Most forecast accuracy reporting has a flattering design flaw. It measures the final forecast against the final actual, which is the one comparison guaranteed to look good.
Getting the forecast right in the last week of the quarter does not help anyone. By then the quarter has already happened.
Accuracy is a function of time, not a single number
A forecast that is 95 percent accurate in week thirteen and 60 percent accurate in week two is not a 95 percent accurate forecast. It is a description that becomes correct once the thing it describes is finished.
The value of a forecast lies entirely in the lead time it buys. Knowing the likely shape of the quarter on day one is what makes it possible to do something about it, and every meaningful intervention has a lead time attached:
- Reallocating marketing spend takes weeks to show up as pipeline. - Escalating a stalled deal requires enough runway for an executive conversation to matter. - Fixing a coverage gap in a segment means generating pipeline against a creation-to-close cycle you do not control. - Changing the hiring plan affects a quarter that has not started yet.
None of those are available in week thirteen. So a forecast that only becomes reliable at the end has converted a planning instrument into a reporting artifact.
What good looks like
Across the market, forecast accuracy on new and expansion revenue, excluding renewals, usually lands around 90 percent. That number sounds strong until you ask two follow-up questions: how much manual effort produced it, and when in the quarter did it become true.
The common pattern is heavy manual effort that produces a good final number and goes stale the moment conditions shift. ORM targets 95 percent without manual adjustment, and holds it from day 1 through day 90 of the quarter, updating as the quarter progresses so the number reflects current information rather than the assumptions in place when the quarter opened.
The distinction that matters there is not the extra five points. It is that the accuracy is available early and does not depend on someone maintaining it by hand.
Measuring it honestly
The fix is a measurement change before it is a modeling change, and it is not difficult.
Snapshot the forecast at fixed points in the quarter and store the snapshots. Day 1, day 30, day 60, day 90 is enough. At quarter close, compare every snapshot against the actual.
| Snapshot | What a healthy model looks like | What a reporting artifact looks like |
|---|---|---|
| Day 1 | Within a defensible band of the actual | Far off, treated as too early to know |
| Day 30 | Tightening, with the gap explained | Still far off |
| Day 60 | Close, differences attributable to named deals | Beginning to converge |
| Day 90 | Accurate | Accurate |
Teams that run this measurement for two quarters usually find the same thing: the model was not forecasting, it was tracking. Deals were being counted as they became certain, which is bookkeeping performed slightly in advance.
What makes early accuracy possible
Early accuracy requires the model to account for revenue that is not visible yet, because on day one a large share of the quarter is not in the CRM. That means modeling in-quarter created revenue rather than treating it as upside, described in the invisible pipeline, and applying the intra-quarter shape so that a low week-six reading is interpreted correctly rather than as a shortfall, described in the 13-week quarter.
It also requires the model to be responsive to change. A forecast built on assumptions that were true when the quarter opened will drift as conditions move, and the most common reason forecasts miss is exactly that: something changed in the business or the market and the model did not pick it up. That mechanism is covered in why SaaS forecasts miss.
The question worth asking in the next forecast review
Not "what is the number." Ask when the number became reliable, and what anyone did differently because of it.
If the honest answer is that it became reliable around week eleven and nobody changed a decision, the forecast is not producing value, whatever its final accuracy says. For the underlying definition see forecast accuracy.
Frequently Asked Questions
Why is a last-week forecast not useful?
Because by the last week the quarter has already happened. A forecast creates value by being actionable, and every action worth taking, reallocating spend, escalating deals, adjusting coverage, requires lead time that no longer exists in week thirteen.When should a forecast be accurate?
From day one. ORM targets 95 percent accuracy on new and expansion revenue and holds it from day 1 through day 90 of the quarter, updating as the quarter progresses rather than converging only at the end.How do I measure whether my forecast is accurate early?
Snapshot the forecast at fixed points, for example day 1, day 30, day 60 and day 90, and compare each snapshot against the eventual actual. A model that is only accurate in the final snapshot is describing the quarter rather than forecasting it.What does 90 percent forecast accuracy usually hide?
How much manual effort produced it and when in the quarter it became true. The common pattern is heavy manual work that produces a good final number and goes stale the moment conditions shift, which is a different thing from a model that is accurate early and stays accurate.What does a forecast that only tracks look like?
Its day-1 and day-30 snapshots are far from the actual and it converges late. Deals are counted as they become certain, which is bookkeeping performed slightly in advance rather than forecasting.Frequently Asked Questions
Why is a last-week forecast not useful?
Because by the last week the quarter has already happened. A forecast creates value by being actionable, and every action worth taking, reallocating spend, escalating deals, adjusting coverage, requires lead time that no longer exists in week thirteen.
When should a forecast be accurate?
From day one. ORM targets 95 percent accuracy on new and expansion revenue and holds it from day 1 through day 90 of the quarter, updating as the quarter progresses rather than converging only at the end.
How do I measure whether my forecast is accurate early?
Snapshot the forecast at fixed points, for example day 1, day 30, day 60 and day 90, and compare each snapshot against the eventual actual. A model that is only accurate in the final snapshot is describing the quarter rather than forecasting it.
What does 90 percent forecast accuracy usually hide?
How much manual effort produced it and when in the quarter it became true. The common pattern is heavy manual work that produces a good final number and goes stale the moment conditions shift, which is a different thing from a model that is accurate early and stays accurate.
What does a forecast that only tracks look like?
Its day-1 and day-30 snapshots are far from the actual and it converges late. Deals are counted as they become certain, which is bookkeeping performed slightly in advance rather than forecasting.
See how ORM turns these insights into action
ORM builds custom revenue forecast models for B2B SaaS companies. Not dashboards. Prescriptive analytics that tell you what to do next.
Schedule a Demo