Why most budgets fail
The typical budget is built once a year, in a spreadsheet, by extrapolating last year and adding a growth percentage. It is approved, filed, and consulted twice: once in month three when it is already wrong, and once at year end when somebody notices how wrong.
That is not a budgeting failure. It is a design failure. A budget built as a single annual event, disconnected from the accounting system and never revised, cannot do the job people expect of it — which is to inform decisions during the year rather than to score them afterwards.
A rolling forecast fixes this because it is maintained rather than set. It is revised monthly against actuals, always covers the next twelve months, and produces the only genuinely useful question in planning: what has changed since last month, and what does that mean.
Which businesses this applies to
Businesses that need to know whether they can afford something before committing to it — a hire, a location, a piece of equipment, a marketing programme. Businesses with a bank facility, where covenant compliance depends on the forward view rather than the historic one.
Businesses with investors or a board expecting forward reporting. Seasonal businesses, where an annual figure conceals the months that actually matter. And businesses whose owner has an intuitive plan that has never been written down, which is a surprisingly common and surprisingly risky position.
The work, step by step
What this looks like in practice:
- Build the revenue model from drivers, not from a growth rate. Volume, price, capacity, pipeline, seasonality. A forecast built on a percentage cannot tell you what to do differently.
- Separate fixed and variable costs properly, since that distinction is what makes scenario modelling possible at all.
- Model headcount explicitly — usually the largest cost and the one most often modelled as a single line that grows.
- Build the balance sheet and cash flow, not just the profit and loss. A plan that shows profit and ignores working capital is the plan that runs a business out of money.
- Set the phasing. Monthly rather than annual, reflecting seasonality and known timing rather than dividing by twelve.
- Agree the scenarios — base, downside, and whatever specific case is being considered.
- Establish the rolling process: actuals in, forecast revised, variance explained, next twelve months always in view.
- Report variance with explanation, because a variance number without a cause is a curiosity rather than information.
Driver-based, and why it matters
The difference between a driver-based model and an extrapolated one is that the first can answer questions.
An extrapolated budget says revenue will grow 15 per cent. When it does not, you know only that it did not. A driver-based model says revenue is customers multiplied by average order value multiplied by frequency — so when revenue misses, you can see whether it was customers, value or frequency, and those three have entirely different responses.
Building it takes longer at the start and pays for itself the first time something goes wrong. It also makes scenario modelling trivial rather than a rebuild: if the question is what happens when the largest customer leaves, a driver model answers it in minutes.
- Revenue as volume × price, with volume itself broken down where it can be
- Direct costs as a function of volume, not as a percentage of revenue
- Headcount modelled individually, with joining dates and full employment cost
- Overheads separated into genuinely fixed and step-fixed
- Working capital modelled on days — receivable, payable, stock
- Capital expenditure and financing shown explicitly rather than buried
The rolling process
A rolling forecast is a monthly rhythm rather than an annual event, and the rhythm is what makes it work.
Actuals for the month go in. The forecast for the remaining months is revised in light of what actually happened. A new twelfth month is added so the horizon stays constant. Variance against both the original budget and last month’s forecast is explained — the second comparison is usually the more informative, because it isolates what changed recently.
The whole cycle should take a couple of hours once it is set up. If it takes days, the model is too complicated and will be abandoned by month four, which is the fate of most forecasting models built by people who will not have to maintain them.
Common mistakes
The expensive mistakes in this area are consistent:
- Extrapolating last year plus a percentage, which produces a number and no insight.
- An annual budget never revised, obsolete by the second quarter and consulted only to explain a variance.
- Profit and loss only, ignoring working capital — the plan that runs profitable businesses out of cash.
- Dividing the year by twelve in a business with obvious seasonality.
- Headcount as a single growing line rather than individual hires with dates.
- A model too complex to maintain, abandoned by month four.
- Variance reported without explanation, which is arithmetic rather than management information.
The timing
The annual budget is built two to three months before the financial year starts, so it is agreed before the year rather than during it.
The rolling forecast is updated monthly, immediately after the management accounts are closed, so the revision reflects actuals rather than estimates. Scenario work happens whenever a significant decision arises — and the value of having the model already built is that a decision does not have to wait for it.
Deliverables
- A driver-based model producing profit and loss, balance sheet and cash flow
- Monthly phasing that reflects actual seasonality
- Base and downside scenarios, with the ability to add more
- A rolling forecast process the business can maintain in a couple of hours a month
- Monthly variance reporting with explanation
- Handover and training so the model is yours rather than ours
What to have ready
Nothing exotic, and most of it you already have:
- Financial statements and management accounts for the last two years
- The current chart of accounts
- Any existing budget or forecast and its assumptions
- The commercial drivers: customers, volumes, pricing, capacity
- Headcount plan and full employment costs
- Planned capital expenditure
- Details of facilities, covenants and financing
How this is priced
The initial model is a fixed-fee build, scoped on the complexity of the revenue drivers and the number of entities or segments.
Ongoing maintenance is either done by you — we build it to be maintainable and hand it over — or included in a CFO retainer where we are already engaged monthly. We would rather build something you can run than something that requires us every month.
Related
Frequently Asked Questions
What is the difference between a budget and a forecast?
A budget is the plan you commit to at the start of the year. A forecast is what is now likely to happen given what has actually occurred. A rolling forecast is updated monthly and always looks twelve months ahead, which is what makes it useful during the year rather than only at the end of it.
Why is our budget always wrong by mid-year?
Usually because it was built by extrapolating last year and then never revised. A budget set once and filed cannot inform decisions during the year. The fix is a rolling forecast updated monthly against actuals.
What is a driver-based model?
One where revenue is built from its components — volume, price, frequency, capacity — rather than from a growth percentage. It matters because when revenue misses, a driver model tells you which component missed, and those have entirely different responses.
Should the forecast include the balance sheet?
Yes. A plan showing profit and ignoring working capital is exactly the plan that runs a profitable business out of cash. Receivables, payables and stock need modelling on days, and capital expenditure and financing shown explicitly.
How long does the monthly update take?
A couple of hours once it is set up. If it takes days the model is too complex and will be abandoned by month four — which is what happens to most forecasting models built by people who do not have to maintain them.
Can we maintain it ourselves?
That is how we prefer to build it. We hand over the model with training, and it is yours. Where we are engaged as outsourced CFO we maintain it as part of the monthly cycle, but the model should never be something only we can operate.
Do we need a budget if we have never had one?
If you are making decisions about hiring, locations or capital, yes — because otherwise those decisions rest on a bank balance, which reflects the past rather than the commitments already made. A first budget need not be elaborate; it needs to exist and to be revised.
Tell us what drives your revenue. A model built on drivers rather than growth rates answers questions rather than just producing numbers.
Check my compliance status 058 101 9570
Last reviewed 27 July 2026. Rates, thresholds and deadlines change — the e-invoicing provider deadline has already moved once. Confirm current requirements with the Federal Tax Authority before acting, or ask us to check your position.