Problem
Our dashboards pivot on known templates. Users are increasingly starting from custom templates or their own codebase, so the template pivot is going stale and a growing share of usage is unattributed.
What we'll do
Cohort users by how the project was initialized, and use that as a first-class dimension:
| Cohort |
Definition |
| Known template |
Recognized template |
| Blank template |
No metadata in azure.yaml |
| Unknown template |
Metadata present in azure.yaml, but not recognized by azd |
| Own codebase |
User brought their own code |
Done when
- The cohort dimension exists in the cooked tables and in the dashboard.
- Usage, success rate, and retention can be reported per cohort.
- We can see the trend of each cohort over time.
Problem
Our dashboards pivot on known templates. Users are increasingly starting from custom templates or their own codebase, so the template pivot is going stale and a growing share of usage is unattributed.
What we'll do
Cohort users by how the project was initialized, and use that as a first-class dimension:
azure.yamlazure.yaml, but not recognized by azdDone when