More Cadence, Same Decision Timing: The Rolling Forecast Before/After

The label changed; the decision calendar did not

On paper, it is a rolling forecast. The deck has 15 columns, a five-quarter horizon, a weekly refresh, and a tab labeled 'driver-based view.' But the meeting follows the old month-end script: last month's actuals versus budget, a variance bridge, and a list of corrective actions. The future columns sit to the right, built by formula, while no one makes a new capital or resource choice three months out.

The label has changed; the substance has not. The process has gained cadence and column count, but the moment when assumptions are debated and decisions are made remains the same. The rolling forecast is being consumed as a more frequent P&L review, not as a forward-looking decision forum.

Column mechanics can hide the unchanged script

A rolling forecast only changes operating rhythm if it moves the point at which a manager commits to hiring, inventory, pricing, or capital. The most reliable sign of adoption is not the number of columns or the refresh frequency. It is whether a forward-looking assumption appears on the agenda before the prior month's actuals are fully closed.

The extra columns are easy to build. Microsoft Dynamics 365 financial reporting can format headers specifically for rolling forecasts and export reports to Excel worksheets. Business Central column definitions control how columns appear in a report. A PDF export can carry up to 15 columns, while printed, previewed, and saved versions display a maximum of five columns. That distinction matters when someone says the pack has become wider. A 15-column PDF export can still be driven by the same five-column printed mental model.

Business Central 2025 release wave 2 added dynamic date headers. That is a useful systems feature but also a hidden risk. Dynamic date headers can make a column look forward-looking because the header advances with the current period. The header changes, but the calculation behind the column may still be a prior-month actual plus a growth factor. The date label is updated automatically; the assumption date is not. A 'driver-based view' tab in the workbook does not guarantee that the meeting spends time on drivers; it may only mean the formulas are stored on a separate sheet.

Concrete example: the lease decision stayed on day six

Suppose a mid-market distributor in Canada runs Business Central. The finance team expands its financial report from five printed columns to fifteen columns in PDF for the monthly forecast pack. The extra columns are month-by-month P&L projections labelled M+1, M+2, M+3, and so on. The first page still opens with last month's actuals versus budget, then a variance bridge against a static annual plan. The future columns are formulas that smooth prior actuals by a fixed growth rate.

The planned warehouse capacity decision, which requires a lease commitment about 90 days ahead, is still approved only after the month-end close confirms the latest volume. Before the rolling forecast, the lease decision was made on day 6 after close. After the rolling forecast, it is still made on day 6 after close. The column count rose; the decision timing did not.

Why the before/after test matters

A process can pass the steering-committee demo and still fail operationally. The confirmation bias is that cadence and column count are treated as evidence of transformation. A weekly refresh of a report that still waits for actuals is not a forecast; it is a variance tracker with more windows.

The diagnostic below is built around a before/after comparison. It does not rely on screenshots of the pack. It relies on the calendar.

A validation order that catches the unchanged decision calendar

Record the first ten minutes of the rolling forecast meeting. Write down the first question asked. If the first question is 'Why did revenue come in below budget?' rather than 'What price change should be tested for next quarter?', the meeting is still a month-end review.

List the decisions made in the meeting. Separate decisions that commit resources from decisions that explain prior results. A resource decision has an owner, a date, and a cost. A variance explanation has none of those.

Compare the approval dates for two or three resource-heavy choices before and after implementation. Examples include a pricing change for key customers, a shift in production staffing, or a warehouse capacity commitment. If the approval dates are unchanged, the rolling forecast has not moved decision timing.

Inspect the future columns. For each column beyond the current month, check whether the cell contains a driver assumption such as volume per site, price per unit, or headcount per shift, or whether it is a flat growth factor linked to prior actuals. CFA Institute modelling best practices include separating assumptions from calculations. That separation is the difference between a driver-based view and a formula-based view.

Run a delayed-close test. Ask: if the month-end close slips by three days, does the forecast review move with it? If the answer is yes, the forecast is still anchored to the close, not to the decision calendar.

These five checks share one question: was a decision date pulled forward? If no date moved, the rolling forecast is still a month-end pack with extra columns.

A limitation of this before/after test

A limitation of this before/after test is that it measures timing, not quality. A team could move a weak, unsupported assumption into an earlier meeting and pass the test while degrading the decision. That is why the driver-based test and scenario analysis matter as a second layer. Forecast approaches are generally based on historical results, historical base rates and convergence, management guidance, or analyst discretion. Scenario analysis is used to consider multiple outcomes rather than a single point estimate. If a team moves an early decision but only tests one scenario, the forecast has become earlier without becoming more resilient.

A rolling 12-month cash flow forecast model can be built in Excel from scratch using key assumptions and supporting schedules. The model structure is teachable, but it does not schedule the lease decision. The systems layer is rarely the blocker. The blocker is that the meeting script was never redesigned.

Monday Morning Takeaway

Dynamic date headers, 15-column PDF exports, and Excel forecast work can make a report look forward-looking. But if the decision still waits for the close, the only thing that has changed is the width of the pack.