Protocol

Explain forecast error by component

One error number can hide five different systems.

When it fits

  • A project was late and the only lesson is 'we underestimated.'

When to avoid it

  • Categories are models; keep them simple enough to use and refine them only when they change decisions.

Why it matters

After delivery, compare forecast with actual by components such as active work, wait/dependency time, rework, scope change and realized risks. Keep categories stable enough to compare across projects. Fix the component that repeatedly dominates instead of applying one global pad.

Steps

  1. The post-estimate review identifies the dominant source of forecast error.

An example

A two-day slip turns out to be almost entirely approval latency, so the next improvement targets the review queue rather than adding 20% to development effort.

Check your result

The post-estimate review identifies the dominant source of forecast error.

Keep this limit in mind

  • Categories are models; keep them simple enough to use and refine them only when they change decisions.

Connected ideas

Use before
Maintain a local forecast-error baseline

Evidence and sources

Supports

The 2026 Green Book recommends explicitly accounting for optimism bias in cost, benefit and duration estimates and using historical forecast errors from similar proposals where available.

The guidance is for UK public appraisal; the transferable principle is empirical correction, not a universal percentage uplift.

The Green Book (2026) · Optimism bias

Supports

Supplementary Green Book optimism-bias guidance recommends basing adjustments on data from past or similar projects and collecting local data to improve future estimates.

Reference classes must be genuinely comparable; irrelevant historical projects can make an estimate worse.

Supplementary Green Book Guidance: Optimism Bias · Introduction and making adjustments

All sources (2)