The definition of churn rate was the subject of a shareholders' lawsuit against Netflix in 2004. The shareholders accused Netflix of reporting an artificially low churn rate using:
Number of Customers churn / (Number of customers at beginning + number of customers gained).
Where the plaintiffs preferred:
Number of Customers churn / (Number of customers at beginning + number of customers at end of period)/2.
Netflix succeeded in having the suit dismissed, since there is no official way to calculate churn.
This has happened to me before in full web pages where there is no clear visual indication of the end of the page. This and makes me question the scrollbar decision from a usability perspective - there is no direct visual signal of progress.
That's very interesting context and all the more reason to spend time thinking about this stuff (or face the litigious consequences). I hadn't heard of this case before writing this article but I don't find the similar definitions all that surprising. We tried to stay pretty close to the sorts of definitions one might learn in an intro financial analysis course.
There isn't a good definition. At Dropbox, we try to answer the question: what percentage of signups 56 days ago were active 1-28 days ago? For larger time periods, just look at the sum of each day over the window (sum-of-retained/sum-of-signups. 28/56 day windows are useful to keep the same number of weekends.
This reports the most recent signups for which there is good data, but is lagging. You could look at another action that causes people to be retained that happens earlier in the funnel to run an actionable test in a reasonable period of time.
Looking at cohort analysis historically will get you a good understanding of the percentage of users that are active after N time periods.
What if there is seasonality in your Churn (lets say in your example monthly, which is very reasonable) and you are growing?
If you calculate at the wrong time you will severely skew the numbers.
-
Its my opinion that if you have daily churn numbers and you want to be the most accurate, simple formula is no longer viable. You should model daily churn against daily sales and create a revenue model. (takes about 2 minutes in excel, less if you have the data already in a spreadsheet).
-
If you want a simple formula, you should use the first two formulas you described in the article and just see how the numbers "feel".
In my opinion, anything beyond that introduces unnecessary levels of complexity that may actually make your modeling less valuable.
-
In any event, the article is great and really got me to think about churn again. Well done, and I really liked your thoughts.
I think we basically agree; you may just be taking issue with my very explicit representation of the formula. All that's really happening here is just a weighted average of daily churn rates. I really think you need to average over a period of time to deal with normal volatility.
While the number '30' is in the formulation is not intended to mean that this metric can only be measured for a month. Rather it's just there to normalize the metric to always be comparable to the monthly rate. It would be very reasonable to take this metric for a month and for every week in the month and see if any of the weeks are substantially higher or lower.
I would be cautious with using this, or anything based on daily churn, to be used in a formula for predictions. If you want to predict what customers will do over x days it is far better to measure what customers have done over x days. What you lose in currency you more than make up for in having taken a direct measurement. Though I would happily use this metric to play in a model with computed weights.
http://www.globenewswire.com/newsroom/news.html?d=62086
http://www.docstoc.com/docs/33875708/In-Re-Netflix-Inc-Secur...