In the world of software development, teams estimate backlog items in different ways. The most common approaches are story points and time-based estimates, which are then used to plan work into sprints of a fixed duration. After a few sprints, teams start to measure their velocity with some level of confidence. But that’s where the problem begins.
Estimations typically account for both effort and complexity. However, as a team becomes more familiar with a project, the perceived effort often decreases, even if the underlying complexity remains the same. Experience, domain knowledge, and better tooling make things faster. So what exactly are we measuring anymore?
It gets even messier when team composition changes. People leave, people join, and suddenly past estimates
lose their reliability.
Do we average things out?
Do we estimate for the worst case scenario to protect the sprint goal?
Or do we just accept that estimations are an illusion?
These are questions I, and many other developers, keep running into.
In the end, estimations are just another metric.
Stupid... but necessary.