Static budgets vs. rolling forecasts

Rolling forecast vs budget: two documents, two jobs

A static budget is a plan for a fixed period, usually a fiscal year, approved before the period starts and not revised inside it. Actual results are compared against it and the variance is the story.

A rolling forecast is a projection over a constant horizon, usually twelve or eighteen months, revised on a fixed cadence so that as one month closes another is added at the far end. It is never finished and never a commitment.

People argue about the two as if they compete. They answer different questions. The budget answers what did we agree to. The forecast answers what do we now believe will happen. A business that confuses them either treats every revision as a broken promise or treats its budget as a suggestion.

What a static budget is genuinely good at

A static budget is the better tool wherever a fixed number has to be defended for a full year.

Capital planning is the clearest case. If you are deciding in November whether to buy a building, refit two locations and add a service line, you need one authorised number per project and a total that does not move. A forecast that shifts the capital figure monthly cannot function as an approval. Debt covenants work the same way, because lenders test against fixed thresholds.

Compensation is the third case and the most underrated. Bonus plans need a target set before the period that cannot be moved once results start arriving, and a rolling forecast makes a terrible bonus target because it is revised by the same people whose bonuses depend on it.

Where static budgets break

A static budget breaks when the gap between plan and reality gets large enough that people stop referencing it.

That happens faster than most owners expect. A budget built in November using October data is five months old by April. If one location underperforms by 20%, a lease renewal lands 30% higher than assumed and a large customer changes terms, the April variance report is archaeology.

The specific failure is that variance stops being actionable. When a line is 18% off plan you need to know whether the assumption was wrong from the start or whether something changed in March, and a static budget cannot tell you the difference.

The second failure is quieter. People start keeping their real numbers somewhere else. Once the operator of an entity has a private spreadsheet with what they actually think will happen, the budget has become a formality.

What a rolling forecast changes: cadence and horizon

A rolling forecast replaces one annual event with a repeating small one, and the two parameters that matter are cadence and horizon.

Cadence is how often you revise. Monthly is right when the business genuinely changes monthly: seasonal demand, project-based revenue, volatile input costs, anything with occupancy or utilisation swings. Quarterly is right when the business moves slowly and monthly revision would just be noise. Choosing monthly cadence for a stable business is the most common way groups make rolling forecasting expensive without making it useful.

Horizon is how far out you look, held constant. Twelve months is the default and works for most owner-led groups. Eighteen earns its place when you have long lead times on capacity decisions.

The horizon staying constant is the point. A budget's horizon shrinks all year, so by November you are planning six weeks ahead. A rolling forecast always shows the same distance, which is why it keeps supporting decisions late in the year.

Re-forecast discipline: what has to happen every month

A rolling forecast without discipline is worse than a static budget, because it consumes real time and produces a number nobody trusts.

Discipline means four things. The re-forecast happens on a fixed date, not when someone gets to it. Each entity's operator updates their own drivers rather than finance guessing on their behalf. Only material changes are made, so the exercise is not a monthly rebuild. And each version is saved, so you can look back at what you believed in March and see whether you were right.

That last one is the part everyone skips, and without version history you cannot measure your own forecast accuracy. A forecast whose accuracy is never measured drifts toward optimism.

The time cost is the real objection and it is fair. AFP/APQC benchmarking found that only 25% of FP&A time goes to value-added analysis, with 42% spent gathering data. The FP&A Trends Survey 2024, covering more than 2,400 practitioners, put the share of time spent on high-value work at 35%. Those numbers describe teams with dedicated staff; an owner doing this alongside running the businesses has less slack, not more.

If the monthly re-forecast takes more than half a day for six entities, the model has too many lines. Forecast the ten things that move the outcome and let the rest run on last year plus a factor.

The political cost of re-opening a budget

This never appears in the methodology guides and it is usually the reason rolling forecasts fail inside owner-led groups.

Revise a forecast downward and the operator of that entity hears an accusation. Revise it upward and they hear a new target. Either way the revision arrives as a judgement, and after two or three cycles operators start submitting numbers they can beat rather than numbers they believe.

The structural fix is to separate the two documents by purpose and say so explicitly. The budget stays fixed and stays the basis for bonus and approval. The forecast is revised freely and carries no compensation consequence. Once operators know that revising a forecast cannot cost them money, the numbers get honest.

The second fix is to require a reason rather than a number. A change from 400,000 to 340,000 is a negotiation. A change because occupancy fell from 88% to 79% is a fact, and facts are harder to argue with.

So which one? An honest test

Ask two questions of each business.

First, how far does actual revenue land from plan by month six? Within five points and a static budget with quarterly variance review is sufficient. Fifteen or twenty points and the budget will be dead by spring.

Second, how many decisions do you make each quarter that depend on a forward view? Hiring, lease commitments, equipment, whether to fund a slow entity from a strong one. A business making one such decision a year does not need a monthly forecast; a business making one a month does.

For most groups the answer is mixed, and there is nothing wrong with running a static budget on the predictable entities and a rolling forecast on the two that move.

Running both without doubling the work

The practical arrangement is one set of numbers with two views: the budget locked as a version, the forecast live. Both use the same account structure, so comparison is arithmetic.

Report three columns: actual, budget, current forecast. The actual-to-budget variance is the accountability conversation. The forecast-to-budget gap is the management conversation, and it is the more useful of the two.

Keep the forecast at a coarser grain than the budget. A budget may carry two hundred lines per entity; a forecast needs twenty or thirty. Detail in a forecast is not accuracy, it is more places to be wrong, and a driver-based approach keeps the line count down while making assumptions visible.

A worked example: a group of six entities

Consider a group of six: three operating businesses in the same trade, a property entity holding two buildings, a management company, and a dormant holding entity. The FY budget was approved in November. By the end of Q1 the picture has split.

Three entities need no re-forecast at all. Their budgets are still good and revising them monthly would generate work and no information. Entity B needs a monthly re-forecast because the outcome determines whether the group funds it. Entity C needs one revision to capture the new contract and can then go quiet.

The group view is where the interesting number sits. The current forecast is below budget for the year and the entire gap is one entity. A single consolidated variance would have shown a modest shortfall and hidden the fact that one business is in trouble and another is ahead, which is why multi-entity groups need entity-level detail alongside the roll-up.

For the entity in trouble, the annual forecast is not the urgent document. What matters there is the next quarter of cash, which is a 13-week cash-flow forecast.

Can QuickBooks or Xero do this?

Honestly, only the first half.

QuickBooks Online and Xero both let you load a budget per entity and report budget versus actual against it. The actuals side is good: correct, reconciled, drillable to transaction level. For a single entity with a static budget, that report is most of what you need.

What neither does is planning. There is no way to hold multiple forecast versions, no scenario switching, no driver inputs feeding several lines at once, no rolling horizon. Budget-versus-actual is a reporting feature, and the difference shows the first time you want to compare what you believed in March against what you believe now. For a group it is harder again: budgets live inside each entity's file, so a group-level comparison means exporting several reports and combining them.

So most groups end up in a spreadsheet, which is reasonable if it is a maintained model, not a monthly rebuild. The Intuit Enterprise Technology Benchmark found that 64% of multi-entity firms say the close takes too long and 76% say their technology struggled when they added entities. The planning problem usually sits downstream of a reporting problem.

A re-forecast on the 10th needs last month closed by the 10th. APQC's Open Standards Benchmarking across roughly 2,300 organisations puts the median monthly close at 6.4 calendar days, top quartile at 4.8, bottom quartile beyond 10. If your group closes on day 18, a monthly cadence is not available to you whatever tool you buy, and fixing why the close takes so long comes first.

Where to start

Do not convert the whole group. Pick the entity whose budget is furthest from reality and put it on a monthly rolling forecast with twenty lines and a twelve-month horizon. Leave the stable entities on their budgets.

Run it for a quarter, keep every version, and then compare what you forecast in month one against what happened. That comparison tells you whether the cadence is earning its keep, which is a better basis for extending the approach than any argument about methodology.

If the constraint turns out to be that your actuals arrive too late to re-forecast against, that is the part cruisr works on: current books and a group-level close on QuickBooks Online or Xero. Get in touch.