A friend asked for suggestions on a metric for backlog grooming. I’ve never written down these numbers, but this is the metric I use in my head.
When are changes happening to the acceptance criteria and acceptance scenarios of a story?
Sort these changes into the following bins:
1. When the story is first written.
That’s OK, as it may be needed to remember what the story is about.
2. Prior to 2 iterations before the story is developed.
Premature. This separates the story definition from the development. It uses stale information, inhibiting learning during development. It encourages a “throw it over the wall” mentality.
3. Within 2 iterations before the story is developed.
This is the sweet spot. The immediately preceding iteration is even sweeter.
4. After the development iteration starts.
This indicates that the story wasn’t really ready for development. Work harder at Example Mapping by the Three Amigos in the future.
5. After the story is accepted.
This indicates that not all the important scenarios were anticipated. Sometimes this is inevitable, but it’s a good indication to check your assumptions. Discuss this in the retrospective and seek double loop learning. “How can we improve our process so this doesn’t happen in the future?”
Graph a histogram of these bins. You want to see a sharp peak in bin 3. Don’t worry too much about bin 1 unless it’s higher than bin 3, or you’re finding that the software you’re creating doesn’t meet actual user needs.
