"When will it be done?"
For ages this has been one of the first questions customers have.
Over time we have stepped away from traditional planning as it was unreliable and left behind estimation poker as it was not good enough. In Kanban systems, probabilistic forecasting turned out to be a good approach -- yet there are still open questions.
What if there are multiple points where work can enter (or leave) the system, if certain steps might be skipped, or if work may also flow backwards? Are there ways to do reliable forecasting in such a system? The power of maths may offer a way and in this talk you can learn how it works using Markov Chains.
Video & slides