The FD looked at me across the table and said: "Our forecast was wrong before we finished Q1."

She wasn't embarrassed. She was resigned. Her business had run the same static annual budget process for seven years. Every year the forecast was dead by March. Every year the board trusted the numbers a little less. And every year, decisions got delayed because nobody believed the data underneath them.

She is in large company: 43% of finance leaders admit their forecasts carry 10%+ error. Most have accepted it as the weather.

It isn't the weather. Across multiple businesses, on budgets from £10M to £200M+, I have delivered 96%+ forecast accuracy - and never once was software the reason. It was four disciplines, all adoptable inside a quarter.

Accuracy is a trust problem before it is a numbers problem

Here is what an inaccurate forecast actually costs. Not the variance itself - the board meeting. Decisions deferred "until we have better numbers". Finance leaders spending political capital defending last month instead of shaping next quarter. A leadership team quietly running on gut feel because the official numbers arrived discredited.

A forecast nobody believes is more expensive than no forecast at all - because it still costs money to produce.

Fix accuracy and something bigger than accuracy gets fixed: the numbers become the place where decisions happen.

The four disciplines

1. Document every assumption, explicitly

Every forecast line traces back to a named assumption: this growth rate because of this pipeline, this cost because of this contract. When a number moves, you know which belief was wrong - and you fix the cause. "The forecast was off" with no explanation is banned.

This is the discipline that does the most work. An assumption that isn't written down can't be wrong, so it never gets corrected.

2. Review variances weekly

Monthly variance analysis is archaeology. By the time the monthly pack lands on the board table, the problem is six weeks old and the moment to correct it has gone. A weekly look - short, informal, "which assumption broke?" - catches drift while it is still steering, and turns variance review from a blame exercise into maintenance.

3. Kill the static annual budget

The annual budget is a number somebody guessed in October, defended until Christmas and mourned by March. Replace its grip with two things, each doing its own job:

  • Full reforecasts, two or three times a year, done properly - assumptions rebuilt, drivers revisited, leadership engaged. A few forecasts done rigorously beat a perpetual re-spin that nobody has time to think about.
  • A weekly 13-week cash view - because cash needs continuous sight in a way the P&L does not. That discipline has its own article.

4. Make operational leaders own their numbers

Finance consolidates and challenges. Sales owns the revenue forecast. Ops owns the cost line. The people closest to the decisions make the best predictions - and once a leader has signed a number, the conversation about missing it changes entirely.

What the board meeting feels like afterwards

The FD with the dead-by-March budget implemented all four disciplines in a single quarter. Within six months her board had stopped questioning the numbers - and started using them. Planning conversations moved from "are these figures right?" to "what do we do about what they show?"

That is the real product. Not the 96%. The 96% is just what makes it possible.

Key takeaway: Forecast accuracy is not a software feature or a talent lottery. It is four habits - named assumptions, weekly variance, proper periodic reforecasts, operational ownership - and a quarter of honest effort.

Start with one

If you adopt a single discipline this month, adopt the first. Write the assumptions down. It costs nothing, it needs no meeting, and it quietly exposes where the other three are missing.

How confident is your board in the numbers your finance team presents?