When will these “Work Items” be completed? How confident are you in that date? Working in the IT/IS domain, these questions are very familiar to you, right? If you’re learning about pull systems and principles of flow, and how to effectively apply them to your workflow, in my opinion, you’re taking a big step toward creating a (more) predictable system and being able to confidently answer these questions.
However, if your efforts to answer these questions are still relying greatly, for example, on any of the following:
1) Estimating levels of effort or duration needed to complete every work item.
2) An historical “average” only (ex. avg. lead time, avg. throughput, etc.).
3) Summing “best case estimates” and summing “worst case estimates” to produce a “range.”
4) Using any method similar to 1, 2, or 3 and adding a “buffer.”
Then, you’re missing out on what I’ve found to be one of the greatest benefits of an effective pull system, which is the ability to forecast using a probabilistic approach. More specifically I’m referring to simple, yet still powerful, ways to create meaningful forecasts of when work items will be completed.
In this workshop we’ll begin by contrasting probabilistic approaches with deterministic approaches. Then, we’ll cover some basic concepts that will be used to create a set of simple tools that provide powerful probabilistic forecasts for answering when an individual work item (ex. story) might be completed, or when a set of work items (ex. stories making up a feature, or a number of features making up a release) will be completed. The examples will be based on real world data and require nothing more than an MS-Excel spreadsheet. No formal previous statistics knowledge is required, however, as a “heads-up” we’ll introduce and discuss the following concepts: scatter plots, percentiles, histograms, random numbers, and conclude with a very gentle introduction to (Monte Carlo) simulations.
