Blog · Forecast and pipeline · Sales teams
A sales force composite combines sales representatives' estimates into one forecast. Use a consistent submission basis, preserve review adjustments and compare the frozen estimates with actual results.
A sales force composite builds a sales forecast by adding the estimates submitted by representatives for their accounts or territories. Its value is the customer information behind those estimates; its weakness is that inconsistent definitions and individual bias can survive the roll-up. Collect a comparable set of estimates, preserve the review trail and test the frozen forecast against actual results.
Hyndman and Athanasopoulos describe the sales force composite as aggregating forecasts made by salespeople close to customers. They also explain why incentives and optimism can bias those judgments. Treat that as a reason to check submissions, rather than an assumption that any particular rep is unreliable.
Sales force composite = Sum of representative estimates on a common basis
Reviewed composite = Submitted composite + Documented management adjustments
This is a rep-estimate process. For its place alongside run rate, weighted pipeline and historical conversion, use four sales forecasting methods compared. Do not quietly switch between those methods inside a submission column.
Use one row per representative, exclusive territory and forecast period. If representatives share accounts, define the allocation first so their contributions sum to one forecast for that account.
| Column | Required content |
|---|---|
| Forecast version | Fixed snapshot date and reporting period |
| Rep and territory | Stable identifiers and effective ownership dates |
| Measure | For example, accepted order bookings excluding sales tax |
| Currency | USD here; retain a conversion basis for other currencies |
| Submitted estimate | Rep's expected total for the stated period |
| Adjustment | Signed amount added or removed during review |
| Adjustment reason | Evidence and the person approving the change |
| Reviewed estimate | Submitted estimate plus adjustment |
The actuals file needs the same territory identifiers, period and sales measure. Finance should confirm the bookings definition before results are used in an evaluation. A bookings estimate cannot be tested against invoiced revenue without a reconciliation between the two.
Give representatives the period boundaries, cutoff time and inclusion rules before they submit. Specify whether they estimate the full period or only the remaining period. An in-quarter full-period forecast includes actual bookings to date; adding those bookings again would double count them.
Require an explicit zero where a rep expects no bookings. A blank means missing information and should be resolved before aggregation. Check the territory roster against the submission roster so an omitted rep cannot silently reduce the forecast.
In this illustrative US industrial supplier, three representatives submit full-quarter bookings estimates for July 1 through September 30, 2026. All estimates and management reviews are frozen on July 1, before the quarter's results are known. Territories are exclusive, and all amounts are USD.
| Rep | Territory | Submitted estimate (USD) | Review adjustment (USD) | Reviewed estimate (USD) |
|---|---|---|---|---|
| Adams | East | 300,000 | -20,000 | 280,000 |
| Brooks | Central | 240,000 | 0 | 240,000 |
| Chen | West | 180,000 | 0 | 180,000 |
| Total | 720,000 | -20,000 | 700,000 |
The East adjustment removes an order whose documented delivery condition places acceptance outside the quarter under this company's bookings policy. It does not change the rep's original submission. Preserve that reason with the version being reviewed.
Place the table headings in row 1, submitted estimates in C, adjustments in D and reviewed estimates in E. In E2, fill down through E4:
=C2+D2
The totals are:
=SUM(C2:C4)
=SUM(D2:D4)
=SUM(E2:E4)
The reconciliation must return zero:
=SUM(E2:E4)-SUM(C2:C4)-SUM(D2:D4)
Here, 700,000 - 720,000 - (-20,000) = 0. If you roll up a larger table by territory and forecast version, Excel's SUMIFS function adds values meeting multiple criteria. Use equally sized ranges and select one version at a time. Covirage is not affiliated with Microsoft.
After quarter-end, finance supplies actual bookings on the agreed basis. Keep both submitted and reviewed versions: otherwise you cannot distinguish rep judgment from the management changes.
For this article, signed error is forecast minus actual, so a positive number means overforecasting. This sign convention must be stated because other references use the reverse.
| Rep | Actual bookings (USD) | Submitted error (USD) | Absolute submitted error (USD) | Reviewed error (USD) | Absolute reviewed error (USD) |
|---|---|---|---|---|---|
| Adams | 270,000 | 30,000 | 30,000 | 10,000 | 10,000 |
| Brooks | 260,000 | -20,000 | 20,000 | -20,000 | 20,000 |
| Chen | 150,000 | 30,000 | 30,000 | 30,000 | 30,000 |
| Total | 680,000 | 40,000 | 80,000 | 20,000 | 60,000 |
In a separate comparison sheet, put submitted forecast in B, reviewed forecast in C and actual in D. The row formulas are =B2-D2, =ABS(B2-D2), =C2-D2 and =ABS(C2-D2).
| Check | Submitted version | Reviewed version |
|---|---|---|
| Forecast less actual | 720,000 - 680,000 = 40,000 | 700,000 - 680,000 = 20,000 |
| Aggregate signed error / actual | 40,000 / 680,000 = 5.9% | 20,000 / 680,000 = 2.9% |
| Absolute row errors / actual | 80,000 / 680,000 = 11.8% | 60,000 / 680,000 = 8.8% |
Percentages are rounded to one decimal place. The final line is weighted absolute percentage error: it preserves errors that cancel in the signed total. An aggregate actual of zero makes those percentage checks undefined; report dollar errors in that case.
The reviewed version was closer in this illustration. That does not establish a durable improvement. Forecasting: Principles and Practice explains why evaluation needs genuine forecasts on unseen outcomes. Compare several frozen periods at the same lead time, then use forecast bias by rep for persistent directional patterns. The worked example on five reps and four quarters extends that review across periods.
Bring the frozen rep submission file, review adjustments and matching actuals export to Covirage. Covirage maps the columns, and its tools compute and check the roll-up and forecast errors; the external AI model explains the result with every figure traced to its rows. Create a free Covirage account and bring a sample of your own export.
See forecast analysis for the supporting workflow.
It is generally classed as judgmental forecasting because representatives estimate future sales using customer knowledge and commercial judgment. The submitted estimates are numerical, but adding them does not remove the judgment behind their expected timing and amounts.
A quota states the result a team is expected to achieve. A composite estimates the result representatives currently expect. Keep separate fields for the target and the forecast so a gap remains visible instead of being removed through an unsupported adjustment.
Managers can review estimates against documented evidence, such as a contract date or duplicate account ownership. Preserve the submitted amount, adjustment and reason separately. Test whether the reviewed version improves later results rather than assuming every intervention adds accuracy.
Compare forecasts frozen at a consistent lead time with the eventual actuals on the same basis. Inspect signed errors by representative and territory across several periods. Also inspect absolute errors because overestimates and underestimates can cancel in the total.