Letting the Customer Prove You Wrong
Why good discovery is designed to disagree with you.
At this stage, you have a specific customer in mind, a one-page model, and a list of untested beliefs. Now, we take that list to the people who can confirm or challenge it. This is not open-ended exploration. It is structured hypothesis testing. Before asking questions, we write down our assumptions, rank them by the risk they pose if wrong, and decide what evidence would disprove each one. The challenge is that customers do not always act as they say, and well-meaning people often tell us what we hope to hear. If we design discovery to collect agreement, we gather the wrong evidence and do so with confidence. By the end of this chapter, you will know which beliefs held up with real customers, which ones did not, and whether the need you identified is matched by anyone ready to act. Where there are structured tools to help, you will find a note pointing to them.
The Meridian Team had moved beyond the business-model stage. They carried forward a set of hypotheses, and one finding they had worked hard to earn: the decision to buy compliance tooling was split between the compliance function and IT procurement, and the line between the two sat in a different place at almost every company. What they had not yet tested was the risk hiding inside that split. Could compliance, the people they could reach most easily, actually move a purchase, or would IT and procurement decide it? Five colleagues had worked together to design a compliance platform for mid-market supply chains. Their model now fits clearly on a single page. At this stage, their main task was to test this with real customers.
The first conversations went well. The team met with compliance leads and operations managers at the types of firms they hoped to serve. They described the problem and shared their vision for a solution. The people across the table agreed. They said the problem was real and their current tools were a mess. One person said it sounded like exactly what her team had been waiting for.
The Meridian Team returned energized and ready to validate the opportunity. On the surface, it seemed they had succeeded.
In reality, they collected agreement, but agreement alone does not confirm a real opportunity. The people they spoke with were supportive and recognized the problem, which was never in doubt. The more important questions were left unasked. Would these firms actually change their behavior? Who would approve that change? What would it take to make it happen? This is a good moment for us to pause and reflect on what we might be missing.
Most importantly, that risk was never tested in any conversation. The team knew the decision was shared, but they chose to speak with compliance leads, who were easiest to reach. They let those friendly conversations take the place of the harder question: who can actually move a purchase? When they asked, ‘Compliance would own this, right?’ the compliance leads agreed. It is easy to accept an answer that matches our perspective, and even easier when it matches what we hope to hear. This reminds us to dig deeper, even when the answer seems clear.
In the end, the team spent a week confirming what they already believed instead of testing those beliefs. The real purpose of this stage is to challenge assumptions. This is a common pitfall, and one we can learn from as we move forward.
Why Discovery Goes Wrong
How Careful Teams Talk Themselves Into the Wrong Answer
What happened to the Meridian Team is common. It rarely comes from lack of effort. Most teams who lose their way in discovery are working hard and meeting real customers. The problem is in what they listen to and what they accept as proof. Three patterns explain most of these missteps.
The say/do gap. What people tell us and what people do are not the same. We met this in the first stage as the intention-behavior gap. Here it shows up in the room. Someone can describe a frustration in vivid detail and then go right back to the workaround they have used for years. They are not lying. They simply do not know their own future behavior as well as they think they do. When we ask, “Would you use something like this?” we are asking them to predict a future self. That self is optimistic, agreeable, and unreliable. So we ask about the past instead. What have you done about this problem? What are you doing now? What did you try the last time it came up? Behavior that has already happened is the only behavior we can trust.
The comfort of agreement. People want to be kind, especially those closest to us. A customer who likes us or sees our commitment to the idea will often encourage us. Agreement can feel like validation, but most of the time, it is simply politeness. Earlier, we warned about noise that feels like a signal. This is that same noise, just in a friendlier form. The warmer a conversation feels, the more carefully we should read it.
Hearing what we came to hear. We enter these conversations hoping our idea will work. That hope is natural, but it quietly shapes our questions. We ask in ways that can only return a yes, and then we record the yes as evidence. Discovery done this way does not test anything. It only seeks permission. The purpose of this stage is the opposite. We are here to find out where we are wrong, while mistakes are still inexpensive.
We are here to find out where we are wrong, while mistakes are still inexpensive.
None of these patterns mean our customers are difficult or dishonest. They show us that we must design the conversation with care, because the easy path will mislead us every time. That care begins with knowing exactly who we are talking to.
Knowing Exactly Who, and What Their Day Looks Like
A Persona Means Little Until You Follow It Through a Real Day
Before we can test an assumption about a customer, we need to be specific about who that customer is. Vague customers give vague answers. ‘Small business owners,’ ‘busy parents,’ or ‘compliance teams’ are not people we can interview. They are categories, not individuals. Our task at this stage is to turn a category into a single, recognizable person and then follow that person through a typical day.
Two tools help us with this. The first is a customer profile, which makes us commit to the details. Who is this person? What are they trying to accomplish? What gets in their way? The second is a map of their experience, which takes that person through the problem step by step. The profile tells us who. The map shows us what their day looks like. We need both, because the barriers we care about do not live in the profile. They show up in the day itself.
This is where the process becomes most useful, and where the gap between what people say and do becomes clear.
Amara is building a mobile tool that puts real-time clinical guidance in the hands of community health workers in rural districts of Ghana. When she started, she had a clear picture of her customer in mind. This worker would pull up the guidance on a phone during a patient visit, check it against what she was seeing, and decide with more confidence. That picture felt right. It was also the worker Amara hoped to find. The one in front of her was someone else.
So she did two things. She wrote down what she actually knew about these workers, and she mapped a real day from the moment they set out to the moment they finished. The day looked nothing like her picture. The workers covered long distances on foot. They saw more patients than the schedule allowed. Power was unreliable, phones were often shared, and there was no unhurried moment during a visit to stop and look something up.
Then she asked them how they made the hard calls. They told her they would check the guidance whenever they were unsure. When she walked them through specific recent cases, a different answer emerged. In the moment, with a patient in front of them, they decided from memory and experience and kept moving. They looked things up later, if at all. They were not being dishonest. They were describing the version of their day they believed in, the way many of us would.
The barrier Amara thought she was solving was the absence of good guidance in the field. The barrier that shaped behavior was the absence of any moment during the visit when checking a phone was possible. The map showed her that. The profile alone never would have.
A persona is only as good as the real day it represents.
This is the lesson to carry forward. A persona is only as good as the real day it represents. When we ground the person in a mapped experience, the barriers we need to design around become visible instead of remaining guesses.
→ Supported by the Customer Profile and Customer Experience Map worksheets and guides · New Venture Innovation Process
Deciding What Would Prove You Wrong
The Step That Turns a Friendly Chat Into a Real Test
Most mistakes in discovery happen before the first conversation. They come from what we choose to ask and what we hope to hear. If we want the conversation to test our thinking instead of flatter it, we need to prepare in advance. That preparation has three steps.
First, we write down our assumptions. Every venture is built on beliefs that feel like facts. For example: the customer has this problem, they will change their behavior to solve it, or this person controls the money. We pull these beliefs from the Opportunity Statement, the model, and the experience map, and put them in plain view. If we have not named a belief, we cannot test it.
This is where the experience map proves its value. The friction it reveals, the point where the customer gets stuck, is not just background detail. It is the assumption we most need to test. A common mistake is to list every belief from the model and ignore the friction the map uncovered. The model feels comfortable. The friction is more challenging. If the map identifies a barrier, we need to include it on our list.
Second, we rank our assumptions by risk. Not every assumption matters equally. Some can be wrong without much cost. Others, if wrong, could end the venture. Which belief, if false, would hurt us most? That is where we focus our questions.
There is usually a clear answer, and it is usually the one we are tempted to skip. Every venture rests on a single central claim: that removing this particular barrier will change what the customer does. Teams tend to test the beliefs around that claim, the ones about price, or trust, or who signs off, and never put the claim itself at risk. Those edge beliefs matter. But a venture that confirms everything except its own foundation has not tested the thing that matters most. The central claim goes on the list first.
Third, and this is the step most teams skip, we decide what would prove each belief wrong before we talk to anyone. What would we need to hear or see to admit we were mistaken? We write that down first. Then we build questions that could lead to that answer. A question that can only return a yes is not a test. It is a wish.
A question that can only return a yes is not a test. It is a wish.
The Meridian Team learned this the second time around. The risk they had walked past was not whether the decision was split. They already knew it was. It was whether compliance, the part of that split they could reach, could move a purchase on its own. Their whole plan to enter through compliance depended on the answer. So they treated it as their highest risk and asked themselves a hard question. What would tell us we are wrong about this?
Their answer was clear once they wrote it down. If, in past purchases of tools like theirs, compliance had not been able to move the decision without IT or procurement leading it, their entry plan was in trouble. So they stopped asking, ‘Compliance would own this, right?’ Instead, they asked people to walk them through real history. Tell us about the last tool your team brought in. Who first pushed for it? Who signed off? Who could have stopped it? The stories that came back were sobering. In most firms, compliance could raise the need and champion a tool, but the purchase did not move until IT or a procurement committee carried it out. Compliance had a voice. It rarely had the final say.
That was uncomfortable, and it was the most valuable hour of the project. An entry plan built on compliance alone proved shaky, and they found that a change was still cheap to make.
This is the heart of the stage. When we decide in advance what would prove us wrong, we turn a friendly conversation into a real test. We also remove the easy yes that quiet confirmation seeks. The goal is not to defend our assumptions. It is to find the weak ones before they cost us anything.
→ Supported by the Assumption Testing Matrix worksheet and guide · New Venture Innovation Process
Questions That Test, and People Who Can Disconfirm
Who You Ask Matters as Much as What You Ask
Once we know what would prove us wrong, three things determine whether a conversation is valuable. The questions we ask. The people we ask. And how we behave in the room. These are often treated as separate skills, but they are one discipline. A weakness in any of them can undo the other two.
The questions come first, and we have already seen their form. We ask about the past, not the future. We ask for specific recent events, not general habits. A good question invites a story. A weak question invites an opinion, and opinions are easy to give but rarely useful.
The questions that mislead us often sound rigorous. For example, “What would have to be true for you to switch?” or “What would your ideal version look like?” seem thorough. In reality, they ask the customer to imagine a future that does not exist. The answers are guesses, not evidence. There is a simple test. If a question asks someone to imagine instead of recall, rewrite it. Ask what they did the last time they faced the problem.
Who we ask matters just as much. We set a rule for ourselves: the people we interview must match the profile and persona we created. If we talk only to those who are easiest to reach, we have not tested anything about the customer we set out to learn about. The candidate list is the persona made real. Every name on it should resemble the person we described.
Matching the persona is necessary, but not enough. We also have to include the people who can say no. Sung-jin learned this the hard way. His venture was a home health monitoring service for adult children keeping an eye on an aging parent who lived alone, and he already had a base of early customers paying month-to-month. When it came time to interview, he went to the customers he had. They were glad to talk, and they had good things to say.
The problem was that those customers had already chosen to stay. They could explain why the service was worth keeping, but not why others had signed up and then canceled, or never started at all. So he made a harder list for himself. He reached out to customers who had left after a few months, and to people who had considered the service but chose something else or nothing. These conversations were harder to arrange and harder to hear. They were the only ones who could explain what kept people from staying. The people most likely to tell us no are often the ones we need to hear from most.
There is a quieter lesson in the act of reaching out, and it is worth noting as we go. How long did it take this person to reply? How long from the first message to a meeting on the calendar? Did they ask to stay in touch and follow the venture’s progress? None of this is small talk. It is our first honest look at the sales cycle. A customer who replies quickly and wants to keep talking behaves very differently from one who takes weeks and never follows up. When we later figure out how the venture finds and wins customers, this early record of who responded, how fast, and how warmly will be our first real data. The people who lean in now are often the first to buy.
Then there is the room itself. Fatima had spent a long time rebuilding access to senior executives, and that access was paying off. They took her calls. They were warm and generous with their time. Nearly every one of them told her the idea was excellent.
That warmth was the danger. Senior people who like us are not trying to mislead us. They are being gracious, and graciousness sounds a great deal like evidence. So Fatima changed how she ran the conversation. She stopped describing her solution at the top because a pitch only teaches people what to praise. She asked instead about what they had already done and paid for. In one call, an executive called the idea a clear winner. Fatima asked when he had last paid for something that solved this problem. He paused, then admitted he never had. The enthusiasm had not survived a single question about his own behavior.
The moment we start selling, we stop learning.
How we behave in the room matters as much as who is in it. We keep our solution out of the conversation for as long as possible. We do not signal the answer we hope for. We let silence sit until the other person fills it with something real. The moment we start selling, we stop learning.
Treat all of this as one discipline. Choose the right people, drawn from the persona and including those who might disagree. Ask questions about the past. Create a neutral room. Keep a record of how hard each person was to reach. If we skip any part, the conversation drifts back toward the comfortable yes we are trying to avoid.
→ Supported by the Interview Preparation and Interview Candidates worksheets and guides · New Venture Innovation Process
What Discovery Tells You
Sorting What Held From What Broke Is the Real Work
A stack of interview notes is not learning. It is just raw material. Learning happens when we compare what we heard with what we believed at the start. Every assumption we wrote down before the conversations gets one of three marks: it held, it broke, or it needs to change shape. Sorting these results is the real output of this stage, and it is harder and more honest work than the interviews themselves.
A held assumption is a small gift. We can build on it with more confidence. A broken assumption is even more valuable, because it saves us from building on sand. The assumptions that need reshaping, where we were half right, are often where our sharpest thinking happens. None of these results is a failure. The only failure is finishing discovery and believing exactly what we believed before we started.
Each result also points us in a direction. A broken assumption about the customer sends us back to the persona. A broken assumption about the problem sends us back to the Opportunity Statement we wrote at the start. A broken assumption about how the venture makes money sends us back to the model. Discovery is not a step we finish and leave behind. It reaches back and rewrites earlier work, and the behavioral thread we started with runs through all of it.
Renée felt the full weight of this. She had spent 20 years in military leadership development, and her venture was designed for mid-career professionals transitioning from military service to civilian careers. She had lived her own version of that transition, and the people she interviewed confirmed that the problem was real. They knew it well. They could describe it in detail. For a while, that felt like validation.
Then she asked the harder questions. Were they out looking for help with this, or assuming they would sort it out on their own? Had they tried anything? Would they put their own time and money into structured support? The answers were quieter. Most of them had made their peace with the transition. They expected to manage it themselves, the way they had managed hard things before. The need was real, and the demand was not. People can carry a problem for years without ever looking for a cure.
That finding sent Renée back to the Opportunity Statement. The customer she had described was not actively searching for anything, and her whole framing had assumed they were. She did not abandon the venture. She went looking for the people who were searching, the ones for whom the transition had crossed from something to endure into something they were ready to invest in. That was a revision, and it was the most important thing she learned.
The only failure is finishing discovery and believing exactly what we believed before we started.
One more practical note about this stage. The summary that brings everything together does not arrive on the day the rest of the work is due. It cannot. Interviews take weeks to schedule and complete. People reschedule. A pattern that looks clear after three conversations changes after the next five. So we capture findings as they come in, conversation by conversation, and let the full picture form in its own time. A discovery summary written too early is just another set of assumptions in a new form. The honest version waits until the conversations are done.
→ Supported by the Interview Findings and Discovery Summary worksheets and guides · New Venture Innovation Process
What We Carry Into the Market
Discovery does not end with a verdict. It ends with a clearer and more humble picture of the customer than we started with, and a short list of beliefs we can stand behind because we tried to break them and could not.
Look at where the conversations left everyone. The Meridian Team walked in believing compliance could carry a purchase on its own, and walked out knowing that in most firms, it cannot move without IT or procurement. Amara stopped designing for the health worker she hoped to find and started designing for the one whose real day left no moment inside a visit to stop and check anything. Sung-jin learned more from the customers who had canceled than from the ones who stayed. Fatima watched a warm endorsement collapse the instant she asked what the executive had ever paid for. And Renée found a real problem with no real demand and went looking for the people who were already searching for it.
None of them ended where they began. That is the point. Each person chose a little discomfort early to avoid much greater expense later. Each one finished with something sturdier than an opinion.
What they carry into the next stage is a customer they understand and a demand they have started to confirm. The work ahead asks a different question. We now have some confidence that someone will act. What we do not yet know is how many people there are, where they are, and how much of that market the venture can realistically reach. We cannot size a market honestly until we know whose market it is. Discovery earns us the right to ask how big it could be.
We leave this stage with fewer illusions and better questions. The goal was never to be right. It was to find out in time.
© 2026 Venture for All® · Learn · Innovate · Thrive