Tracking a budget honestly
A project that has spent 40% and committed another 50% has 10% of freedom left, not 60%. Most trackers report the 60%.
Committed spend is where budgets go wrong
The moment an order is placed the money is gone in every sense that matters. It is no longer available for anything else, and cancelling usually costs something.
A tracker recording only actuals reports the project as comfortably under budget and keeps doing so until the invoices arrive. Recording at the point of commitment rather than the point of invoice is the single biggest improvement most project budgets can make.
Remaining is not baseline minus actual
remaining = baseline − actual − committed. Leaving out the middle term is
how a line that is already fully spoken for shows a healthy balance.
Worth checking on any budget report you receive: ask whether the remaining figure includes commitments. Often nobody knows, which is itself the answer.
Contingency is not budget
Rolling contingency into the working lines makes everything look funded and hides the moment you start spending the reserve.
Held separately, "how much contingency have we used?" has an answer — and it is usually the earliest reliable signal that a project is in trouble. A project halfway through that has used most of its contingency is telling you something the schedule has not said yet.
A forecast defaulting to baseline is a lie
If a line has already spent and committed more than it was given, forecasting it at baseline is arithmetically impossible. It happens constantly, because the baseline is the number sitting in the template and nobody edits the cell.
Default the forecast to the greater of the baseline and what is already gone, and an overrun cannot hide in a blank field.
Scenarios are not predictions
A scenario says: if rates rise 8% and we add this scope, the budget becomes that. It says nothing about likelihood, and the distinction is lost the moment a single number leaves the room.
Present the spread rather than the middle. And test contingency against the worst credible combination, not the central case — comparing a reserve against your expected outcome tells you nothing about whether the reserve is real.