Glossary
Three-Point Estimate
An estimating technique that captures optimistic, most likely, and pessimistic values for a cost or duration, rather than committing to a single figure.
A three-point estimate replaces a single deterministic value with three: the optimistic value (O), the most likely value (M), and the pessimistic value (P). These three values define a range that reflects the estimator's genuine uncertainty about the outcome. They are the fundamental inputs to Monte Carlo simulation — every activity duration in an SRA and every cost element in a CRA is expressed as a three-point estimate. The simplest way to combine them into a single representative value is the PERT formula: (O + 4M + P) / 6, which gives more weight to the most likely value.
Several variants of the weighting formula are in circulation. The classical PERT formula (O + 4M + P) / 6 derives from the beta distribution and is the version cited in the Project Management Body of Knowledge. A simpler weighted variant — (O + 3M + P) / 5 — appears in some UK training materials and procurement guidance, and gives slightly less weight to the most likely value (60% vs 67%). Triangular weighting (O + M + P) / 3 gives equal weight to all three points and is rarely used in cost or schedule risk analysis because it ignores the empirical evidence that real activity durations cluster around the most likely value rather than the simple mean. Which formula you use matters less than people often assume: when the inputs are well-calibrated and a Monte Carlo simulation is run over the full distribution, the choice of point-estimate formula is an intermediate artefact, not the headline output.
Three-point estimates are important because they force estimators to acknowledge and quantify uncertainty rather than committing prematurely to a false precision. A single-point estimate of 10 weeks says nothing about whether that could be 8 weeks or 14 weeks. A three-point estimate of 8 / 10 / 16 weeks tells a very different story — one where the upside is modest but the downside is significant. This asymmetry is typical of real project estimates and is invisible if only the most likely value is reported.
The hardest part of three-point estimating is getting calibrated inputs. People are naturally bad at estimating probability ranges — we tend to anchor on the most likely value and set the extremes too close to it, producing ranges that are too narrow. This is especially common under schedule or commercial pressure. A practical test: ask the estimator whether they genuinely believe there is only a 5–10% chance of the actual value falling outside their stated range. If the answer is no, the range needs to be wider. For critical activities or cost elements, independent review or comparison against historical data is the best check.
Frequently asked
- What is a three-point estimate?
- A three-point estimate replaces a single deterministic duration or cost figure with three values: optimistic (O — the best plausible case), most likely (M — the mode), and pessimistic (P — the worst plausible case). These three values define a probability distribution for the uncertain quantity. Three-point estimates are the primary inputs to Monte Carlo simulations in schedule and cost risk analysis, and they allow a model to represent the asymmetric uncertainty that is typical in project estimating — most activities can overrun by more than they can underrun.
- What is the formula for a three-point estimate?
- Two formulas are in common use. The PERT (beta distribution) formula gives a weighted mean: (O + 4M + P) ÷ 6, which weights the most likely value four times. The triangular distribution formula gives an unweighted mean: (O + M + P) ÷ 3. PERT is the traditional formula used in programme scheduling; the triangular distribution is more common in cost risk models and is the default in tools like @RISK and Crystal Ball. The PERT formula produces a narrower range than the triangular and is generally considered to understate uncertainty on complex projects.
- What is the difference between PERT and triangular distribution in three-point estimates?
- Both use the same three input values (O, M, P) but weight them differently. The PERT beta distribution concentrates probability around the most likely value, producing a smooth bell-shaped curve and an expected value of (O + 4M + P) ÷ 6. The triangular distribution gives equal weight to all three inputs and produces a triangular probability density function with an expected value of (O + M + P) ÷ 3. In practice, the triangular distribution is preferred for cost risk because it better reflects the fat tails observed in actual cost overruns; PERT remains standard in many scheduling tools.
- How should you calibrate three-point estimates?
- The most common calibration failure is anchoring — estimators set the pessimistic value only slightly above the most likely, producing falsely narrow distributions. Good calibration asks: what would a realistic worst case look like if everything that could go wrong did? Reference class data (historical outturn distributions for similar activities) should be used to sense-check the range. A useful heuristic: for construction activities, the P − O range should typically be at least 30–50% of the most likely value on a complex project. Calibration workshops with independent facilitation consistently produce better-calibrated estimates than individuals working alone.
Related terms
Putting these techniques into practice?
SOMA provides independent project controls consultancy for UK programmes. We can help you apply QRA, EVM, schedule risk analysis, and more.