Completion Analysis
Predicts a project's actual completion date from velocity, blockers, and sprint history.
Completion Analysis predicts when a project will actually finish, returning a predicted date, the variance from the committed deadline, and a confidence level based on real project state.
How it works
Taro builds the prediction from three layers of data. First, team velocity: the average tasks completed per sprint over the last 30, 60, and 90 days, weighted toward recent sprints. Second, known blockers and unresolved dependencies, where each blocked task adds delay proportional to how deadline-critical it is and how long blockers have historically stayed open for the team. Third, sprint history patterns, including the team's typical mid-sprint scope additions. These signals combine into a single predicted finish date. The prediction recalculates every time the project state changes, and Taro feeds actual delivery dates back into the model to self-correct over time.
Key capabilities
- Predicted finish date derived from current velocity, blockers, and sprint history rather than the planned date.
- Variance against the committed deadline, surfaced as soon as it becomes visible.
- Confidence level reflecting how much consistent historical data backs the prediction. Typical ranges: 55 to 70 percent with 1 to 2 sprints, 75 to 85 percent with 4 to 6 sprints, 85 to 92 percent with 8 or more sprints.
- Blockers counted and weighted by deadline proximity, with external blockers weighted higher than internal ones.
- Automatic recalculation on every project state change.
- Manual anomaly flagging to exclude unusual sprints from the velocity baseline.
Related intelligence
Workload Distribution
Risk Prediction
Treat a lower-confidence prediction as a directional signal rather than a firm commitment, and use higher-confidence predictions when setting stakeholder dates.