Why we kill ventures on a spreadsheet.

There is a version of founder culture that treats conviction as a virtue in itself. The founder who keeps going when everyone else would stop. Who ignores the signals and backs their vision anyway. Who survives long enough that the world catches up.

This narrative produces some of the most celebrated companies in technology. It also produces an enormous amount of quietly destroyed capital, most of which never makes the case studies.

We do not romanticise persistence. We respect it when it is backed by evidence and distrust it when it is a substitute for evidence. The difference matters more than most founders want to admit.

When we evaluate a venture, we build the economics before we build anything else. Not a financial model in the fundraising sense, formatted to look serious and projected to Year 5. A real model: unit economics at the level of one transaction, run at three different volumes, with every assumption named and owned.

The purpose of the model is not to confirm the thesis. It is to find the point at which the thesis breaks. Because every thesis has one. The question is whether you find it in the spreadsheet or in the market.

Finding a break in the model costs nothing. Finding it after eighteen months of operation costs everything.

Most founders treat the model as a hurdle to clear on the way to building. They construct it backwards from a number that looks fundable, populate it with industry benchmarks that carry no accountability, and mark the key assumptions in light grey so they are technically present but practically invisible.

This is not modelling. It is narrative presentation with a spreadsheet attached. And the problem with it is not that it deceives investors. The problem is that it deceives the founder.

We have killed ventures in the model. More than once. The decision is not dramatic. There is no single moment where the numbers collapse. It is slower than that: an assumption you thought was conservative that turns out not to be, a retention figure that does not hold at volume, a cost of acquisition that makes sense at the early adopter stage and becomes the entire margin at scale.

You pull on the assumption. You rebuild that section of the model with a more honest number. You watch what happens to the rest of it. Sometimes it adjusts and holds. More often, it reveals that one more assumption downstream was carrying weight it should not have been.

At some point the model is telling you something clear: the business does not work under realistic conditions. That is not the model being pessimistic. That is the model being honest in a way that the founder pitch is structurally unable to be.

The model is the only version of the business that cannot be charmed.

Killing a venture at this stage requires a specific kind of discipline. Not courage, exactly. The courage required to shut down a venture you have been developing for months is real, but it is not the hard part. The hard part is resisting the temptation to adjust the assumption rather than face what the model is saying. Every founder knows how to make a model work. The question is whether you are making the model honest or making it agreeable.

When a model breaks, we do not move on immediately. We interrogate the break. A model that fails is not a closed door. It is a specific question: why does this not work, and is that why addressable?

Sometimes the break is structural. The unit economics of the category simply do not support a venture scale business, and no operational efficiency changes that. That is a real answer, and it is a useful one. We move on.

Sometimes the break is conditional. The model does not work at current market conditions but holds if a specific variable shifts: regulatory change, infrastructure buildout, a distribution channel that does not yet exist but is clearly forming. That is a different conversation. The question becomes whether we can time the venture to the shift rather than launch it prematurely.

And sometimes the break reveals something more useful than either: a different version of the venture, targeting a different segment or running on a different revenue model, that actually works. The kill becomes a pivot, grounded in evidence rather than instinct.

We have been asked, more than once, whether this approach is too conservative. Whether we miss opportunities by killing ventures that a more risk tolerant operator would have pushed through.

The honest answer is: probably, in individual cases, yes. There is almost certainly a venture we killed in a model that a more persistent founder would have taken to market and figured out.

But the expected value of that persistence is negative across a portfolio. The ventures that survive a broken model by willpower alone are the exceptions. The businesses that fail because someone kept going after the evidence said stop are not exceptional. They are the majority.

Discipline is not the absence of ambition. It is the thing that makes ambition sustainable.

We build with conviction. The conviction has to be earned. Earning it means building the model, finding where it breaks, asking whether the break is real, and being honest about the answer. That process is the work. Everything that comes after it is execution.