Skip to main content

Hi everyone!

Was trying to create a waterfall retention chart on Amplitude, comparing retention by groups of sign ups (particularly, people who signed up in the same week).

I want to be able to quickly see how the sign up date affects retention.

What I’m trying to do is basically this, where the value on the rows is the sign up date / week:

Is there a way to automatically create segments, without having to create a cohort for each sign up week? Right now the only way around that I found is to create a manual cohort every week with users who had been new the week prior and add it to the chart.

Thanks!

Hi @lulufromspot! At the moment, there is no way to automatically create segments within the Retention Analysis and then apply them to the chart. Creating a cohort for each signup week and manually adding those cohorts to the segmentation module is the best workaround as of now. However, I think this is a really great idea! I’d encourage you to submit a feature request for this functionality here & categorize it as an “idea”. 


Hi @jmagg

I’m not sure if my question is similar, but I’m trying to track retention of users who used any events that is part of a given features (for instance Feature X has 5 events) I want to track users who triggered any of these 5 events and when returned did also triggered one of these events.

For now, I’m switching events to inactive except the 5 features event. Therefore, I can use Any Active Event. :grinning:

Let me know if if you have another workaround. 


Hi @Fouad! It sounds like you want to understand if a user did any of the 5 events that make up feature X and then returned to do one of these 5 events again. Is that right? If yes, I’d recommend creating a custom event (more info here). A custom event allows you to group multiple events together with “or” statements. So when one of the events that make up the custom event gets triggered, the custom event gets triggered. Here is an example chart I’ve made for you in the demo environment. What do you think?


Yay It works, I tried it. Cool thanks @jmagg 


Reply