Calculating Churn Rate ATR
If you must use a churn rate (and this is unavoidable for most) it should be the simplest method that most literally represents the phenomenon without coloration or confusion.
There are two key rules to accomplish this:
1. Measure subscribers not dollars
The clearest and most essential churn is the loss of a customer (eg. subscriber, account, logo, etc.). It is much less obvious what it means for a dollar to "churn". Not that there are myriad problems with measuring dollar churn (eg. MRR, ARR) which we won't cover here.
The key issue is that dollar churn is always lower than subscriber churn because smaller and lower-paying subscribers always churn at higher rates. The result is that dollar churn always looks better than customer churn, and that means that it is actually concealing actual churn. The bigger the difference between dollar churn and subscriber churn, the more churn it is concealing.
→ But isn't dollar churn a more accurate financial metric?
The answer is: only if you are looking only at one moment frozen in time.
Remember that growth for subscription businesses is ultimately built on account expansion. That means that when there's a difference between subscriber churn rate and dollar churn rate, the company cannot expand smaller accounts.
In the world of SaaS, what matters for your valuation are your prospects for long-term growth. Using dollar churn essentially obscures the most important growth signal, long-term subscriber retention.
The same holds true for measuring the "user" churn rate. This is analogous to dollar churn and has all the same problems.
If you are measuring subscriber churn, then you are already on the right track with the numerator in the churn rate equation.
2. Get the denominator right
The simple rule for the denominator is only to include customers that could have churned in that period. This is the number of subscribers Available to Renew (ATR).
For example: if 12 customers churned in the period, and we have 1,200 total customers that appears to produce a 1% rate of churn. But, if only 120 customers were available to renew during the period then it's really a 10% rate of churn.
This method works for any period. For example, the annual churn rate would be calculated as the total number of subscribers churned during the year, divided by the total number of subscriber ATRs for the year, which is simply how many available subscriber renewals occurred during the year.
For example, if a subscriber joined halfway into the year, and they have a monthly renewal cycle, and did not churn, then that customer would have represented 6 subscriber available renewals for the year.
This simple approach partly addresses the distortion problem related to sales growth described above because we are only including customers that are available to renew. However, some distortion may always remain if there is sales acceleration or deceleration.