Practically predictable – Forecasting delivery when nothing is certai
summary
Uses a throughput forecaster (cycle time = WIP / throughput) to produce probabilistic date ranges instead of single deadlines, pairs that with a risk burndown that quantifies likelihood and effort to reduce unknowns, and runs a three‑week spike period that cuts measured risk by 55%. Applies value‑stream mapping, Kanban sizing and a bit of Co‑Pilot help to halve cycle time in six months, secure £10.5m investment, and ship an MVP to a hospital that halves medicine loss on first release.
AI-written summary · about
chapters (10)
- 0:00Introduction to forecasting
- 1:26Case study of failed estimate
- 4:15Probabilistic forecasting tool
- 7:25Risk burndown for transparency
- 9:32Reducing risks in three weeks
- 9:53Breaking down big work
- 11:19Value stream mapping
- 13:28AI for efficiency gains
- 14:30Results and investment success
- 15:33Summary and key takeaways
related talks
43mPredictable Delivery
40mWhat are the odds? Transforming forecasts through probabilities
44mAgile metrics for predicting the future
44mPredictably Agile
41mYour Project Behaves Like a Hurricane--Forecast It Like One
23mThe Shape of Uncertainty
45mHow to not shoot yourself in the foot with estimation
51mGet Predictability Regardless of Estimates