Historical Probability Models

There are exactly two ways to assign probability to a deal: ask the rep what they think, or calculate what history says. The first method produces forecasts with 23% average error. The second produces forecasts with 8-12% average error. The methodology is not complicated. Take four quarters of closed-won and closed-lost deals. Group by stage at the time of outcome. Calculate the win rate per stage. Stage 1: 12%. Stage 2: 24%. Stage 3: 41%. Stage 4: 67%. Stage 5: 84%. Those percentages become the probability assigned to every current deal at each stage. The rep's opinion is removed from the probability calculation. The rep's judgment is applied to deal strategy — a different and more appropriate use of their expertise.

1. Extract Historical Outcomes Pull every deal resolved in the last four quarters. Record the stage at which it was won or lost. Four quarters provides enough data for statistical significance in most organizations. Fewer than two quarters is unreliable.
2. Calculate Stage Conversion Rates For each stage, divide the number of deals won by the total deals that reached that stage. This is the stage probability. Update quarterly. The rates drift as your sales process, team composition, and market conditions change.
3. Apply to Current Pipeline Assign each current deal the probability that matches its stage. Multiply deal value by probability. Sum the pipeline. This is the weighted forecast. Compare it to your commit-based forecast. The gap between the two is the measure of your current forecasting bias.