Sales forecasts are consistently wrong in the same direction, and the error survives repeated attempts to fix it. The cause sits in the incentives rather than in the estimating.
Optimism and pessimism carry different penalties
A representative who forecasts a deal that does not close has a difficult conversation at quarter end. One who omits a deal that closes looks like they lacked visibility.
Neither error is free, but the first can be explained by circumstances outside their control and the second suggests they do not understand their own accounts.
Given that asymmetry, including a marginal deal is the safer choice, and the aggregate of many such choices is a systematically inflated pipeline.
Stage definitions describe seller activity, not buyer commitment
Most pipeline stages are defined by what the seller has done: a demonstration delivered, a proposal sent, a negotiation opened.
None of those establish that the buyer has decided anything, so a deal can advance through the entire process on seller activity alone.
Probabilities attached to those stages then encode an assumption about buyer intent that the stage definition never tested.
Managers adjust the number rather than the input
Experienced sales leaders know the pipeline is optimistic and apply a discount based on historic conversion, which produces a usable number for the current quarter.
The adjustment is applied at the top and never returns to the representatives, so the underlying estimates are never corrected and the bias is preserved.
This also hides which deals were wrong, since the discount is applied in aggregate and cannot identify the specific opportunities that inflated it.
Slipped deals distort the following quarter
A deal that does not close is usually moved forward rather than removed, on the reasonable grounds that the buyer has not said no.
The next quarter therefore begins carrying opportunities that already failed to close once, and their presence makes the target look more achievable than it is.
Deals can carry across several quarters this way, accumulating in the pipeline while contributing nothing, until an eventual clean-out that appears as a sudden collapse.
Evidence-based stages change the behaviour
Defining each stage by something the buyer did — a stakeholder introduced, a procurement process started, a security review scheduled — makes advancement depend on observable facts.
Representatives can no longer advance a deal by working harder on it, which removes the mechanism that produced the bias in the first place.
Reported pipeline falls immediately when this is introduced, which is the main reason organisations recognise the problem and decline the remedy.