Management projections are an argument, not a forecast. What a decade of assessing them taught me.

I spent about a decade in corporate banking and did two stints as CFO. In banking the job was to take whatever numbers a company put in front of us, work out how much of it we actually believed, and build our own projections from there. We never used their plan as given. That work changes how you read guidance and I don't see it discussed here much, so here it is.

First thing you learn is that the gap between what a company projects and what it delivers is not random. It leans the same way and it leans in the same places. Revenue slips by a quarter or two rather than disappearing. Cost savings land later and smaller than the plan says, and working capital always looks better in the forecast than it does in the outcome. Once you have seen enough plans against enough outcomes you stop reading a projection as a forecast and start reading it as an argument the company is making.

On ranges. The bottom of a range is usually the number they are confident about and the top needs several things to go right at the same time. So the midpoint is not the middle of anything real. If you are plugging the midpoint into a model you are using a figure that was chosen for how it presents rather than for how likely it is.

The check that does the most work is track record. Pull what a management team said eight quarters ago and put it against what happened. Every team has a house style and eight quarters is enough to see it. Some are consistently conservative, and once you know that you can adjust up. The ones I paid most attention to were the teams that missed their own low end, because a forecasting problem inside the business usually tells you more than the miss itself does.

The CFO side taught me the other half. I built those projections myself, and what that teaches you is how much judgment sits inside a number that looks precise on the page. Change two or three assumptions that all seemed reasonable at the time and the answer moves a long way. Nobody outside can see which assumptions got chosen or why.

This is why I don't start from the guidance number. Make your own guidance, pick the forward growth rate you actually believe, and check it against what the business has done over the last five years. Set the return you need to make it worth owning. See what value falls out, then change the growth to something you'd be less comfortable with and see how far the number moves.

Now guidance ends up as one input you weigh against your own view rather than the thing everything else gets built around.