The costs are not symmetric
An over-forecast commits spend against revenue that never arrives. Headcount gets approved and campaigns get funded against a number that turns out to be wrong. None of that reverses inside a quarter, so the miss lands on margin and cash.
An under-forecast leaves money on the table in a quieter way. Capacity that was never approved cannot be built retroactively, and a team that beats its call every quarter loses the right to be believed when it asks for investment. Finance stops planning against the number and starts planning against its own adjustment of the number.
Where each one comes from
Over-forecasting comes from optimism baked into two fields. Close dates slide, and deal amounts sit above what those deals actually close for. ORM's position is that most deals close for less than the value they carry in the CRM, illustrated by a pipeline averaging $80,000 per deal against closed-won deals averaging $40,000. ORM also finds that 20% of the pipeline carrying in-quarter close dates on day one actually closes in that quarter, so a forecast that trusts day-one close dates is overstating by construction.
Under-forecasting is usually deliberate. Reps hold deals back to protect the next quarter's comp, managers apply a standing haircut on top of calls that were already conservative, and the two stack into a number nobody intended.
How to tell which one you have
| Signal | Points to |
|---|---|
| Signed error positive four quarters running | Over-forecasting |
| Close dates pushed inside the quarter on the same deals | Over-forecasting |
| Commit category converting near 100% | Under-forecasting |
| Quarter beats the call by a similar margin every time | Under-forecasting |
Fixing the direction
Direction is the fixable half of forecast error. Random spread requires better inputs and better models. A consistent lean requires a calibration you can compute from four quarters of history and apply.
Start with the inputs producing it. Tighten what qualifies a deal for commit, require a verifiable date change reason rather than a silent push, and reconcile pipeline amounts against realized amounts by segment. Then measure the signed error again, because a lean corrected at the input level should disappear rather than move. Track it against forecast accuracy, watch deal slippage as the leading indicator, and rebuild the calls themselves using sales forecasting criteria that do not depend on rep sentiment.
Frequently Asked Questions
Which is worse, over-forecasting or under-forecasting?
Over-forecasting is worse in the short run because spending is committed against revenue that never arrives, and headcount and infrastructure cannot be unwound inside a quarter. Under-forecasting costs less immediately and still costs real growth, since capacity that was never approved cannot be built retroactively.
How do you tell which direction your forecast leans?
Track signed error, forecast minus actual, over at least four quarters. Consistently positive means over-forecasting and consistently negative means under-forecasting. Absolute accuracy scores like MAPE cannot answer this because they discard the sign.
What causes most over-forecasting in B2B SaaS?
Two mechanics. Close dates that sellers keep pushing forward, and deal amounts carried at pipeline value rather than realized value. ORM's position is that most deals close for less than the value they carry in the CRM, illustrated by a pipeline averaging $80,000 per deal against closed-won deals averaging $40,000.
Does applying a haircut fix a forecast that leans high?
A haircut removes the symptom for one quarter. It does not fix the entry criteria, close date discipline, or amount hygiene producing the lean, and it stacks unpredictably when reps already discount their own calls. Correct the inputs, then verify the lean is gone.
Put these metrics to work
ORM builds custom revenue forecast models that turn concepts like over-forecasting vs under-forecasting into prescriptive action for your team.
Schedule a Demo