How to measure and how to balance
Quite a few organizations are in a middle of a paradigm shift from linear SW development model to agile or lean development. This brings extra challenges for measuring SW development and for assessing and knowing actual status of a situation.
Agile and Lean measures flow. If the development team can provide a steady and reliable flow of features, all is fine. Central measurement is calendar time from idea to provisioning of outcome. Linear state-gate driven development measures accrued and earned hours against estimated workload. When the gap between accrued hours and earned hours split, the project is in trouble.
Both models have good rationale in them and of course, pitfalls too. When managing a program, the hybrid model brings a multitude of status related problems. Too often the teams are not disciplined enough to estimate their work properly. If they are matured enough to give reliable estimations in story points, they typically forget a whole bunch of work. There is also plenty of non-visible work like support to other teams, that has not been recorded. Thus, the estimated velocity is not accurate enough, it is less than the team thinks.
The tradition of recording accrued work has been forgotten, as the flow is the king. When the teams slip from delivering their stories and features in reserved time, the flow gets broken. The team must start estimating how much the slippage eats velocity from following sprint.
The development team concerns only about the next sprint of cycle. Features rule. This leads to a false assumption that the team can deliver when it has finished off the feature development. The DevOps movement is trying to solve this issue. The commissioning phase tasks should be taken into Feature DoR -statements: the feature should not be done until it is fully releasable.
One reason why agile development struggle is that there is not enough analysis and planning resources. The analysis becomes a bottleneck and the teams fail to understand what is being expected from it. One reason to this is that the amount of the analysis work has not been measured nor estimated.
In transition phase between the paradigms, the organizations need to be extra careful to enforce disciplines. Even though some of them feel like 90’s, they are still proving their strength. Start collecting accruals and measuring the size of the project. Manage by data not by hunch.
Mitro Kivinen, FiSMA Executive Director