A surprisingly simple recipe for predictable velocity (you won’t like it)
I recently noticed a rising urge to measure predictability in software teams. Allow me to explain how to achieve it:
Set a target velocity. Explain to your team that hitting this number as closely as possible means success.
Make them slap a generous buffer on every estimate they make. (This is critical.)
Deliver the precise amount of story points needed early in the sprint.
Do a victory lap around the office before embarking on a 5-day weekend.
I hope you see the issue.
Keep your delivery measures from becoming targets. Targets change incentives. Incentives change behaviors. Once we’re told that being predictable means success, we’ll do everything possible to reduce the risk of failure. That’s basic human nature.
Remember — you picked Agile for a reason. You were smart enough to recognize how little you know upfront. But you had a goal and were determined to walk through the fog of uncertainty one step at a time.
Embrace the fact that perfect predictability in a foggy forest is an illusion. As long as VUCA (volatility, uncertainty, complexity, ambiguity) exists, you’ll only get perfectly predictable by cooking the books.
PS: Unconvinced? Luckily, a grand case study spanning almost 50 years highlights the dysfunctions resulting from predictability being the target in an uncertain and complex environment — the 5-year economic plans of the Soviet Union. An excellent summary of the resulting crooked internal dynamics can be found here.
