[HN Gopher] How to Measure Cohort Retention
___________________________________________________________________
How to Measure Cohort Retention
Author : gmays
Score : 41 points
Date : 2022-11-19 03:22 UTC (19 hours ago)
(HTM) web link (www.lennysnewsletter.com)
(TXT) w3m dump (www.lennysnewsletter.com)
| kqr wrote:
| One thing people often forget about retention is that the data
| are censored. When customers stop being active, you know it, but
| you don't know when the customers that are active today will stop
| being active. If you don't account for that ignorance, you get
| data that are biased one way or the other.
|
| Time-to-event ("survival") analysis has tools to deal with this,
| which are eerily underdiscussed. Here's one example: https://two-
| wrongs.com/survival-analysis-for-customer-retent...
|
| Another common problem, especially when analysing what effect
| interventions have on retention, is ignoring selection bias. Most
| interventions are of the form where they will be preferentially
| offered to or accepted by users who are either more or less
| satisfied than the average. This will in turn make it seem like
| retention is positively or negatively affected by the
| intervention, when really it's just selecting for user groups
| that were different to start out with.
| sokoloff wrote:
| > When customers stop being active, you know it
|
| That is true for time-based (eg. monthly) subscriptions, but
| not for the general repeat customer case. (Home Depot or
| Starbucks would have no timely idea if I suddenly decided to
| boycott them, while Comcast would.)
| kqr wrote:
| Right. I was going off the definition in the article that
| active customers are those that have performed an action
| within a time span, i.e. at the end of each such time span
| you can definitely count customers as "no longer active" or
| "still active... for now".
| Jorge1o1 wrote:
| While this article is mostly focused on
| collecting/gathering/visualizing cohort metrics, rather than on
| analysis, I would like to plug my college professor Peter Fader's
| research on empirical Bayes modeling and analyzing Customer
| Lifetime Value.
|
| One thing I learned from his class is that survival rates will
| naturally trend upward over time, which marketers erroneously
| attribute to (1) improving the product, (2) better customer
| service, (3) network effects / lock-in, etc.
|
| However, if you have a heterogeneous customer base with latently
| better and worse customers, inevitably your worse customers will
| churn before your better customers, showing that "decrease" in
| churn.
|
| These models also let you do cool things like conditional
| expectation: "if a customer has survived 13 months, what's the
| probability they churn in the 14th?"
|
| Here's a paper of his from 2004:
| https://repository.upenn.edu/cgi/viewcontent.cgi?article=141...
| kqr wrote:
| This sounds a lot like the type of survival analysis mentioned
| in the other comment. The Kaplan--Meier estimator can be made
| conditional by, well, conditioning on earlier parts of the
| curve.
|
| If you count not "time-to-event" to "total-revenue-to-event"
| you get a lifetime value estimator!
___________________________________________________________________
(page generated 2022-11-19 23:01 UTC)