The placebo of dashboards¶

The dashboard is seen by many as an ultimate solution to all important aspects  of software development. More accurate estimations, better team performance, better product quality, the answer to everything seems to be a dashboard.

A tool collecting data from multiple sources, visualizing it and making easily accessible is very important if not a prerequisite of a successful software project, but the devil is in the details.

A dashboard is only as good as short is the feedback loop it creates.

The shorter feedback loop, and the closer to the real life action (code commit, test run, deployment) the better the dashboard. Unfortunately many solutions offer nice visualizations, which are not actionable. In this case the analysis required to translate a metric to a real action takes experience and time, and sometimes multiple people, which seriously limits the advantage of a dashboard.

In the worst case scenario, confusing metric can have negative impact, since all false positives cost time, require explanation, and create misconceptions within the organization.

We need cockpits not dashboards.

If dashboards are collecting, analyzing and visualizing data, cockpits enable actions. Even if not fancy visually, a list comparing code merges with test results, provides very short feedback loop, and helps to improve the code quality more efficiently than a set of charts reflecting number of failed tests or defects on the backlog.