Building a forecast that holds up

Most early-stage forecasts look alike. A curve going up, a margin improving, profitability at month eighteen. Nobody believes it, you least of all.

The problem isn't the optimism. It's that the number arrives without the reasoning that produced it.

Nobody believes your numbers, and that's fine

An investor knows you won't hit that figure. They know it before opening the file.

They are looking at something else. How you got there. Which assumptions you made. And whether you know which ones are fragile.

A forecast isn't judged on its result. It's judged on the quality of the questions it makes possible.

Start from assumptions, not from the result

Many founders write the revenue they want first. Then they work backwards to the assumptions that produce it. That is a target, not a forecast.

Do the opposite. Write your assumptions, one per line, in plain words.

• How many prospects you reach each month, and through which channel.
• What share becomes a customer.
• How much each one pays, and for how long.
• What it costs you to serve them.

Revenue falls out on its own. It becomes a consequence, and each assumption becomes separately arguable.

Three lines are enough to start

A ten-tab forecast at your stage is a negative signal. It shows time spent in a spreadsheet rather than with customers.

Start with three things.

• Money in. What you collect each month, split by offer.
• Money out. What you pay out, split between fixed and variable.
• Cash. The available balance at the end of each month.

You will add detail when a decision depends on it.

The four-question test

Your forecast is ready when it survives four questions.

• What changes if you sign half as many customers?
• What changes if each one pays for half as long?
• Which assumption breaks the whole model if it's wrong?
• Which one can you test next month, cheaply?

If you can't answer on the spot, the file isn't what needs more work. The reasoning is.

What it changes

A forecast built this way works twice.

It works for you first. It tells you which assumption deserves testing this week rather than in six months.

It works in meetings next. When someone challenges a number, you don't defend the number. You show the assumption. You say why you made it. You say what you would do if it fell. That answer is what earns trust.

Romy runs this calculation on your numbers, and flags what changes when they move.