Skip to content

Rolling Forecast Review

Review cadence — instantiates Moving-Target Tracking

A scheduled and event-triggered ritual that re-forecasts where the target is heading and refreshes the scenario spread, so plans always ride current evidence rather than a fixed period boundary.

The trouble with a once-a-year forecast is that it ages the moment it is signed: by month nine everyone is steering toward a picture of the future assembled before any of the year actually happened. Rolling Forecast Review replaces that with a cadence — a standing ritual that, on a fixed schedule and on defined trigger events, re-projects the target's trajectory forward a constant horizon and refreshes the spread of scenarios around it. Its defining move is not any single forecast but the recurring refresh: it always looks the same distance ahead, and it treats the forecast as a living estimate that is expected to be replaced, not a commitment to be defended. What it produces is not a control action and not a drift test — it is an updated, uncertainty-tagged view of where the target is going, handed to whatever plans and loops consume it.

Example

A subscription software company runs a rolling 18-month revenue and headcount forecast, reviewed at the close of every month and re-opened immediately whenever a defined trigger fires — a large customer churns, a pricing change ships, a quarter closes materially off-plan. Each review does not re-litigate the annual budget; it slides the 18-month window forward one month and refreshes the scenario spread: a base case, an upside if two enterprise deals land, a downside if net revenue retention keeps slipping. Hiring plans, cash runway, and quota-setting all read from the current review rather than from January's frozen numbers.

When a competitor cuts prices mid-year, the trigger re-opens the forecast the same week: the downside scenario widens, the base case shifts, and the runway view updates before the old plan can steer the company into over-hiring against a target that has quietly moved. The value is not any one number's accuracy — it is that no decision is ever more than a month stale, and every decision sees the range, not a false point.

How it works

What distinguishes a rolling review from a one-off forecast is that the refresh itself is the mechanism:

  • It runs on a constant horizon that always slides forward, so the organization never runs out of forward view the way a fixed year-end date does.
  • It fires on two clocks at once — a regular calendar cadence and event triggers — so both slow drift and sudden news pull a fresh forecast.
  • It refreshes a spread of trajectories, not a single line, so consumers see the plausible envelope and how it has widened or narrowed since last time.
  • It is deliberately decision-agnostic: it updates the picture and hands it off; the acting on it belongs to planning and control mechanisms downstream.

Tuning parameters

  • Review cadence — how often the scheduled refresh runs. Faster keeps forecasts current but consumes attention and invites over-reaction to noise; slower is cheap but lets staleness accumulate.
  • Trigger set — which events force an off-cycle review. A broad, sensitive set catches shocks early but can thrash the process; a narrow one is calm but slow to a surprise.
  • Horizon length — how far each refresh looks ahead. Longer supports long-lead decisions but carries more forecast error; shorter is sharper but leaves long-lead plans unguided.
  • Scenario breadth — how many trajectories the spread carries and how far apart. Wider honestly represents uncertainty but can paralyze; narrower is decisive but risks false confidence.
  • Regeneration depth — whether each review rebuilds forecasts from raw evidence or merely nudges the prior one; deeper resists anchoring but costs more effort.

When it helps, and when it misleads

Its strength is keeping the whole system's picture of the target fresh and honest about uncertainty. By refreshing on both a clock and on events, and by carrying a spread rather than a point, it stops plans from steering confidently toward a destination that moved months ago — and it makes the widening of uncertainty visible before it becomes a crisis.

Its failure modes are about cadence and candor. Refresh too often and the organization chases noise, re-planning around every wiggle; refresh too rarely and it is back to a stale annual number. Its most common quiet failure is anchoring: a "review" that merely tweaks last period's forecast a few percent rather than rebuilding it, so the number never really moves even as reality does.[1] A related misuse is treating the forecast as a target to defend — sandbagging the base case so it is easy to beat — which corrupts the very estimate the rest of the loop depends on. The discipline that guards against this is to regenerate from current evidence rather than nudge, to keep the forecast strictly separate from the commitment, and to review the spread and its drift, not just the headline point.

How it implements the components

Rolling Forecast Review fills the archetype's forecast-refresh slot — the parts that keep the forward picture current, not the parts that act on it:

  • update_cadence — it is the cadence: the scheduled-plus-triggered rhythm on which the target's projection is refreshed.
  • target_trajectory_ensemble — each review regenerates the spread of plausible target trajectories, not a single forecast line.

It does not test whether the target has already drifted from plan (tracking_error_metric — that is Rolling Window Comparison), decide the next action from the forecast (the planning cadence is Rolling Planning Cycle; the optimizing controller is Model Predictive Control), or govern the objective's own version history (Objective Versioning and Change Log). It refreshes the forward view and hands it off.

Notes

A refreshed forecast is inert until something consumes it. Rolling Forecast Review deliberately stops at supplying the current forward view — pairing it with a planning cadence or an anticipatory controller is what converts the refreshed picture into action. Its closest sibling, Rolling Planning Cycle, runs on the same rhythm but re-plans; keeping the forecast review and the plan revision as distinct steps is what lets the forecast stay an honest estimate rather than becoming a negotiated commitment.

References

[1] Anchoring — the tendency for an estimate to stay near a previously seen number regardless of new evidence. In forecasting it shows up as reviews that only ever adjust the prior forecast by small increments, so the projection lags reality; regenerating from evidence rather than editing the last number is the standard corrective. ↩