Why forecasts miss
Two failure modes dominate. Sandbagging: reps lowball to beat the number, so leadership flies blind. Happy ears: reps believe every deal closes, so the forecast is a wish list. Both come from forecasting on feelings.
Forecast three ways and compare
No single method is trustworthy alone. Run all three and look at the spread:
- Weighted pipeline. Each deal's value times its stage probability. Fast, but only as honest as your stage hygiene.
- Commit / Best-case / Pipeline. Reps categorize each deal. Commit = "I'd bet my job." Best-case = "realistic stretch." The gap between them is your risk.
- Historical run-rate. What you've closed per quarter recently, adjusted for trend. A sanity check against the bottom-up numbers.
When the three disagree wildly, that's the conversation to have, not which spreadsheet to believe.
Probability must come from data
Don't let stage probabilities be made up. Pull historical conversion: of all deals that reached "Proposal," what fraction actually closed? That is your proposal probability, usually far lower than the optimistic default the CRM ships with.
Inspect the deals that move the number
80% of forecast error lives in the largest 20% of deals. For every deal above a threshold, require: a confirmed close date, an identified economic buyer, and a mutual action plan. No mutual plan, no commit category.
Track forecast accuracy itself
Each quarter, log what you forecast versus what closed, by rep. Chronic over-forecasters and sandbaggers both reveal themselves over a few quarters, and that's a coaching signal worth more than any single deal.