Three-point estimation
Three-point estimation produces a single expected value from three estimates: optimistic (best-case), most likely, and pessimistic (worst-case). The most common formula is PERT — Expected = (O + 4M + P) / 6 — which weights the most-likely estimate four times more than the extremes. The technique was developed by the U.S. Navy in the 1950s for the Polaris missile project.
Three-point is useful when the underlying task has wide variance (research spikes, integration work with unknown upstream dependencies, fresh-domain work). Single-point estimates collapse that variance to false precision; three points preserve the band and let downstream planners reason about it. The PERT weighting assumes a beta-distribution-like shape — most outcomes cluster near the mode with long tails. For software estimation, three-point is most valuable as a discussion tool: forcing a team to articulate the optimistic and pessimistic scenarios surfaces risks that single-point estimates suppress. It is not, in practice, more accurate than calibrated expert judgment — Jørgensen 2004 found expert judgment ≈ formal models on most software tasks.
Related terms
- Cone of Uncertainty
The Cone of Uncertainty is a model of estimation variance over the lifecycle of a software project, first articulated by Barry Boehm in his 1981 book Software Engineering Economics.
- Planning fallacy
The planning fallacy is the cognitive bias toward predicting that one's own task will take less time than past tasks of similar scope.
- Story points
A story-point estimate is a unit-less measure of relative effort assigned to a user story.