How to build a budget that doesn't lie to you

Most budgets are built in December, believed in January, quietly abandoned by March, and never mentioned again. The problem isn't discipline. It's that the budget was constructed in a way that made it impossible to use.

The two ways budgets go wrong

Last year plus a percentage

Take last year's P&L, add 15% to revenue, add 8% to costs, done. It's fast and it's internally consistent, and it tells you nothing — because it contains no statement about how the business will actually operate. When you miss, you cannot tell why, because there's nothing underneath the numbers to check.

The number you need to hit

The reverse: start from the outcome you want, then build a revenue line that reaches it. This produces a budget that is really a wish, and everyone in the business knows it by February. Once a budget is known to be fictional, it stops constraining any decision — which was its entire job.

Build it from drivers instead

A driver-based budget starts with how the business physically works, and lets the financial statements fall out of it. For a services business, the chain usually looks something like:

• How many people are delivering, and how many hours can each realistically bill?
• What's the effective rate after discounts and write-offs?
• What does utilisation actually run at — not what you'd like, what it ran at last year?
• How many new clients do you need, and what does your close rate say about the pipeline required to get them?

Revenue is then a consequence of those assumptions, not an input. That matters enormously, because when you miss, you can find out which assumption broke. Utilisation at 58% instead of 68% is a diagnosis. "Revenue was 12% under" is not.

Three tests before you trust it

1. Can you name every assumption?

If a line exists that nobody can explain, it's a placeholder someone typed. Delete it or justify it.

2. Does it reconcile to cash?

A profitable budget that runs out of money in month eight is not a budget, it's a trap. Every operating plan needs a cash view sitting beside it, with timing — receivables, tax payments, debt service, capital purchases.

3. Is there a version where things go badly?

One case is a prediction and predictions are wrong. Build three: base, a downside where your largest customer leaves or growth stalls, and an upside. The value isn't in the numbers — it's that you've already thought about what you'd do in each, before you're under pressure.

Make it live

A budget reviewed once a quarter is a document. A budget reviewed monthly against actuals is a control system.

The monthly ritual is short. Actual versus budget by line. Anything more than 10% off, ask why. If the answer is "the assumption was wrong," update the assumption and reforecast the rest of the year. If it's "we didn't do the thing we said," that's an operating conversation, not a finance one.

A budget you revise when reality changes is more honest than one you defend all year to avoid admitting it was wrong.

Rolling beats annual

The annual budget has a structural flaw: by October it covers two months. A rolling twelve-month forecast, updated monthly, always looks a full year ahead. Same effort, considerably more useful — you're always making decisions with a year of visibility rather than whatever is left of the calendar.

Start with the drivers. Keep it small enough that you'll actually update it. A budget with fifteen lines you maintain beats one with two hundred that you don't.