I have spent a lot of hours over the years sitting in rooms with co-founders, incubator cohorts, and early stage teams working through TAM, SAM and SOM. I have built the slides. I have argued about the wedges. I have watched other people do the same exercise more times than I can count. And the thing I have come to believe, after enough of these rooms, is that the exercise almost never predicted what actually happened to me or to the founders I was sitting next to. The numbers on the slide and the numbers in the bank account were operating on completely different physics. I kept doing the exercise anyway, because it was the exercise everyone agreed you were supposed to do. But it was not telling me anything I could use.
It took me a long time to articulate why, and the explanation that finally made sense to me is not really about TAM SAM SOM as a tool. The tool is fine for what it is. The problem is that it is answering a question that does not belong to the founder. It belongs to the investor. And those two people, despite the language pretending otherwise, are running entirely different experiments with entirely different stakes.
Two People At The Same Table
An investor is allocating capital across a portfolio. The mathematics of that allocation rewards them for finding outliers, because the returns from a single outlier can pay for every other bet in the portfolio that goes to zero. The TAM SAM SOM exercise is genuinely useful to an investor, because it tells them whether the upside, if it arrives, is large enough to make the bet worth making at all. They do not need the upside to be likely. They need it to be possible and large.
A founder is not allocating capital across a portfolio. A founder has one bet, which is their own time, energy, savings, relationships, and the years of their life that they will not get back. They do not get to diversify across twenty companies. They get one. The mathematics of that bet does not reward outlier hunting in the same way, because the cost of being wrong is not a line item in a fund report. It is a real person, in their late thirties, looking at a depleted bank account and a CV with a hole in it.
This is the part that the investor framing quietly glosses over. The slide deck pretends both people at the table are doing the same exercise, when in fact one of them is running a portfolio and the other is staking their life. The reason TAM SAM SOM never paid off for me as a predictive tool is that it was never designed to predict the founder’s outcome. It was designed to predict whether a bet on the founder was worth making for someone whose downside looks nothing like the founder’s downside. I was using an instrument calibrated for someone else and wondering why the readings did not match my reality.
What The Numbers Actually Look Like
Before going further it is worth being concrete about what the base rate actually is, because the conversation about validation tends to happen as if every idea is on the path to a Series A and the only question is how big the eventual exit will be. The data does not support that picture at all.
Around 0.05% of startups raise venture capital in the first place. The other 99.95% are funded by savings, credit cards, family, bank loans, or revenue. Roughly 78% of startups are self-funded by their founders. Of the tiny minority who do raise venture capital at the seed stage, CB Insights tracked over a thousand US tech companies through their entire funding lifecycle and found that around 67% stalled before any exit, and only about 1% reached a billion dollar outcome. The US Bureau of Labor Statistics, which is the cleanest base rate available because it does not select on funding status, shows that just over half of all small businesses close by their fifth year and around 65% by their tenth.
Stack those numbers honestly and the modal outcome for a founder is not a venture backed company that fails to exit. It is not even a venture backed company. It is a self-funded business that operates for some number of years, generates some amount of revenue that may or may not cover the founder’s costs, and eventually winds down quietly without anyone writing a post mortem about it. CB Insights have a useful word for the in-between version of this, which is the “zombie”: a company that is alive on paper, technically operating, but stagnant in a way that the formal failure statistics do not capture.
This is the outcome the validation conversation should be having, because it is the outcome the validation conversation almost never has. Every TAM SAM SOM slide I have ever seen was implicitly modelling the 0.05% case. None of them were modelling the 99% case, which is the case the founder was overwhelmingly likely to actually live through.
A Methodological Swap, And Why It Is Not One
I want to be careful here, because there is a sharp version of this critique that catches a sharp version of the response. The sharp critique is that everything I am about to propose is also speculation. Any working backwards exercise from a personal floor involves assumptions about pricing, about acquisition, about how long it takes to find the first customer. None of those numbers are knowable in advance. So if the objection to TAM SAM SOM is that it is built on guesses, the proposal that replaces it is also built on guesses, and the founder has not gained anything except a different set of slides.
The sharp response to that is that not all speculation is the same. The two exercises differ in what they are anchored to. TAM SAM SOM is anchored to a ceiling that only exists if the company reaches scale, which in turn only happens to a tiny minority of companies, almost all of whom needed external capital to get there. Every assumption in the chain is load-bearing for a future that statistically does not arrive. If the future does not arrive, the entire model evaluates to nothing, because the model was never about any of the years before the future.
The floor exercise is anchored to a number that exists right now, in the actual world, attached to a person who actually exists. The floor is the founder’s real monthly cost of living, or the real income they would need to walk away from a job, or the real runway their savings represent before things get uncomfortable. These are not forecasts. They are the present. The speculation built on top of them is about whether a plausible business can plausibly reach a number that already exists, rather than whether a plausible business can plausibly reach a number that only matters in a future the founder is statistically unlikely to inhabit.
That is the distinction I want to name explicitly, because without it the post sounds like a methodological swap and it is not. It is a swap of which kind of guess deserves to govern the decision. One kind of guess answers a question about a real person in the present. The other answers a question about a hypothetical version of that person in a future that overwhelmingly does not happen. They are not interchangeable.
The Two Numbers That Are Not Speculation
The part of the exercise that matters most is also the part that involves no guessing at all. It is two numbers that the founder already knows, or can find out in an honest hour with a spreadsheet, about their own life as it currently stands.
The first number is the minimum monthly revenue the founder needs the business to generate for working on it to be sustainable. Sustainable here is a real word, not a euphemism. It means the founder can pay rent, eat, see a doctor when they need to, and not destroy their savings or their relationships in the process. For most people in most cities this number is the sum of a handful of line items the founder is already paying every month. It exists today, attached to a real bank account, and the founder either knows it or can work it out before lunch.
The second number is how long the founder can tolerate operating below the first number before something has to change. This is the runway question, but framed honestly. Not “how long until the seed round,” because the seed round statistically does not arrive. How long until the personal cost of continuing exceeds what the founder is willing to pay. Six months. Eighteen months. Three years. The number is different for everyone and there is no right answer, but the founder has to know theirs.
These two numbers are the foundation of the exercise, and the reason the rest of it has any meaning at all is that it is being measured against them.
The Two Numbers That Are
Once the floor is on the table, the next question is whether a plausible business can plausibly clear it. This part is speculation, and I want to be honest about that, because pretending otherwise is the move that quietly empties the exercise of meaning. The next two numbers are guesses about a market the founder does not yet operate in, and they will be wrong in the specifics. The reason they are still useful is that they are now being measured against numbers that are not guesses, and that anchoring is the only thing that disciplines speculation into something a founder can act on.
The third number is a plausible unit price. Any reasonable guess will do, because the exercise is robust to being wrong about this by quite a lot. If the price is off by a third, the answer to the next question is off by a third, and the conclusion almost never changes.
The fourth number is the number of paying customers the business needs at any given time. The real question sitting behind this number is how much work it takes to find that many people, convince them to pay, keep them paying, and replace the ones who leave.
The exercise as a whole answers a question that TAM SAM SOM cannot answer at any level of precision, which is whether the founder can consent to the work in front of them on the basis of an outcome that is actually likely, rather than an outcome that is theoretically possible.
The Exit, And What It Is Actually For
I want to handle the exit question carefully, because the easy version of this post would dismiss the lottery ticket and that would be wrong. Plenty of founders do strike it rich. The exit case is not a fantasy. It happens. It just happens to a vanishingly small fraction of people, and the people it happens to are not, on average, identifiably different from the people it does not happen to in any way you can detect in advance.
The point I want to make is not that wanting the exit is irrational. It is that the consent to swing for the fence is only meaningful if the founder has actually priced the base case first. A founder who has done the floor exercise, looked at the realistic outcome, decided they can live with it, and then chosen on top of that to swing for a much larger outcome is making an informed bet. They know what they are buying and what it costs. The upside is a bonus on top of a base case they have already accepted.
A founder who has only ever looked at the upside case has not made an informed bet. They have made an uninformed one, and the thing they did not inform themselves about is the version of the future they are statistically most likely to actually live in. The problem is not the ambition. The problem is the missing exercise underneath the ambition.
This is, I think, the real argument for the floor question. Not that founders should be less ambitious, or that they should aim for lifestyle businesses instead of going big, or any of the slightly preachy things this kind of post tends to drift into. The argument is just that consent without information is not consent, and the information that has been missing from the validation conversation is the information about the base case. Once it is on the table, the founder can choose anything they like. They just cannot choose without it.
The Engineering Connection
There is a reason this is the question I find myself asking in almost every conversation with a founder before any code gets written, and it is not because I have appointed myself the keeper of anyone’s life decisions. It is because the engineering decision and the floor decision are the same decision. What gets built, how much it costs, and how much runway it consumes are all answers to a question about the floor, and without that question being answered I cannot tell a founder whether what they want is actually feasible inside the constraints they have. The floor is the context that turns “is this buildable” into a question with a real answer rather than a guess about a guess.
That is the version of validation I have come to trust, after years of doing the other version and watching it not predict anything I cared about. It is less impressive on a slide and it involves no acronyms. What it does instead is replace a conversation about a future that may or may not arrive with a conversation about what the founder can actually do this year, and what outcome from this year would count as a good one. That second conversation is the one that turns out to be useful, because it is the one the founder is actually in a position to make decisions about. Everything else is a forecast, and forecasts have never been the thing that decided whether an idea was worth working on.
