What does a P80 cost estimate mean for a budget approval?

A P80 cost figure is a point at which 80% of the modeled total-cost outcomes are at or below that amount. It is conditional on the model's scope, data and assumptions. The label does not mean the project is 80% complete or that management can promise an 80% real-world chance of success without assessing the model's credibility.

GAO's cost-estimating guidance uses cumulative probability distributions to show confidence levels associated with different cost amounts. It also calls for risk and uncertainty analysis to be supported and updated. For a business approval, request the distribution and assumptions behind the percentile, not only a number copied onto a slide.

Identify whether the output covers total project cost, remaining cost or a particular work package. Confirm the price date, currency, scope and treatment of tax or inflation. A percentile for one subset cannot be treated as coverage for commitments excluded from the analysis.

Suppose a hypothetical model reports a base estimate of $500,000, P50 total cost of $540,000 and P80 total cost of $620,000. Moving from the base to P80 adds $120,000. Moving from an already approved $540,000 budget to that same P80 adds $80,000. Both statements can be correct if their starting points are clear.

Do not add the full $120,000 again to a budget that already contains the $40,000 uplift to P50. Identify which risks or allowances are embedded in the base and how they were treated in the model. A reconciliation should explain the complete proposed funding amount without stacking overlapping reserves.

These hypothetical figures are assumed outputs, not observed industry benchmarks. They illustrate the approval bridge only. A business needs its own supported model or an appropriately qualified estimate before assigning confidence labels to actual spending limits.

Ask how uncertain quantities, rates, schedule effects and dependencies were represented. Examine whether important common causes were included and whether scope changes are within the analysis. An elaborate simulation can repeatedly calculate the same incomplete description of the project.

Review the evidence for the distributions. Historical project data, supplier ranges and management judgment have different limitations. The number of simulation runs does not repair weak input assumptions. Have the estimator explain which uncertainties dominate the difference between the base and the higher percentile.

Keep schedule confidence separate. An 80th-percentile cost estimate and an 80th-percentile completion date do not, by themselves, establish an 80% probability of meeting both. Joint cost-and-time performance requires a connected assessment. Avoid combining separate labels into a stronger statement than the analysis supports.

Management still chooses the budget and its tolerance for an overrun. Show what risk remains above the chosen amount, what additional funding could be available and which scope or timing responses are feasible. The percentile is evidence for that decision rather than an automatic policy setting.

Phase the expected payments into the cash plan. A total allowance does not establish that money will be available when a supplier milestone arrives. State who can authorize use of the contingency and when emerging risks require a revised estimate or separate approval.

Retain the estimate date, model version and assumptions with the authorization. Refresh the analysis when material scope, schedule or evidence changes, and explain the movement from the prior distribution. The original P80 label should not continue unchanged after the project it described has materially changed.