Your annual budget is wrong by February. Use it anyway, differently.

The annual budget process in most businesses of this size runs for six to eight weeks, consumes a serious amount of leadership attention, and produces a document that is wrong within a quarter. Everybody knows this. The response is usually to start earlier next year and add more detail, which makes the document longer and no more accurate.

The problem is that one artefact is being asked to do three incompatible jobs: set targets, allocate resources, and predict the future. It cannot do all three, and predicting the future is the one it is worst at.

Separate the three jobs

The budget should be a resource allocation and a commitment: here is what each part of the business may spend, and here is what it has agreed to deliver. Set it once, keep it stable, and measure against it all year. Changing the budget mid-year to match reality destroys the only thing it is good for, which is accountability.

The forecast is the honest current view of where the year ends. It changes every month, because the world changes every month, and nobody is judged for updating it. Keeping these two separate is the single change that improves planning most, and it costs nothing but discipline.

What a rolling forecast looks like in practice

• Twelve to eighteen months ahead, re-cut monthly so the horizon never runs down to nothing in December.
• Driven by a small number of assumptions: heads, average revenue per head or per customer, churn, win rate, price. Twenty drivers, not two thousand rows.
• Updated, not rebuilt. If the monthly update takes more than a day, the model is too detailed to maintain.
• Accompanied by a bridge that explains the movement from last month's version in a handful of named steps.
• Owned by the budget holders, with finance running the model. A forecast finance produces alone is a finance opinion, not a plan.

Cut the detail, keep the drivers

Most budget models carry detail that adds no accuracy: every cost line forecast individually to the month, headcount plans naming individuals eighteen months out, revenue built customer by customer for a book that turns over each year. This detail feels rigorous and is the main reason the process takes six weeks.

Build the model on the handful of drivers that actually move the outcome and let everything else follow as a ratio or a run rate. A tighter model is faster to build, faster to update, easier to explain to a board, and, because the assumptions are visible rather than buried, considerably easier to argue with. Scenarios also become a five-minute exercise rather than a fortnight.

Then hold people to the budget and the forecast differently

A budget holder is accountable for spending within the budget and for the honesty of their forecast. Those are different obligations. Punishing people for a forecast that moves guarantees you will be told what leadership wants to hear until the point where the truth can no longer be hidden, which is always the worst possible moment to find out.