From intuition
to assumptions
A model is not the truth. It is a deliberately simplified version of reality, built to answer one useful question.
I used to think the hard part of modeling was choosing the right formula. This week reframed the problem: first define the decision, then identify the few relationships that matter enough to model.
Every cell should be traceable to either an assumption, an observed value, or a relationship between the two.
Build a simple three-scenario revenue model and test which assumption moves the forecast most. If the output is too sensitive to one input, that input deserves better evidence.