Sales forecast range: three cases instead of one number

The forecast range consists of three named cases: worst case, realistic case and best case. Each case stands for an assumption about the business, not for a confidence bound. For a time window, the best case can lie below the worst case if stock shrinks faster than inflow replaces it. Finished periods show whether a sales forecast's edges hold.

Symptom: how do you notice that a number is not a forecast?

In the meeting before the quarter, somebody asks for “the number”. Sales leadership names an amount. Three months later the result lies below it, and nobody can say which assumption did not hold.

Or there are three numbers: a cautious one, a middle one, an ambitious one. Nobody in the room knows how the three differ, apart from their height.

Three observations belong to this pattern:

• The sales forecast is a single amount without an assumption.
• The three cases differ only in tone.
• Nobody has looked at how earlier forecasts stood against the result.

For the revenue leader who defends the forecast before the board, the consequence is the same. The number can be named, but it cannot be defended.

The assumption gives a case its name, not its height. That is the core of this article.

Cause: what is a forecast, if it is not a single number?

Three roles, three assumptions

A sales forecast is a statement about the future and therefore rests on assumptions. The forecast range makes the assumptions visible. Worst case, realistic case and best case are roles here: the lower edge, the middle case and the upper edge.

Which assumption fills a role is determined by the question. For the range of a time window, this mapping applies.

In this mapping the worst case is the rEFV. rEFV (Risk-Adjusted Expected Funnel Value) is bEFV weighted by SLA Health. The weighting uses measured factors per band. More on this is in the entry rEFV, the risk-adjusted expectation.

In this mapping the best case is the cEFV. cEFV (Calibrated Economic Funnel Velocity) is the calibrated revenue forecast for the end of a period. cEFV calibrates rEFV with the value growth of existing deals and the inflow of new deals from completed prior periods. cEFV is the best case of the forecast range, not the realistic one.

No confidence bounds

Worst case and best case are named assumptions about the business. They are not confidence bounds and not probabilities. Both have to be defensible without talking about mathematics.

Where the assumption of a case is not measurable, the case is reported as not measurable. The case is not filled with an estimate. An invented range is worse than the single number it replaces.

The question determines the assumption

For the coming quarter, a different assumption fills the same roles. There the question is what slips over from the running quarter.

In the worst case nothing slips over; only what is already dated for the coming quarter counts. In the realistic case the share that has slipped over so far is added. In the best case everything still open that is dated for the running quarter is added.

Where the share that has slipped over so far is not measured, the realistic case is reported as not measurable.

With this filling, the order is given by construction: slipping over can only add.

The role is the same, the driver is not. What could be pulled forward from a later quarter is a decision of the sales organisation and not a summand of the range.

cEFV stands in the range, not alone

cEFV is one number: the upper edge of the range of a time window. The lower edge is rEFV, which assumes neither value growth nor inflow. Both edges are assumptions, not limits.

More on the upper edge is in the entry cEFV, the best case of the forecast range.

Measurement: how is a forecast range read and checked?

The names follow the assumption, not the height

For the range of a time window, the three cases are not assigned by size. The best case counts the value growth and the inflow that the prior periods show. Where those values point downwards, the best case can lie below the worst case. The realistic case need not lie between the edges either.

Then there are no three ordered cases. There are two scenarios and a finding: the stock is shrinking faster than inflow replaces it. The cases are not re-sorted in that situation. The finding is the news, not the table.

The lower edge is an assumption, not a floor

The worst case says which assumption forms the lower edge: neither value growth nor inflow. The worst case does not say that the result cannot fall below it. rEFV is an expectation, not a floor.

The edges against the actual

Whether a range holds is shown only by the comparison with the result. The edges are set beside the revenue actually won for every finished period: as a series, not only for the most recent period. Beside every value stands the case count, that is, the number of decisions the value rests on.

A lower edge that the actual of finished periods has undershot is not a limit but an assumption. An edge that has held so far remains an assumption as well.

The EFV Ratio is related. EFV Ratio divides actual Closed Won revenue by rEFV at the start of the period.

The EFV Ratio, actual against rEFV, is read as a series. The EFV Ratio says something about the calibration of the forecast, not about quota attainment. A deviation from 1 is not a finding on its own; what is read is the series, with the case count per period.

What quota attainment is, the article quota attainment, actual against plan explains.

A window in the future, a running quarter

For the range of a time window that is not running yet, there is no measurement. The edges are then constructed, and exactly that is said. A narrower statement does not come from a different calculation.

The running quarter gets no range. The running quarter gets three standings.

The three cases of the range describe a window as seen from its first day. They do not apply to the running quarter, and that is said.

Countermeasure: which four questions belong to every forecast?

This article recommends reading every sales forecast with four questions.

The four questions follow from the sections above.

The comparison with the plan is covered in the article Forecast Integrity, the forecast against plan and actual. There, “forecast” means the rEFV; here it means the range of three cases. No verdict follows from the distance to the plan alone.

Where the actual deviates from the forecast in a finished period, the decomposition of the deviation comes first. Rebuilding the procedure does not come first.

What matters is not how sharp the number is. What matters is how early somebody notices the deviation.

For the finance leader who checks the sales forecast, that means: every case comes with its assumption, and every edge comes with its series against the actual.

A forecast for an upcoming window is a range of three named assumptions. The assumption gives the case its name, not its height; the edges are checked against the actual.