Why forecasts fail
Forecasts fail when they hide uncertainty, ignore dependencies, and don’t track variance trends.
A “date” is not a forecast unless it includes assumptions and confidence.
A simple confidence rubric
High: critical path stable; dependencies confirmed; variance within band.
Medium: one or more risks trending; assumptions plausible but unverified.
Low: critical path unstable; dependency breaches likely or already occurred.
What to measure
Schedule variance trend (planned vs. actual).
Cost variance trend (planned vs. actual).
Milestone hit rate and critical-path stability.
Operationalizing it weekly
Publish forecast changes with a reason code (dependency slip, scope change, resourcing shift, defect discovery, etc.).
Attach evidence links at the point of change (not buried in a deck).