The Science of not estimating
summary
Argues that story points are guesses and replaces them with probabilistic forecasting built from real delivery data: export dates from JIRA export, compute cycle times, apply bootstrapping and run Monte Carlo simulation (thousands of trials) to get percentiles (for example 80% chance to finish five stories by a date). Needs only ~10–11 past stories to start, adapts as team data changes, and produces a range of probabilities instead of a single, misleading number.
AI-written summary · about
chapters (15)
- 0:00Introduction and thanks
- 1:03Problems with story points
- 3:51Why estimates are guesses
- 6:38Stop guessing and alternatives
- 8:24Transition to probabilistic forecasting
- 11:13Bootstrapping and forecasting
- 15:27Starting with no data
- 18:18Tools and implementation
- 20:05Conclusion and next steps
- 24:01Q&A session begins
- 33:47Team frustrations with estimation
- 40:12Why we estimate and alternatives
- 43:26Forecasting for mixed teams
- 47:50Replacing estimation sessions
- 52:25Data requirements and risks
related talks
30mSoftware Estimation Made Easy
44mAgile metrics for predicting the future
46mEstimates and complexity: a tale in 42 and a half minutes
43mPredictable Delivery
47mCreating efficient, accurate, software estimations
05mIgnite - How to Use Scrum Poker to Slay Estimations
35mGo Big with Supercharged Agile
58mThe State of NoEstimates