Problem
We report month-over-month, which is dominated by event seasonality — Build, Ignite, Universe. A conference bump and real growth look the same, so the numbers do not support a decision.
What we'll do
Compare the same month across years: for example July 2025 vs July 2026, August 2025 vs August 2026, and so on. Comparing like-for-like months removes the seasonal effect and shows how usage is actually changing over time.
Apply this to users, operations, and success rate.
Depends on retained history — see the raw-event retention issue in this epic.
Done when
- The dashboard has a year-over-year view for users, operations, and success rate.
- We can state growth for a month without having to caveat it with "there was a conference".
Problem
We report month-over-month, which is dominated by event seasonality — Build, Ignite, Universe. A conference bump and real growth look the same, so the numbers do not support a decision.
What we'll do
Compare the same month across years: for example July 2025 vs July 2026, August 2025 vs August 2026, and so on. Comparing like-for-like months removes the seasonal effect and shows how usage is actually changing over time.
Apply this to users, operations, and success rate.
Depends on retained history — see the raw-event retention issue in this epic.
Done when