Because of the measurement issues noted above, a favourite way of analysing churn is via cohort analysis. This shows either customers numbers (or ARR) based on when the customer was acquired (the cohort). By looking at the performance of cohorts over time, it is easy to distinguish timing and mix issues on the reporting of aggregate level churn.
In interpreting these charts, rising revenue in a cohort indicates negative churn (a good thing) and falling revenue represents positive churn (a bad thing). The steeper the slope the higher the churn.