Test Coverage and Fermentation

Measuring test coverage makes sense - but maybe not in the way you think. Here's why:

Test Automation coverage is like fermentation bubbles

1. Seeing it doesn't mean things are going to be fine
- but not seeing it may indicate that the process itself may not yield good results.

2. could see it and still get an unconsumable product:
Coverage only tells you that code was executed by tests, not whether there are quality risks or issues.

3. When you don't see it on new stuff, this most certainly tells you that you'll have problems:
Your testing process and your development are out of sync, and that's a pretty good indicator that your development process itself is low quality.

It also tells you that the team is definitely not practicing TDD - because just like fermentation bubbles, test coverage is a side product, not a goal, of TDD.

If you ran a pickle factory - what do you think would happen if you set a target for amount of fermentation bubbles?

Focus on the pickles - not the bubbles.
But when there are no bubbles (tests) - you need to investigate.