What needs to be finished first
Three values (O, M, P) per estimate item are collected or will be collected in the same session, including assumptions for each value.
What needs to be ready before start
Estimation sheet with columns for item, O, M, P, expected value (PERT), standard deviation; estimate-item document; assumption list; calculator or sheet formula.
A facilitator who collects values and makes the formula transparent; an estimation team (1-8 people); a scribe for assumptions and risk classification.
Estimate-item description; assumptions per O/M/P; list of risks that drive P; desired confidence level (1, 2, or 3 sigma).
15-45 min
Prepare a sheet with formulas: expected value = (O + 4M + P) / 6, standard deviation = (P - O) / 6, range with ±1 sigma (68%), ±2 sigma (95%). Make each column definition visible.
The one question this method answers
What weighted expected value and confidence interval result from the three estimate values, and is that enough for the planning decision at hand?
Marker: Phase
| Step | Duration | Action | Hint |
|---|---|---|---|
1Phase 1: Capture three values | 10-20 min | Collect O, M, P per item with the estimation team (see Three-Point Estimation runsheet). Document assumptions for each value. | If O/M/P values are taken from a prior estimate, update assumptions. Outdated assumptions lead to a wrong expected value. |
2Phase 2: Apply formula | 5 min | Per item, calculate expected value with (O + 4M + P) / 6. Calculate standard deviation with (P - O) / 6. Enter values in the sheet. | 4M weighting is the core heuristic of the method. Equal weighting produces Three-Point, not PERT. |
3Phase 3: Derive confidence interval | 10 min | Calculate expected value ± 1 sigma (68% confidence) or ± 2 sigma (95%) and choose the level that fits the planning situation. | Normal distribution assumptions are sensitive for single items. The confidence interval is a guide, not a statistical guarantee. |
4Phase 4: Aggregate and align risk | 10-15 min | For project-level estimation, sum expected values. Calculate standard deviation of the total (square root of the sum of variances). Compare with the risk list: do assumptions match the risk assessment? | Summed standard deviation is only valid for independent items. With dependencies (all items share the same engineer), dispersion is higher and assumptions must be documented. |
What comes out at the end
Estimate table with columns for item, O, M, P, expected value (PERT), standard deviation, confidence interval, assumptions, and risks. Aggregated totals at the end of the table. Linked to the estimate-item document.
Create a snapshot for each estimation round with date and assumptions version. For re-estimation, add a new row and keep old values with status. Aggregated values must be reproducible from item-level data.
Compact working template for PERT Estimation with context, input, output artifacts, and next step.
Weights three-point estimates, typically with the formula O plus 4M plus P divided by 6.
When and for what do we use this method?
Which data, observations, decisions, or materials are available?
Short notes along the runsheet.
Owner, date, and success signal.
Concrete filled scenario, fictional example
Concrete filled scenario, fictional example
Estimate item: Migration from MariaDB to PostgreSQL for around 12,000 tenants. Unit: person-weeks.
| Item | O | M | P | PERT | Sigma | 95% range |
|---|---|---|---|---|---|---|
| Schema mapping | 1 | 3 | 6 | 3.2 | 0.8 | 1.6 - 4.8 |
| Data migration script | 2 | 5 | 12 | 5.7 | 1.7 | 2.3 - 9.1 |
| Test migrations | 1 | 2 | 5 | 2.3 | 0.7 | 0.9 - 3.7 |
| Cutover weekend | 0.5 | 1 | 3 | 1.3 | 0.4 | 0.5 - 2.1 |
| Total | 4.5 | 11 | 26 | 12.5 | 2.2 | 8.1 - 16.9 |
Assumptions: Items are treated as independent (shared engineer is a violation of this assumption; a 20% ramp is recommended).
Recommendation: Plan value 13 person-weeks, range 8-17, with a 10-20 dependent-task increase when adjusted for dependency.
Open risks: Compliance audit date is not fixed (spike @ben by 25.05.).
Recognize symptoms and steer against them
Expected value is calculated as (O+M+P)/3, and M weighting is missing.
Use formula (O+4M+P)/6 strictly. Without 4M weighting it is Three-Point, not PERT. Validate sheet formulas before the workshop.
The 95% confidence interval is communicated as a guaranteed range.
Confidence interval is based on normal distribution assumptions, which do not hold strictly for single items. Present it as guidance, not as statistical certainty.
Standard deviations are summed (instead of taking the square root of summed variances), or items share resources but are treated as independent.
Use square root of variance sum for aggregation. For shared resources, add a 10-30% sigma uplift. Document the assumption explicitly.
The table is full of numbers, but no one can explain why O is so low or P so high.
Assumptions per O/M/P are mandatory. A PERT formula without an assumptions log is number choreography. After re-estimation, no one will understand what was assumed.
Stakeholders treat the PERT expected value as a delivery promise while sigma and range disappear from presentations.
Consistently report expected value, range, and confidence level together. Stakeholders using only expected value misuse the method.
Done signals checkable in under a minute
Go to the profile for purpose, similar methods, and sources or continue to the next method in the catalog.