The rolling forecast is the most common answer to “we already do continuous planning.” A CFO hears the argument that the annual plan is obsolete by April, nods, and says the company moved to a twelve-month rolling forecast two years ago. That is a real improvement over an annual budget, and it is worth keeping. It is also not the thing. A rolling forecast refreshes the answer to a fixed question more often. Predictive planning changes the question. The difference is easy to miss from inside the finance function and expensive to miss from the board.
What a rolling forecast does
It replaces the fixed fiscal-year horizon with a moving one, usually four to six quarters ahead, and re-projects the numbers monthly or quarterly as actuals land. The better versions are driver-based: revenue is a function of volume and price, cost is a function of headcount and rates, and when a driver moves the model moves with it. The output is a fresher set of numbers for the same set of lines, on the same assumptions about what drives them.
That last clause is the whole issue. The rolling forecast takes the driver tree as given. It will tell you that revenue is tracking below the last projection because volume came in light. It will not tell you that the reason volume came in light is that a competitor shipped something that changes what your customers are buying, because that is not a driver in the model. It is a change in the story the model was built to describe.
None of this is a knock on FP&A. The rolling forecast was built to answer a finance question, how are we tracking, and it answers it well. The trouble starts when the executive team hears “continuous planning” and assumes the strategic question, should what we said still be what we say, is being asked somewhere in the same process. It is not. Nobody built a tab for it.
Where it fits inside the Loop
Woodring’s Loop runs four phases: Scan, Story, Stake, Steer. The rolling forecast lives in Steer, and it is good there. Steer is continuous correction against active commitments as the world moves, and a model that re-projects the numbers every month against fresh actuals is exactly the instrument that phase needs. The Convergence Review uses the rolling forecast as one of its inputs every month. Nobody running the discipline properly wants to get rid of it.
What the discipline refuses is letting Steer stand in for the other three phases. A company with a rolling forecast and nothing else has a continuous correction mechanism attached to a story that was written once, in September, and a set of commitments that were sized once, in November. The forecast rolls. The scenario and the stakes do not.
What it misses: Scan
A rolling forecast reads internal actuals. It does not read the world. The signals that tell you a load-bearing assumption broke, a rate move, a credit spread widening, jobless claims turning, diesel repricing, are outside the model until somebody manually changes a driver. By then the signal has aged into an actual and the forecast is reporting a variance that a Scan would have flagged as a risk two months earlier. This planning season is the clean example. The Fed raised its target ceiling to 4.00% on September 17. A rolling forecast built in early September does not know that until someone edits the rate driver, and in most companies nobody is assigned to.
What it misses: Story
A forecast has one future in it. Even a forecast with upside and downside cases has one future with error bars. Story is the discipline of holding three to five distinct worlds at once, each on a different cleavage, each with indicators that would tell you it is the one arriving. A rolling forecast cannot do this because it is built to converge on a number, and a number is the opposite of a scenario set. The forecast can be re-pointed at a scenario once the scenario is written. It cannot write the scenario, and it will not notice that the one it was pointed at has collapsed.
What it misses: Stake
The rolling forecast adjusts projections. It does not adjust commitments. When the forecast shows revenue tracking lower, the response is usually to trim discretionary spend to protect the margin, which is the one move the model can see. What the model cannot see is that the facility lease, the hiring wave, and the multi-year supply contract were all sized to a scenario that is no longer converging, and that the Stake Sizing Matrix would say to stage or hedge them. Those decisions do not live in the forecast. They live in a room, monthly, with the people who can re-open a commitment, and most companies with a rolling forecast do not have that room.
The tell
Open your rolling forecast and look at the assumptions tab, or the driver inputs, or wherever the things the model takes as given are kept. Check the last-modified dates. If the numbers roll every month and the assumptions were last edited at the start of the fiscal year, you have a rolling forecast that has become the plan. The model is doing exactly what it was built to do. It is being asked to do a job it was never designed for, which is to notice that the world it describes has changed.
The second tell is in the meeting. If the monthly forecast review ends with a revised number and a list of spend adjustments, that is Steer without Stake. If it ever ends with a commitment taken off the table, or a scenario declared dead, or a new one opened, something more than a forecast is running.
What to do Monday
Keep the rolling forecast. Add three things around it. First, a short public signal list, eight to twelve series from the Signal Library, each mapped to one driver in the model, read weekly by one named person. Second, a written scenario set of three worlds with indicators, so the forecast has more than one future to be pointed at. Third, a monthly Convergence Review where the forecast is an input and the output is a decision about commitments, not a revised number. That is the rolling forecast promoted from the plan to its proper job inside a plan.
The longer argumentabout why rolling forecasts, OKRs, and FP&A modernization all orbit the annual cycle without replacing it is on this site. The full discipline, phase by phase, is in the book.