A rep sandbags a deal to guarantee they beat quota next quarter. Their manager pads the team number to have a buffer when the VP asks for more. The VP shaves the roll-up because the board always discounts whatever gets submitted anyway. By the time a number reaches the boardroom, it has passed through three or four people who each had a rational, self-protective reason to move it — and somehow everyone acts surprised when it doesn't match reality.

Nobody is lying, and the number is still wrong

The easy explanation for a bad forecast is that someone was dishonest. The more accurate one is that the forecasting process, as most orgs run it, rewards exactly the behavior that produces an inaccurate number. A rep who forecasts precisely and occasionally misses looks worse in a pipeline review than a rep who reliably sandbags and always "beats" their number. A manager who submits an unpadded team forecast has no room left if one deal slips. Every individual is optimizing rationally for how they're evaluated. The aggregate result is a number that reflects the incentive structure more than it reflects the pipeline.

This is why forecast accuracy rarely improves just by asking people to be more honest. The incentive to distort was never about honesty in the first place — it's about self-preservation inside a system that punishes precision and rewards a comfortable margin of error, as long as that margin runs in the safe direction.

Where the political layer gets added

The negotiation doesn't stop at the rep level. A VP walking into a board meeting has their own incentive to manage expectations — committing to a number they're confident they can beat, rather than the number that's actually most likely. The board, in turn, often applies its own haircut to whatever gets presented, because experienced board members have learned that submitted forecasts run optimistic. Each layer is quietly correcting for the layer below it, and the correction factors themselves are rarely made explicit or consistent quarter to quarter.

The result is a forecast that isn't really one number moving through an org — it's several different numbers, each shaped by what the person presenting it needed the board or their manager to believe. Two companies with identical pipelines can report wildly different forecasts purely because their internal politics apply different correction factors at different layers, which makes cross-company forecast comparisons close to meaningless without knowing the internal negotiation that produced each one.

What reduces the distortion

The fix isn't a stricter policy demanding honesty, since the incentive to distort survives any policy that doesn't change what actually gets rewarded. It's separating forecast accuracy from performance evaluation as much as an org realistically can — tracking and rewarding calibration (how often a rep's stated confidence matches the outcome) as its own metric, distinct from whether they hit the number itself. A rep who forecasts a 60% deal at 60% and it closes 60% of the time across a large enough sample is forecasting well, even on the quarters it doesn't close.

This requires genuine cultural buy-in, not just a new field in the CRM — a manager has to actually reward a rep for a well-calibrated miss the same way they'd reward a beat, or the incentive to distort simply reasserts itself the next quarter. Few orgs get all the way there, but even a partial shift — tracking calibration alongside attainment, without fully decoupling comp from it — starts to surface which parts of the funnel are genuinely predictable and which have been padded into false confidence for years.

None of this eliminates uncertainty in the pipeline — it just stops adding a second, avoidable layer of distortion on top of it. The same principle applies at the very top of the funnel: a forecast is only as trustworthy as the pipeline quality feeding it, and pipeline built on real signal rather than volume for its own sake is easier to forecast honestly, because the deals in it were closer to genuine fits from the start. That's the upstream half of the forecasting problem, and it's worth fixing before trying to fix the politics downstream of it.