What the three inputs actually mean
Three-point (PERT) estimation asks for three numbers instead of one: optimistic (best case, everything goes right), most likely (what you'd honestly expect), and pessimistic (worst case, things go wrong). Combining them into a single expected value - weighted toward the most-likely case - produces a defensible number along with an implicit confidence range, instead of one guess presented as certainty.
The formula
Expected value = (O + 4M + P) / 6
The most-likely estimate counts four times as much as either extreme, which is why the result usually lands close to your "normal" case rather than splitting the difference between best and worst. The spread between optimistic and pessimistic also gives you a standard deviation - (P − O) / 6 - which is what the range shown above the calculator comes from.
When this beats planning poker
Three-point estimation and planning poker solve a similar problem differently. Poker produces a single relative estimate through team consensus and discussion - it's built for sprint-ready stories, where the conversation surfaces hidden complexity. Three-point produces a probability-weighted number with an explicit confidence interval, which fits better when a stakeholder needs a schedule for one epic or feature, not a story-by-story sprint plan. Full comparison: three-point estimation vs. planning poker.
What a wide spread is telling you
If pessimistic is far from optimistic, the math still works - but a spread that's wide relative to the estimate itself is usually a signal that the work isn't well understood yet, not a math problem. That's worth a scoping conversation before the number goes into a schedule anyone commits to.