A model is not a forecast

There is a particular kind of spreadsheet that feels like knowledge and is actually a mirror. You build a revenue model: visitors times a conversion rate times a price, summed across products, compounded across months. The bottom-right cell says something enormous. It feels like a forecast. It is not. Almost every input was a number you typed because it seemed reasonable — a two percent conversion here, a thousand visitors there — and the giant total is just those guesses multiplied together and handed back to you in a confident font. You cannot hit it or miss it. It restates its inputs.

This matters because a model dressed as a forecast quietly changes how you behave. You start defending the number instead of testing it. You feel busy and on-track because the projection still says so, while the one thing the projection cannot contain — whether a single real stranger will pay you — goes unmeasured for weeks. The spreadsheet is not lying, exactly. It is doing precisely what you built it to do. The mistake is asking it a question it was never able to answer.

The fix is not better guessing. It is a labeling habit so simple it sounds trivial: tag every number in the plan as a fact or an assumption. A fact is something you measured — a real transaction, a real click-through, a count you could show someone. An assumption is everything else, no matter how confident it feels. Then you look at your grand total and notice, often for the first time, that ninety percent of it is standing on assumptions and maybe one cell is a fact. That is not a failure. That is the map finally telling the truth about the terrain.

Once the numbers are labeled, the whole plan reorganizes itself around one job: converting assumptions into facts as fast as possible. The assumed two percent conversion is not a thing to defend — it is a thing to go replace with a measured rate from a live page. The modeled price is a thing to test against a real checkout. Every week you should be able to point at one number that used to be a guess and is now a measurement. Progress stops being "the projection still looks good" and becomes "we retired three assumptions and two of them came back worse than we hoped." The second kind of progress is the only kind that compounds, because it is the only kind that is real.

This is also where a quiet discipline earns its keep: you never let an assumption walk out the door wearing a fact's clothes. The confident total is fine to steer by internally, as long as it is labeled — a stretch star, not a promise. But it never becomes a public claim, a marketing stat, or a number on a landing page, because the moment an unmeasured figure faces a customer, you have converted a modeling convenience into a lie. Sell with the things you can actually stand behind. Model with the things you are still guessing. Keep the wall between them, and label which side every number lives on.

So treat the plan as a machine, not a prophecy. Its output is not a target to be admired; it is a to-do list of assumptions ranked by how much they matter. Point the whole operation at the biggest unmeasured guess, go turn it into a fact, and repeat. A forecast tells you a comforting story about a future you cannot see. A labeled model tells you the one true thing you can act on today — which of your beliefs you have actually earned, and which one to go earn next.