the wrong problem is not. Discovery is the low-cost insurance you buy before making the expensive bet, and it only works if you do it before you are emotionally and financially committed to a particular answer.

The cost of skipping this discipline is familiar and expensive: months of heads-down building, a launch day that lands in silence, and a slow realization that you solved a problem nobody urgently had.

The customer discovery loop: four repeatable stages

Customer discovery is not a one-time event but a loop you run continuously, and Talking to Humans breaks it into four connected moves. Each stage answers a different question, and the output of one feeds directly into the next.

You rarely run the loop just once. A round of conversations sharpens your segment, which changes who you recruit, which changes what you hear, which sends you back out with better questions. The table below summarizes the loop and what "doing it well" looks like at each stage.

The takeaway: discovery is only as strong as its weakest stage. Perfect interview technique aimed at the wrong segment still produces garbage, and brilliant recruiting wasted on a thinly disguised sales pitch teaches you nothing. Treat all four moves as one connected discipline, not four separate skills.

Stage 1 — Decide who to learn from before you recruit

Start by defining a specific target customer, because "everyone" is not a segment you can find, interview, or serve. Talking to Humans pushes founders to name a narrow beachhead first and widen later, rather than trying to boil the ocean.

Constable draws a distinction worth keeping straight: the customer who pays and the user who uses may not be the same person. A parent buys software their child uses; a manager approves a tool their team relies on. Each role sees the problem differently, so each is its own discovery target with its own assumptions.

This matters even more in multi-sided markets. In a marketplace or platform, you have buyers and sellers, or riders and drivers, and a hypothesis that holds on one side can collapse on the other. Map every distinct stakeholder before you start recruiting.

Practical moves the book recommends at this stage:

• Write rough personas capturing who you think has the problem, based on your current best guess — while fully expecting to revise them.
• List your assumptions per persona: what you believe about their problem, their current workarounds, and their appetite to change.
• Separate customers, users, and influencers so you can recruit and question each group deliberately.
• Pick one beachhead segment to start — narrow enough that you could describe a single specific person who fits it.

The sharper your segment definition, the easier every later stage becomes. For a deeper framework on prioritizing who to approach first, see our guide on who to interview for customer discovery; the quality of your segment choice caps the quality of everything downstream.

Stage 2 — Recruit and screen the right participants

Finding people to talk to is a creative hustle, and Talking to Humans treats recruiting as a skill in its own right rather than an afterthought. The goal is a steady flow of conversations with people who genuinely match your target segment.

The book offers a menu of tactics rather than one prescribed channel:

• Tap your warm network first. Direct introductions convert fastest and lower the stranger barrier that makes strangers cautious.
• Go where your customers already gather, online and off: forums, communities, Slack groups, subreddits, meetups, conferences, and physical locations.
• Reach out cold by email and LinkedIn, kept short and framed as a humble request to learn, never as a pitch.
• Ask every interviewee for referrals so each conversation snowballs into the next two.
• Recruit with a simple landing page or ad when you need volume or want to test your messaging at the same time.

A note on incentives: small thank-you gestures can help, but heavy incentives risk attracting people who show up for the reward rather than because they feel the problem. Screen for genuine fit, not just willingness.

It helps to reframe the ask in your own head. You are not begging for a favor or fishing for a sale — you are inviting someone to talk about their own experience, which most people are surprisingly happy to do when they sense real curiosity rather than a pitch in disguise. Lead with humility, keep the commitment small, and be explicit that you are trying to learn, not to sell them anything.

On how many people to talk to, Constable is deliberately undogmatic: there is no statistically significant magic number. He offers a rough starting point of around 15 to 20 people per persona, but the real stopping rule is qualitative. Keep going until the answers stop surprising you and clear patterns repeat. When you can reliably predict what the next interviewee will say, you have enough signal for now.

One screening caution the book stresses: friends, family, and anyone eager to please you make poor first subjects. They optimize for your feelings, not the truth, and their encouragement is the most seductive false signal in early discovery.

Stage 3 — Run sessions for stories, not speculation

Run each session as a guided conversation aimed at real stories, not a survey and never a sales pitch. The single biggest shift Talking to Humans asks of founders is to stop talking about their own idea and start listening to the customer's actual experience.

Interview in pairs whenever you can. One person leads the conversation and stays present; the other takes detailed notes and watches for body language, hesitation, and emotion. Capture verbatim quotes, not tidy paraphrases — the exact words often carry the real signal.

A good session has a natural arc. Open by building rapport and understanding the person's context and role. Move into the problem itself: how it shows up in their life, how often, and what triggers it. Then explore how they handle it today and what those workarounds cost them. Only near the end, if at all, do you introduce anything about your own idea — and even then as a prompt for reaction, not a demo to be admired.

Anchor every question in past behavior, not the hypothetical future

The book's most important technique is to anchor questions in what has already happened, because people are unreliable predictors of their own future behavior but fairly honest narrators of what they have already done.

• Ask about the last time they faced the problem: what happened, what they did, and what it cost them.
• Favor open-ended questions over yes/no ones, and avoid leading questions that telegraph the answer you are hoping for.
• Do not pitch your solution. If you must show it at all, save it for the very end, after you have learned.
• Listen far more than you talk — aim to have the customer speaking most of the time, and get comfortable letting silence sit.
• Dig with "why" whenever something interesting surfaces; the first answer is rarely the real one.

The contrast is stark. "Would you use an app that does this?" invites a polite yes that means nothing. "Walk me through the last time you dealt with this — what did you actually do?" surfaces behavior you can trust. If you want a ready-made question set and session flow, our customer discovery interview guide and template turns these principles into a script you can run tomorrow.

Complement conversations with observation and small experiments

Talking is not your only discovery tool. Constable reminds founders that watching what people actually do — and running lightweight experiments — often reveals truths a conversation alone will miss. Observe someone perform the task, or float a small test like a landing page or a manual concierge version of the service, and behavior does the talking.

Above all, go in trying to disconfirm your hypotheses. A session that only ever confirms what you already believed usually means you asked leading questions, not that you were right.

Stage 4 — Make sense of what you heard

Turn raw conversations into decisions by separating what people say from what they actually do. Talking to Humans is emphatic that words are cheap and behavior is expensive — the strongest evidence is what someone has already spent time, money, or reputation on.

Debrief immediately after each session, ideally with your interview partner while memory is fresh. Then look across conversations for repeating patterns rather than fixating on any single vivid quote.

The hardest discipline here is fighting confirmation bias. Enthusiasm feels like validation, but a compliment is often just politeness. Use the rough guide below to translate what you hear into how much you should trust it.

The takeaway: treat commitment and existing workarounds as the currency of truth, and treat praise as noise until behavior backs it up. The three currencies that reveal real conviction are time, money, and reputation — a person who spends any of them on your problem is telling you far more than one who merely admires your idea. Whether you track all this in a spreadsheet or a dedicated workspace like Edmired, what matters is separating signal from politeness honestly.

Guard against false negatives as carefully as false positives. Sometimes a lukewarm response means the idea is weak; sometimes it means you talked to the wrong segment, framed the problem poorly, or caught people on a bad day. Before you abandon a hypothesis, ask whether the evidence is really about the idea or about how you ran the test. The point is honest interpretation, not pessimism or optimism.

Sense-making ends in a decision, not a document. Based on the patterns, you persevere with your current hypotheses, pivot to a new one, or run another round of discovery to resolve what is still uncertain. This loop is the engine underneath our complete guide to startup idea validation, which zooms out to the full validation journey from first conversation to a business worth building.

Common customer discovery mistakes founders make

Most discovery failures trace back to a handful of predictable mistakes, and Talking to Humans flags them repeatedly. Recognizing them in yourself is most of the fix.

• Pitching instead of listening. The moment you sell, the customer stops telling you the truth and starts reacting to your idea.
• Asking about the future. "Would you use this?" invites a polite yes; "what did you do last time?" reveals reality.
• Using leading questions. Questions that hint at your desired answer manufacture the validation you were hoping to find.
• Talking to the wrong people. Friends, family, and fans give you comfort, not signal.
• Hearing only what you want. Confirmation bias turns every warm comment into false proof.
• Treating opinions as commitments. "I love it" is not a purchase, a signup, or a workaround.
• Delegating discovery entirely. The book treats it as a founder's job and a team sport, not something to outsource or hand off and forget.
• Stopping too early. A handful of chats is not a pattern; talk to enough people that surprises give way to repetition.

Key Takeaways

• Customer discovery replaces internal debate with external evidence. As Steve Blank's foreword insists, there are no facts inside your building — get out and gather them.
• Discovery is a four-stage loop, not a one-off task. Decide who to learn from, recruit them, run sessions for stories, and make sense of what you heard — then run it again.
• A narrow segment beats "everyone." The sharper your target persona, the more useful every interview, and the loop is only as strong as its weakest stage.
• Past behavior beats future speculation. Ask what people have actually done, not what they think they might do, and never pitch when you should be listening.
• Behavior is the currency of truth. Existing workarounds and real commitments of time, money, or reputation outweigh any amount of polite enthusiasm.
• There is no magic interview count. Roughly 15 to 20 people per persona is a starting point; the real rule is to continue until patterns repeat and surprises stop.
• Discovery ends in a decision. Every round should conclude with a clear call to persevere, pivot, or keep digging — not just a folder of notes.

Frequently Asked Questions

What is Talking to Humans about?

Talking to Humans by Giff Constable, co-authored with Frank Rimalovski and featuring a foreword by Steve Blank, is a short, practical field guide to customer discovery. It teaches founders how to choose who to interview, find and recruit them, run effective conversations, and interpret what they learn — all to validate a business idea before over-investing in it.

How many customer discovery interviews do you need?

There is no statistically significant magic number. The book suggests roughly 15 to 20 people per persona as a starting point, but the real stopping rule is qualitative: keep interviewing until answers become predictable and clear patterns repeat. When you can guess what the next person will say, you have enough signal to make a decision for now.

What questions should you ask in a customer discovery interview?

Ask about real past behavior, not hypotheticals. Strong questions include "tell me about the last time you faced this problem," "what did you do about it," and "what did that cost you." Keep questions open-ended, avoid leading the person toward your idea, and never pitch — your goal is to hear their story, not to sell yours.

How is customer discovery different from a survey?

A survey collects opinions at scale; customer discovery uncovers behavior and motivation in depth. Interviews let you ask "why," follow unexpected threads, read emotion, and dig into stories a fixed questionnaire would miss. Surveys can help confirm patterns you have already found, but early discovery needs the flexible, qualitative conversation the book describes.

Should founders do customer discovery themselves or delegate it?

Founders should run discovery themselves. Talking to Humans frames it as a founder's job and a team sport — the raw, unfiltered signal from customers needs to reach the people making product and strategy decisions directly. Delegating it entirely or outsourcing it to a research firm strips out the nuance and conviction that make discovery worth doing.