I just came across this article talking about the new segmentation support for the retention charts and also the consideration of user maturity now.
I’m a little bit confused about the last point.
From my understanding, it means that in the retention chart you don’t consider anymore that a user will renew until it has actually renewed, am I correct?
When I look at a retention chart without any segment, it seems like you’re still considering that a user will renew unless it cancels its subscription:
On the other hand, when I add a segment like country still for the last 90 days, I do only see the retention of the past 2 months, and no projection for the current month.
For me, the data dilemma you’re talking about in your article is occurring no matter if we add a segment or not. Would you have an explanation about these two different behaviors?
Best answer by Dan PannaschView original
Hey, thanks for reaching out! I think I see where I caused some confusion -- let me clarify:
I probably could’ve used a better word than “period” to describe this in the article. When I said “whole period” there, I was referring to the “last 12 months”, not necessarily the most recent period/month in the resulting chart. “To retain for the whole timeframe” would’ve been clearer.
In addition, that paragraph is only referring to the new segmentation options, which after rereading the article I don’t think is very clear. It’ll be edited shortly -- thanks for the call out! :)
Now, just to reaffirm the behavior and clear up any confusion there…
This is not true, but is a reasonable conclusion for you to have drawn. 😄 Here’s how it works now:
If any additional questions remain, feel free to add them and I’ll make sure we get everything squared away. Thanks!
Thank you for your answer, it’s clearer now.
I still have one question, how do we know how many subscriptions have been taken into account to calculate Month 1 rate and the Month 2 rate (cf. my screenshot)?
Because to my understanding, it can’t use the number of subscribers displayed, 8428, as some of them didn’t have the chance to renew one or two times since they installed the app during the last 90 days. So is there a way to know on how many subscriptions these numbers are calculated from?
Yeah unfortunately your understanding is correct that 8,428 would not be the correct denominator to use for periods after the first one. We’ll explore adding that context into the tooltip or in some other form in the future so that you can more easily understand the sample size that each period in the cohort represents.
Ok, thanks Dan for the answer!