A forecast is supposed to be a statement about the future. In most organisations it is a statement about what someone is willing to be held to, and those are not the same thing.
Watch how one actually gets made. An analyst runs the numbers and arrives at a figure. Before that figure goes anywhere, a second question arrives, quietly, and it is not whether the number is right. It is what happens to the person sending it if the number turns out wrong.
The answer to that question is asymmetric. Coming in under a forecast is awkward. Coming in over one, having promised the money was enough, is career weather of a different kind. So the number moves, not by much, and not dishonestly. It moves by the amount that makes it defensible.
Then it moves again one level up, for the same reason, by someone applying the same logic to the aggregate. By the time it reaches the board it has been made survivable three times, and nobody involved did anything unreasonable.
Here is the structural part. The person producing the forecast carries the consequence of the variance. The person consuming it carries the consequence of the decision. Two different people, and only one of them is exposed. A forecast that has to serve both jobs will always be optimised for the person who is exposed.
Which is why asking for a more accurate forecast rarely produces one. The accuracy was never the constraint.
Where I have seen this shift, it shifted because the two jobs were separated. Ask for the expected case and, next to it, what the person will commit to. Let the two differ on the record, without penalty. The gap between them is the most useful number in the pack, and at the moment it is the one number nobody is allowed to write down.
What would happen in your organisation if someone submitted a forecast with the buffer shown separately, and nothing bad happened to them?
Cordula Buss · Plan A2C · Helping finance and programme leaders build steering logic that works.