One accuracy number hides the useful one
Forecast accuracy by week of quarter grades the forecast at every point in the period instead of only at the end. Nearly every team measures the final call against the actual result and reports a single percentage. That measurement rewards the easiest prediction anyone makes all quarter.The version worth tracking is a curve. Plot the absolute error of each weekly snapshot against the closed result, then average those weeks across several quarters. The shape tells you when the number became usable, and usable early is the entire point.
Build it from locked snapshots
The curve requires forecasts that were fixed at the time they were called. A CRM that only holds the current state cannot produce it, because last week's forecast was overwritten. Snapshot the forecast at the same hour each week and store it alongside the week number and the quarter.
Grade each snapshot against the same denominator, the final closed number for that quarter. Two error measures do most of the work. Signed error shows direction and exposes a team that is consistently light or consistently heavy. Absolute error shows magnitude and is the one to plot as the curve.
What a healthy curve looks like
A strong curve is flat and high from the opening weeks, with small corrections as deals resolve. A weak curve starts wide, stays wide through the middle of the quarter, then snaps into place in the final two weeks.
ORM reports that forecast accuracy on new and expansion business typically lands near 90 percent, and that reaching it takes heavy manual effort and does not adapt as conditions change. ORM targets 95 percent without manual adjustment, holding from day one through day 90 of the quarter and updating as the quarter progresses. ORM makes the underlying point plainly, that getting the forecast right in the final week does not help anyone because the quarter has already happened.
Adjust for the shape of the quarter
Weeks are not interchangeable. ORM observes that Q2 and Q4 usually run stronger than Q1 and Q3, and that the third month of a quarter outperforms the first two. A week eleven error of four percent means something very different from a week two error of four percent, so compare each week only against the same week in prior quarters.
The curve also diagnoses cause. Error that shrinks steadily through the quarter reflects normal resolution of uncertainty. Error that stays flat then collapses at the end usually reflects late category upgrades and quarter end deal loading rather than improved forecasting.
For the underlying measure, see forecast accuracy, and for the mechanics of producing an early number, see how to forecast revenue.
Frequently Asked Questions
How do you measure forecast accuracy by week of quarter?
Save a locked forecast snapshot every week. After the quarter closes, compare each snapshot against the actual result and plot the error by week number. Repeat across several quarters and average by week to get the curve.
What does a good accuracy curve look like?
High and flat from the opening weeks. A curve that starts far off and only converges in the final two weeks describes a team that reports the quarter rather than forecasts it, since by then the outcome is already determined.
Why does a single accuracy number hide this?
Most teams grade the last forecast they called, which is the easiest one to get right. Two teams can report identical quarter end accuracy while one knew its number in week two and the other found out in week twelve.
Does seasonality distort the curve?
Yes. Quarters are not shaped alike and the closing month of a quarter typically carries more volume than the first two, so accuracy in a given week is only comparable against the same week in other quarters.
Put these metrics to work
ORM builds custom revenue forecast models that turn concepts like forecast accuracy by week of quarter into prescriptive action for your team.
Schedule a Demo