Thanks for reaching out here @arshiagm. I was able to find this thread on dynamic cohorts that might be relevant to you. Please let us know if this helps! 

Hey @arshiagm 

This could be tricky since the most common way is using the “had been new” any time “during Last 7 days”. This approach will generate overlaps due to the fact that as we move ahead in time, some of the new users will still be consider in the time period whenever the cohort gets computed.

One approach could be to define a date you want to send the email and make a static cohort for just the time bound 7 days prior to this date.

Add your power user criteria to the above clause and compute the cohort on Jan 8th for the email.

For the next email batch, change the definition to -

- and then send the email on Jan 15th.

Another approach could be to capture the email sent as a user prop / event and exclude the users who have received this email. You can then use the Last x days clause with the exclusion clause for users who have already received the email for being a power user . This will ensure that you have new users in the cohort at each computation if they satisfy all clauses.

By no means these are the ideal solutions and I could be missing some easier and obvious approach. I’ll share here if I find any or see if other members have more ideas.

I’m still not sure how to limit the users to just 30 unless you have another level of filtering.

Hope this helps.

@Saish Redkar thanks for your suggestions.

Heres my original way I was defining the cohort:

This is what I adjusted it to after getting your feedback.

The thing is this cohort is intended to target power users who are not always necessarily new. There is also too small of a number of power users who signed up in the last 30 days. New users will be a different cohort.

Is there a way to do this without just limiting it down to new users?

Btw sync is on with a defined cadence so ideally this should work on it’s own without manual work. Thank you!

Hello @arshiagm Thanks for your post! Happy to help.

Saish’s approach is the best option to solve for your use case if the query needs to be done on a week by week basis. Additionally, you could also try using an offset (i.e. in the last 7d ays offset by 7 days) to make the cohort more dynamic. Here is an example from our Amplitude demo:

https://analytics.amplitude.com/demo/cohort/nt9jlup 

I understand that by “new power user” you do not mean “new” as defined by Amplitude (link). In that case, I believe you are looking for users who became a power user within a particular week. In that case, what you can do is remove the filter for “new” as defined by Amplitude, then create individual weekly cohorts but exclude the previous weeks cohort from the current weeks cohort. Here is an example: https://analytics.amplitude.com/demo/cohort/lzpwvhq.

I hope this helps! 

@arshiagm I would try out Jacenda’s approach here.

@Jacenda Davis thanks for your help. It’s going to be a new definition of power user basically if I use your suggestion but perhaps that’s ok if that’s the only solution. I need 30 new added per week automatically because there is a automatic sync. Customer.io already has a way to ensure the same person doesn’t get two emails. In this case, what’s the reason to go weekly?

I want to set a cohort that gets 30 new power users every week without needing to adjust it.