Coherence Is Not Evidence
Why a business model that holds together can still be completely wrong
At this stage, you have a customer, a problem, and a clear sense of what you know and what you still need to learn. Our work now is to take all of that and shape the venture’s first real architecture: a single-page business model. The goal isn’t to plan the company in detail, but to lay out, in one place, your best current thinking about how the customer, the value, the operations, and the money fit together—making each assumption visible so you can go test it. The challenge is that a model can feel convincing long before it’s been checked against a real customer. By the end of this chapter, you’ll have a canvas you can stand behind, the discipline to keep it honest, and a set of questions to bring to the people who can answer them. Where a structured tool can help, you’ll find a note at the end of the section pointing you to it.
We left the Meridian Team at a pause, and they came back from it in better shape than they left.
The two weeks paid off. Three calls with compliance people in their network told them something they’d been guessing at, and guessing wrong: the decision about compliance tooling wasn’t one decision. Part of it sat with the compliance function and part with IT procurement, and the line between them fell in a different place at almost every company they looked at. They rewrote that section of the Opportunity Statement, the customer came into focus, and for the first time, the team wasn’t guessing about what mattered most.
So they did the next thing you’re supposed to do. They built the canvas, starting from the customer and the value proposition and working outward, through channels and relationships, the activities and resources the model would take, the partners, the rough shape of where the money would come from and go. Nine blocks, two evenings. They stood back, looked at the whole thing on one page, and it held together.
Which was the problem?
You could read it across, left to right, and it made sense the whole way, customer leading to value proposition leading to channels leading to the people they’d need. A clean story about how a problem and a solution fit, every block propping up the next. They took that as a sign they’d gotten it right.
They hadn’t tested a word of it.
This is the trap, and it tends to catch the careful teams more reliably than the careless ones. When something hangs together in your head, the mind quietly files it under true, no matter whether you’ve checked it against anything outside your own reasoning. A canvas that connects block to block gives you the genuine sensation of having worked something out. That sensation is exactly the problem. Finishing a canvas and testing one are not the same act, however much the first feels like the second.
What the Meridian Team had done was build a hypothesis and mistake it for a finding, which is close to the most ordinary thing a founding team can do. Fill in nine blocks, and you’ll feel it too, the small satisfaction of the pieces clicking into place, fit quietly standing in for fact. And the more finished the page looks, the less anyone wants to go hunting for the single piece of evidence that might force them to pull it apart.
So the stage is doing something other than what it appears to be doing. It isn’t building a business model. It’s building the first honest map of the bet you’re making, with every assumption underneath it set out where you can see it, which is worth a great deal more than a finished model would be this early. A finished model now would mostly be fiction. The honest map is the thing you can actually work from.
Why a Finished Canvas Feels Like a Finished Business
The Better It Looks, the Less You Should Trust It
The trouble at this stage is hardly ever a bad canvas. Bad ones get caught on their own. A block contradicts the block beside it, a number won’t add up, something obvious is just missing, and somebody on the team notices. It’s the good canvas that’s dangerous, the one where every block feeds the next, and the logic runs clean from customer to revenue, and the whole page reads like a company that already exists somewhere.
There’s a name for what happens next. When we build a story that holds together on its own terms, we tend to treat it as true, and not because we’ve tested any of it. We treat it as true because the holding-together is itself persuasive. Psychologists call it narrative coherence bias. We call it the coherence illusion, after what it does to a founding team: it makes a stack of connected guesses look like facts.
It tends to show up in three ways.
Completion feels like confirmation. Filling in all nine blocks is satisfying, and the satisfaction is earned; you’ve put in hours of real thinking, and the full page is the evidence of it. But having finished and being right are not the same thing, and finishing does nothing on its own to make you right. A canvas with nine blocks filled is no further along than one with four filled and five honest question marks. It only looks further along.
Internal logic stands in for outside evidence. When the customer segment leads sensibly to the value proposition, and that to the channels, and the channels onto the activities, it’s tempting to treat the whole chain as confirmed. All you’ve really confirmed is that your reasoning agrees with itself, which is a much smaller thing, and not at all the same as being accurate. You can build a perfectly logical argument on top of a wrong premise, and the logic will hold all the way down. Whether the premise was right is the one question only the customer can answer, and at this point, the customer hasn’t said a word.
The finished canvas defends itself. Once every block is full, the thing quietly changes jobs. It stops being a draft and becomes a position you hold. After that, any evidence that would force a block to change stops feeling like useful information and starts feeling like an inconvenience, a reason to look the other way, a threat to work you’ve already done. The more polished the canvas, the more there is to lose in reopening it. A canvas that took two evenings to build has no interest in being taken apart.
None of these feels like a mistake while you’re making it. Every one of them feels like care. That’s what makes the illusion so hard to catch in your own work.
You can build a perfectly logical argument on top of a wrong premise, and the logic will hold all the way down.
One thing holds up against all three of these, and it runs through every block on the page. Before you trust the line between any two blocks, trace it back to something the customer actually does. Not what they want, not what they tell you over a friendly coffee, not what your model assumes they’ll do once it exists. What they do now, in the real situation you’re designing for. That line running through the canvas is what we call the behavioral thread, and the place to pick it up is where the canvas starts, with the customer.
The Two Blocks That Anchor Everything Else
Get These Two Wrong and Every Other Block Inherits the Error
Nine blocks, and they don’t carry equal weight. Two of them hold up all the others: the customer segment and the value proposition. Everything else on the page is downstream of those two: the channels there to reach this customer, the activities there to deliver this value, the costs and revenue that arise from how the first two are resolved. Get the two right, and the canvas has a spine. Get them vague, and every block built on top inherits the vagueness, dressed up in the false confidence of a finished page.
They’re also where your earlier work comes to rest. The customer, the key behaviors, and the barriers you identified in the Opportunity Statement all carry straight into these two blocks. Sometimes the canvas version starts to drift, a slightly broader customer, a slightly softer barrier, a behavior that has quietly gone missing somewhere along the way. When that happens, the fix isn’t on the canvas. It’s a sign to go back, sort it out in the Opportunity Statement, and then come back. The canvas doesn’t get to overwrite the foundation; it only inherits from it.
Start with the customer segment, and fight the pull toward demographics. The block isn’t asking who your customer is in the census sense. It’s asking what situation they’re in and what they’re trying to get done inside it. A mid-market compliance officer drowning in regulatory updates across a dozen jurisdictions is a customer segment. “Companies with 200 to 2,000 employees” is a mailing list. Two things anchor the block, both come straight out of the Opportunity Statement: the key behaviors and the barriers. The key behaviors are the one to three things the customer actually has to do to get the outcome they’re after. The barriers are whatever is standing between them and doing those things today. On the Exploring track, you’re writing hypotheses, and you say so. On the Building track, you’re writing claims you can back up, behaviors you’ve watched real early customers perform, barriers that turned out to be real, and others that turned out not to matter at all. One more thing belongs here, particularly for the Exploring teams: whether the customer is already out looking. Someone who’s actively trying alternatives and coming away unsatisfied is a very different prospect from someone who has never thought to look. One of them feels the problem. The other may only have a problem you can see.
The value proposition block is where most of the trouble starts. Founders fill it with their solution almost on reflex, because the solution is the part they’re most excited about and most sure of. The trick is to describe the fit rather than the product. Three things go in here. First, the outcome the customer is after, written as a result they’d recognize as success and not as a feature or a capability. Second, how the solution takes down the specific barriers you named and makes the key behaviors possible, which is the single most important connection on the whole canvas, the one spot where the customer’s problem and your solution actually touch. Third, for the Exploring teams, what a solution would have to do to pull that off, written as requirements rather than features; for the Building teams, what it has already shown it can do, backed by what customers are telling you. Then one test for the block as a whole: does this solution reduce barriers and enable the key behaviors outlined in your Opportunity Statement? If you can’t answer that cleanly, what you have isn’t a value proposition. It’s a description of your product.
The Meridian Team had learned this once already, and here they learned it again. Their customer block was genuinely strong; they’d reframed the compliance officer’s situation with real care, and the key behaviors came across cleanly from the revised Opportunity Statement. Then they wrote the value proposition, read it back, and found they had described their machine learning architecture. Again. The block listed what the model could do, the automated monitoring, the cross-jurisdiction mapping, the real-time alerts, and said next to nothing about the barriers that all that was supposed to remove. Priya was the one who caught it. The barrier they’d named wasn’t a shortage of monitoring tools. It was that the sheer volume and speed of regulatory change had outpaced the manual process these officers were still leaning on, and that nothing on the market covered enough jurisdictions to be trusted as the single source of truth. A value proposition built on that would read more like: one person can stay on top of the changes across every jurisdiction they’re responsible for, without dropping back to the manual review they no longer have time for. What the team had actually written described the product and called it a value proposition. The coherence of everything around it kept the gap hidden, right up until somebody read the block out loud against the customer’s real problem.
If you cannot answer it cleanly, you do not yet have a value proposition. You have a product description.
That’s the payoff of anchoring here. These first two blocks are where you find out whether the rest of the canvas stands in the customer’s world or only in yours.
→ Supported by the Behavioral Business Model Canvas worksheet and guide · New Venture Innovation Process
Where the Thread Becomes Real Work
Five Blocks That Earn Their Place Only by Connecting to a Behavior or a Gap
The middle of the canvas is where many teams start to relax. They often believe the hard thinking happened in the customer and value blocks, and that operations, relationships, channels, activities, resources, and partners are just logistics—fill in the obvious answers and move on. But that assumption is exactly what breaks the behavioral thread.
These blocks are not a checklist. Each one answers a single question: given the behaviors your customer needs to perform and the barriers in their way, what does it actually take to make those behaviors possible? Every entry here should connect directly to a specific customer behavior or a gap your team needs to close. Anything that doesn’t is just filler—making the canvas look more finished, but not making your venture any more likely to succeed.
Take channels first, since that’s where the mistake shows up most. A channel isn’t a place you set your product down. It’s the route a customer travels to perform a behavior. So the question isn’t “where can we reach people,” it’s “which channel makes the key behavior easiest to perform,” and those two questions don’t return the same answer. A channel that gets you in front of the customer but makes the critical behavior harder is worse than having no channel at all. Relationships work on the same logic. The relationship is the context the customer is in when they use your solution, and the real question is what kind of relationship the behavior change actually needs, because a customer breaking a long-entrenched habit needs something quite different from one just swapping one familiar tool for another. Explorers put down their best guesses and mark them untested. Builders write what they can already see happening, which relationships and channels move the behavior along, and which ones quietly get in its way.
The activities and resources blocks ask what your team needs to know and what you need to have. Focus on that word: know. This isn’t just an operational to-do list. It’s about the knowledge the opportunity demands—a deep understanding of the problem, the expertise you can’t do without, and an honest look at where your understanding is still thin. This is where all the knowledge and access work from the first stage finally lands. If you gave yourself a low mark earlier on understanding the customer’s world, don’t file it away and forget it. That gap becomes both a key activity and a key resource. The activity is to close the gap; the resource is whatever helps you do it, whether that’s research, an advisor, or a data source. The resources block is where the mentors and advisors you listed earlier become specific: this person, for this gap.
That leads directly to partners. The partner block formalizes two things you’ve already started: the ecosystem players you mapped and the advisors you named. This block only earns its place when each entry points to a specific gap it helps close. A partner who gives you distribution through a channel your customer actually uses belongs here. So does a partner who fills a domain gap your team has recognized. But an impressive logo with no clear purpose is just the canvas flattering itself. The test is simple: for each partner, what exact gap does this one close? If you can’t answer, it’s superficial.
Fatima hit this head-on. She’d left a regional consulting partnership in the Gulf after fifteen years in organizational transformation, so her command of the domain was never the issue. Her customer and value blocks were among the sharpest we’d seen; she understood her customers’ problems better than most founders understand much of anything. And then she got to the relationships and channels blocks, and the canvas stopped cooperating with her.
Her whole model leaned on a single behavior: a senior executive agreeing to sit down for an exploratory conversation about a problem before they’d decided to spend a cent solving it. Her honest self-assessment in the first stage had already revealed an uncomfortable fact. Every executive relationship she had was institutional, built through the firm, kept alive inside paid engagements, run on the firm’s terms. When she walked out, most of those relationships stayed behind. So she had plenty of access to people who remembered her warmly, and almost none to the kind of off-the-clock buyer conversation her key behavior actually called for. The relationship she had wasn’t the relationship the behavior needed.
On a canvas built to make her feel good, that gap would have disappeared. Her network looked formidable. She could have filled the relationships block with confident lines about her access to senior executives, and the coherence of everything around it would have waved them through. What stopped that was tracing the block back to the actual behavior. The relationship she needed didn’t yet exist, and saying so plainly turned her partners and resources blocks into something genuinely useful, a deliberate plan to build the access she was missing: advisors, warm introductions from outside the old institutional frame, a run of low-stakes conversations engineered to produce exactly the relationship her model depended on. The weakest corner of her canvas became the most actionable part of it.
The Blocks You Never Think to Question
Why Cost and Revenue Are Where Confidence Does the Most Damage
At this stage, the cost and revenue blocks aren’t a financial model, and nobody should expect hard numbers from a venture that hasn’t tested its core assumptions. A canvas that hands you precise figures this early is usually just making them up. What these two blocks want from you is a hypothesis about shape. Where is the real money going to go, and which activities and resources are going to eat the most of it? Where will it come from, and which revenue model best fits the way this customer buys?
The word doing the work in that revenue block is buys, not “would pay.” Behavioral scientists have spent decades measuring the gap between what people say they’d pay and what they hand over when the moment comes, and a revenue model based on stated willingness sits on the wrong side of that distance. The questions worth asking are behavioral. What are these customers already paying for in the next-best solutions they use? In their experience, when does money actually change hands, and does that moment line up with when the problem hurts most? A revenue model is a claim about how customers behave, and it deserves to be tested like one.
There’s a reason this is hard to get right, and it has nothing to do with finance. We don’t examine all of our beliefs with the same rigor. The claims that cut against what we already think get cross-examined; the ones that agree with us tend to walk straight through. On the canvas that shows up as founders pouring energy into the blocks they’re least sure of and breezing past the ones they’re most confident in, which is unfortunate, because confidence is precisely where the expensive wrong assumptions like to hide. The customer segment that the domain expert is sure they already understand. The value proposition the founder has fallen for. And, most dependably of all, the revenue model someone imported wholesale from the last industry they worked in. It feels familiar, so it never gets questioned, and familiar and correct are not the same thing.
A revenue model is a claim about customer behavior, and it should be tested like one.
The defense runs against instinct: put your hardest scrutiny on the blocks that feel the most settled. Those critical assumptions you wrote out at the end of the first stage exist for exactly this. They make you put on paper the things you’re surest of, because certainty is the place you’ve stopped looking. The block you want to skip because it’s so obvious is usually the one to slow down on.
Sung-jin had more reason than most to feel settled. He’d spent years running new-product evaluation inside a South Korean consumer electronics company and came to the canvas holding the kind of internal data most founders never get their hands on. His cost and revenue blocks came easily, too easily, and he filled them the way his company had always filled them: a one-time hardware sale through retail channels, manufacturing, and distribution, with the high costs. It was the model he knew cold, and it was wrong in a way nothing else on his clean, confident canvas was ever going to flag for him.
His venture was a smart-home health monitoring product, and his real customer wasn’t the company’s usual buyer. More often than not, it was an adult child seeking a little reassurance about an aging parent living alone. That person doesn’t behave like someone buying a television. They check in, month after month, year after year, because the worry never entirely lifts. The revenue model Sung-jin had carried over was built for a completely different behavior than the one his customer would actually perform, and his cost block had gone wrong in the same direction for the same reason. He’d budgeted like a hardware company when the thing he was really building ran on ongoing costs, the monitoring and the reassurance delivered over time.
The part worth sitting with is that these were his most confident blocks. He’d gone over his customer segment with care, because he knew that was where his internal data and reality had started to part ways. The revenue model he never thought to question at all, because it was the one piece he was certain he already had right. Which is exactly where assumption blindness lives, in the blocks that don’t worry you in the slightest.
The Check That Breaks the Illusion
Trace the Thread End to End, and the Gaps Reveal Themselves
Everything in this chapter has been pointing to a single test. The coherence illusion makes a finished canvas feel validated. Assumption blindness leaves the most confident blocks unexamined. Both of them run on the same move, letting you check the canvas against nothing but its own internal logic. The behavioral thread check does the reverse. It makes you check the canvas against one thing sitting outside it, what the customer actually has to do, and that’s the one thing internal coherence can never fake.
Here’s the thread. It begins in the customer block, with the key behaviors, the specific things the customer has to do to get the outcome they want. It runs into the value proposition through barrier mitigation, the way your solution clears what’s in the way and makes those behaviors possible. From there, it carries into activities and resources, what your team has to know and do to actually clear it. And it lands in channels and relationships, how the customer reaches the solution and performs the behavior in the middle of their real life. Behavior, to barrier removal, to capability, to access. Four hand-offs, one continuous line.
The test is whether you can follow that line end-to-end without it snapping. It isn’t whether each block is plausible on its own; it’s whether each one connects to the next through the customer’s behavior. And a break anywhere is informative. It’s worth knowing exactly what each kind of break means.
Can’t connect the key behaviors to the barrier mitigation? Then your value proposition isn’t addressing the real barrier. It may well be addressing a barrier, just not the one standing between this customer and the behavior they have to perform. Can’t connect the barrier mitigation to your activities and resources? Then your operations aren’t built around the customer’s need; they’re built around something else, usually the product you wanted to make. Can’t connect your activities to your channels? Then you’ve built a solution that the customer can’t actually reach along the path their behavior travels. Each break points to its own particular fault, and not one of them shows up if you only ever read the canvas for internal consistency. They appear when you trace against behavior, and rarely any sooner.
A canvas you cannot trace this way is not a business model hypothesis. It is a list of assumptions that happen to sit close together on a page.
When the Meridian Team finally ran the trace, the value-proposition break they’d already caught turned out not to be the only one. The key behavior was clear enough: a compliance officer staying on top of regulatory changes across jurisdictions without dropping back to manual review, and once they’d corrected the barrier mitigation, it connected to that cleanly. The thread broke instead at the channels. They had assumed they’d reach compliance officers the way their competitors did, through a long, IT-led enterprise procurement cycle, and traced against the actual behavior that assumption came apart. The behavior they needed to support was a compliance officer staying on top of the daily workflow. A procurement cycle puts the tool in front of IT months before it ever reaches the person who’ll use it, which doesn’t touch that behavior at all. The channel was perfectly plausible. It was even what everyone else in the market was doing. It just didn’t connect to the thread, and tracing the thread was the only thing that ever showed them so.
One last thing about the thread, since it isn’t something you check once and file away. The pieces you started with, the customer, the behaviors, the barriers, the enablers, all keep shifting as you learn, and what we ask teams to do is come back to them at every stage and say plainly what has held up, what has been refined, and what has changed outright. That running record is worth more than it looks. A team that can tell you exactly which of its early assumptions survived, which got sharper, and which were just wrong has something most ventures never manage to produce: an honest account of how its read on the customer changed over time. A thread you keep tracing is a thread you can actually trust.
What the Canvas Actually Gives You
You Came for a Business Model. What You Leave With Is a Discovery Agenda
Let’s be clear about what you actually take away from this stage, because it’s rarely what most teams expect. You haven’t produced a business model. Instead, you’ve created the first complete picture of everything that would have to be true for your model to work—every assumption laid out where you can finally see it. That’s more valuable right now than a finished model, because a finished model this early is mostly fiction. A well-built canvas, on the other hand, is something you can actually act on.
What you get from it comes in roughly three parts. It shows you where your earlier framing still holds, the blocks that line up cleanly with the Opportunity Statement, and that your first read of the customer was sound. It shows you where that framing needs work, the blocks that contradict the Opportunity Statement, and flags an assumption you got wrong, one to fix at the foundation before you go any further. And it finally gives every gap you’ve been carrying around a fixed address. The thin patches in your team’s knowledge, the access you don’t have yet, the expertise you’re short on, none of these stay vague background worries anymore. Each becomes a named line in the resources or partners block, with a plan hanging off it.
But the most valuable thing the canvas gives you is also the one we most often have to highlight, because it’s easy to miss. Every hypothesis on the page is a question waiting for a real customer. The canvas isn’t a set of answers. It’s a structured conversation with your own assumptions, and the next step is to bring that conversation to the only people who can truly settle it.
This is where the canvas becomes your agenda. The customer block becomes a question: who truly has this problem, and what are they doing about it now? The barrier mitigation becomes another: do these barriers actually stop the behaviors we named, or did we imagine their weight? For relationships and channels: how do these customers handle the solutions they already use? For knowledge gaps in your activities: what will customers tell us that changes how we understand the problem? The assumptions you marked as most critical and least certain aren’t loose ends—they’re your interview agenda for the discovery work ahead, written before you even realized it.
Renée came to the end of her canvas uneasy, like someone who’d expected to feel a lot more finished than she did. Twenty years of building and running leadership development inside the military. A program has already been worked out in detail. A deep conviction about the people she meant to serve, mid-career professionals making the hard move out of military service and into civilian careers. The canvas was coherent. Every block connects to the next. And yet when she traced the thread and read the thing honestly, what she was holding wasn’t a model she could start building. It was a list of things she didn’t actually know about a customer she had never once sold to in an open market.
She knew the problem from the inside; she’d lived her own version of the transition. What she didn’t know was whether the people she wanted to serve felt it the way she had. Whether they were already out looking for help, or quietly assuming they’d sort it out on their own. Whether the behavior her whole program rested on was one these customers would really perform, or just one she’d assumed they’d want because they so plainly needed it. And it was a real commitment she was asking for: a mid-career professional putting months into structured readiness work, on their own time and their own money. Needing something and being willing to pay for it are not the same. The space between the two is exactly what the canvas exposes, and only customer discovery can close it.
Her canvas didn’t tell her she was right, and it didn’t tell her she was wrong. What it did was tell her, exactly, what she now had to go and discover. At this stage, that’s the most a canvas can do for you, and it’s a great deal.
What You Take to the Customer
This stage doesn’t hand you a business model. It hands you the first full architecture of one, a set of connected hypotheses about how the customer, the value, the operations, and the money fit together, drawn carefully enough that every assumption inside is out in the open. The job was never to make that architecture feel finished. It was to make it honest and testable.
That’s the line between a canvas that helps you and one that quietly leads you astray. A coherent canvas feels like proof. It isn’t. It is the clearest statement you can make of what you believe and haven’t yet checked, and stating it that clearly is what tells you precisely where to go look. Everything after this is built on it. Customer discovery tests the behaviors and barriers you guessed at. The market work defines the landscape that your channels and partners are assuming. The product decisions follow from what customers tell you about the behaviors you named right here. Leave the canvas vague, and everything downstream inherits the vagueness; make it honest, and everything downstream has solid ground to start from.
The Meridian Team, Fatima, Sung-jin, Renée, and Amara all came out of this stage with a canvas that looked complete and an unsentimental list of the places it wasn’t yet true. Not one of them took the page’s coherence as confirmation from the customer. That restraint, building the whole architecture, and still treating the whole thing as a question, is the discipline this stage exists to teach.
A coherent canvas feels like proof. It is not.
A complete canvas still isn’t a validated one. The only thing that validates it is the customer, and the customer is where the next stage of the work begins.
Build the architecture. Then go find out whether it holds.
© 2026 Venture for All® · Learn · Innovate · Thrive