A good metrics:
- Guides/Drives me. When I look at it, I know what to do (how to react to the data highlighted by the metrics)
- Positively influences (motivates) me to take steps towards improvement
- Has no ambiguity regarding how the data can be interpreted
- is Visual: Once understood, it should take less than 3 seconds to get the essence of the metrics
- Tracks real data; not something else which is easy to measure
- Is Automated: Eliminates waste and reduces the feedback latency
For example, xUnit report is a good metric. When I look at the results, I know what to do. If I have a failing test I know that needs to be fixed. It drives me positively to fix the test. If all tests are working, then it motivates me to add more tests. There are no 2 different ways to interpret the data. A failing test means program did not meet the expectations and a passing test means everything is working as expected. Also we are tracking real data generated automatically instead of guesswork.
In my experience focusing on 3 metrics at a time, is very effective. Once we start tracking more than 3 metrics, we loose focus and clarity.
How do you know which 3 metric to watch?
Usually, I use
ToC to figure out what is the biggest bottleneck in my system. Based on the bottleneck, I define 2-3 metrics that we want to measure as we try to address the bottleneck. Hopefully soon, something else will be the bottleneck and we’ll start measuring things related to that.
This is a main reason why I think we cannot have a set of metrics across the organization. Every team has different bottlenecks and they are at a different stage of evolution. Metrics should be dynamic and they should keep evolving.
