What is the best metric for developer productivity?
How do you measure your own and your team's productivity. Points, cycle time, churn rate, something else?
If you could see only one data point or take one action each day to improve your own or your team's productivity what would it be?
2 comments
Personally I think this is a fantastic question and one I'm very interested in seeing what more experienced developers and managers have to say.
The more I manage people (especially developers) the more I'm drawn to the adage of 'a metric that becomes a target ceases to be a useful metric' but at the same time many companies put great stock in trying to develop measurable KPIs - something that I've greatly struggled with through several roles as a development manager.
If I was going to land on anything currently it would probably be something like 'Features released within a iteratively improved estimated timeframe'. In other words a metric that focused on an improvement between now the and last time the measurement was made but not towards a specific target. Also a metric that made people realise that the focus is more around them working as a team and even across multiple teams to achieve a goal.
The more I manage people (especially developers) the more I'm drawn to the adage of 'a metric that becomes a target ceases to be a useful metric' but at the same time many companies put great stock in trying to develop measurable KPIs - something that I've greatly struggled with through several roles as a development manager.
If I was going to land on anything currently it would probably be something like 'Features released within a iteratively improved estimated timeframe'. In other words a metric that focused on an improvement between now the and last time the measurement was made but not towards a specific target. Also a metric that made people realise that the focus is more around them working as a team and even across multiple teams to achieve a goal.
I wouldn't measure one thing. I'd measure a whole heap of different metrics, do some sort of (possibly nonlinear) mapping so bigger numbers indicate (intuitively) better performance, then take deciles. I'd measure performance by how high each developers' worst metric is. I.e. look for consistency across all metrics and ignore high performance in a small number of areas (to make it harder to game). Also, it might be worth training a model to predict that performance metric from other unstructured data -- again to make it harder to explicitly game the system.