Category: Metrics

Metrics are a good tool for data-driven decision making. Standardized¹ metrics and metric sets, such as DORA, have become popular with tech-managers. Maybe the idea of the metrics has been spread with (the same) consultants in the field, or maybe the managers just like comparing their things to the things of others…

Well-defined metrics enable comparing values between teams, locations, and organizations to easily see who is doing the best and who the worst. Look – mine is bigger than yours?

Most likely, you shouldn’t be bothered with such comparisons. Comparing the deployment frequency of a software that controls a nuclear power plant to a tic-tac-toe mobile game is possibly slightly absurd.

The generalized target values for “good” and “bad” in these metrics may not be relevant for your project. As mentioned somewhere before, focusing only on the metrics may mislead you to master the metrics but not your product.

Having your own targets and following your own progress even with standardized metrics is allowed – maybe even recommended!

1. Standardized but likely vague metrics ↩︎

Defining and measuring metrics is easy. Adding yet another metric just to add yet another metric happens subconsciously. It’s easy to focus on the values of the metrics and make decisions or to measure success only by using them. And it’s easy to forget why are they being measured in the first place – if there even is any.

What exactly is being measured, why is it measured, how is it measured, what does a change in the value mean, and what’s the target for the value. These are the questions often forgotten or left unanswered.

Is it something to celebrate, when nothing relevant is being created but superficial work is being done faster, better, in schedule, and with great style – with perfect metrics? A very fast car without brakes and steering is not valuable for most of the drivers.

Metrics can be a good servant but they are not a good leader. Make sure you master them before you end up as a servant.

It’s not always trivial to determine the impact, or the lack of impact, of a change. However numbers don’t lie and we humans are used to be measuring everything except t-shirts with numbers. Measuring something in a numeric form before and after a change should indicate whether the change is good, bad, or ugly, right?

Metrics are those quantitative assessments of relevant items. Metrics are simple; when the value of a metric changes to one direction it’s good, and when the value changes to the opposite direction it’s bad. Assessing the metric over and over again in different times and analyzing the assessed results one could argue if things are getting better, worse, or being stable. And everyone wants things to be better!

To be able to compare the evolution of the metric the measurements of the value needs to be well-defined and repeatable.

But what actually should be measured?