Stage-weighted vs. cohort forecasting

Stage-weighted forecasting sizes your projection deal by deal, using where each one sits in your pipeline today. Cohort forecasting sizes it group by group, using how similar deals have historically converted over time. Both are legitimate, both are simple enough to compute by hand, and both quietly break in different circumstances — which is the real reason to understand how each one works before picking one.

What stage-weighted forecasting does

Take every open deal, multiply its amount by the historical win rate of the stage it currently sits in, and sum the results. Three deals: $50,000 in Discovery (a stage that historically converts 10% of the time), $80,000 in Proposal (35%), and $120,000 in Negotiation (60%). Weighted value: $50,000 × 0.10 + $80,000 × 0.35 + $120,000 × 0.60 = $5,000 + $28,000 + $72,000 = $105,000. That number reflects exactly where the pipeline stands right now — its whole strength is that it is current.

What cohort forecasting does

Instead of weighting by stage, group deals by a shared starting point — most often the month they entered the pipeline — and apply that cohort’s own historical close rate once enough of it has played out. If deals created in a given month have converted at 22% on average across the last several cohorts, a new cohort of $500,000 in first-month pipeline projects to roughly $110,000 of eventual bookings. The strength here is the opposite of stage-weighting’s: it does not care what stage anything claims to be in today, only how deals like it have actually behaved once they were given time to play out.

Where each one breaks

Stage-weighted forecasting is only as honest as the stage data under it. When deals advance because a rep is optimistic rather than because an exit criterion was actually met, the weighting inherits the inflation — a deal sitting in “Negotiation” that has never had a second call is counted as if it earned that 60% rate.

Cohort forecasting has the opposite failure: it needs enough closed history per cohort to be stable, and a young cohort with only a handful of deals can swing wildly on one early win or loss. It also assumes the current pipeline resembles the historical one — if you just opened a new segment or changed your ICP, last year’s cohort behavior is a worse guide than it looks.

Why picking one is the wrong question

Neither approach is more “correct” — they answer different questions from different data, and each is honest about a different failure mode. That is why Pipemetry’s revenue forecasting software offers a model selector on Pro rather than committing you to one approach: a cohort heuristic, a stage-weighted pipeline, an ML deal score, and an ensemble that blends them, with every model’s inputs and assumptions inspectable rather than hidden behind a single number. For the general principle behind showing your work instead of trusting a black box, see how to forecast revenue without a black box.

The practical version

If your stage definitions are clean and enforced with real exit criteria, stage-weighted gives you a live read that updates the moment a deal moves. If your stages are looser but you have a few years of closed history, cohort forecasting will be more honest about what is actually likely. Most teams have some of each condition, which is the real argument for a model you can switch — and inspect — rather than a single fixed formula applied to a pipeline it was never built for.