What is the PERT formula?

PERT mean = (Optimistic + 4 × Most Likely + Pessimistic) / 6. PERT sigma = (Pessimistic − Optimistic) / 6.

What does P80 mean in PERT estimation?

P80 is the estimate value you have an 80% chance of hitting or beating. It equals the PERT mean plus roughly 0.84 standard deviations, accounting for one-tailed normal-distribution risk.

Why should I use three-point estimation instead of a single number?

A single estimate hides uncertainty. Three-point estimation forces you to articulate the range, which surfaces risk early and gives clients a defensible confidence band rather than a guess.

[ Tool · PERT calculator ]

Three-point estimation.

Type three numbers — the best case, your honest guess, the worst case — and get a defensible estimate with a real confidence band. Same math the rest of Estix uses, free, no login required.

Inputs

Estimate the same task three times. Optimistic = the best-case time if everything goes right. Likely = your honest single-number estimate. Pessimistic = the worst case shy of a project-killing surprise.

Optimistic (O)

Most Likely (M)

Pessimistic (P)

[ Confidence target ]

[ Unit ]

mean (μ) = (O + 4M + P) / 6
σ = (P − O) / 6
estimate = μ + z · σ where z₅₀ = 0, z₈₀ ≈ 0.84, z₉₅ ≈ 1.64

[ P80 estimate · hours ]

11.0h

Mean (μ)

9.0

σ (sigma)

2.33

Variance (σ²)

5.44

Range (P − O)

14.0

Estimating a whole project, not one task? Estix.ai rolls these numbers up across hundreds of leaves automatically.

Try free

What is a PERT estimate?

PERT (Program Evaluation and Review Technique) is a way to estimate how long a task will take when there's uncertainty. Instead of one number, you give three: the optimistic case, your most-likely single-number estimate, and the pessimistic case. The PERT mean weighs the most-likely value four times as heavily as the extremes: μ = (O + 4M + P) / 6.

What's a confidence band?

The PERT sigma σ = (P − O) / 6 captures how much uncertainty there is. Multiply σ by a z-score and add it to the mean to get a confidence target: P50 = μ (50% likely), P80 ≈ μ + 0.84σ (80% likely), P95 ≈ μ + 1.64σ (95% likely). A higher target trades a longer estimate for a stronger guarantee — useful when scope is locked but delivery date matters.

When should I use three-point estimation?

Any project where uncertainty is real and visible — software, design, R&D, anything where you've been burned by a single-number estimate before. It surfaces risk early, gives clients a defensible range, and makes it harder to hide behind optimism.

Estimating a whole project?

Most projects have hundreds of tasks. PERT rolls up cleanly: variances add (σ²_project = Σ σ²_leaf), so the project-level σ is the square root of that sum. Doing it by hand for ~50 tasks works once; doing it for every quote is painful. Estix.ai automates the rollup, lets AI draft the scope tree from a brief, and exports a client-ready proposal with a real confidence band.