Hello Nate ! 

You can use the “retention” chart to display the new users (using Amplitude’s “New user” event as the starting event) who returned at a later date (using “weekly” to get the retention per week). Here is an exemple using Amplitude demo: https://analytics.amplitude.com/demo/chart/new/9s7jn2n?source=copy+url

Hope I could help 
Matrak

@Matrak 

Thank you so much!

I have a follow up question:

On day 3 i have 6 out of 69

On day 7 i have 2 out of 60

What is it this “out of” 69 and 60? Its not the total users from the first day? 
and if it is why the number not the same? 

Hi @Nate !

I believe this is linked to the granularity of the chart. To get the weekly retention for the same population, you need to use a custom date range (using “between”). Otherwise it seems to be taking new volumes of new users each week.

I hope I’m right 

@Matrak 

Thanks!
I did it here!
It still shows me different “of” numbers
Any additional ideas why? 

Week 1 - 5 of 57

@week 2 -  5 of 40

@Nate  
This seems to be related to the length of your date range which is over a week. For instance, let’s take a 3 weeks long date range that you analyze weeky (thus computed using a 7 day-window):

• From week 0 to week X-3 (X being the current week, I believe it is the second to last week on the retention chart), you have your whole population over the period of time.
• At week X-2 , some of the earliest population exit the observed population.
• At week X-1, some more exit the population.
• At week X, same thing. Your population is left with the latest new users of your date range.

Here’s an example on Amplitude demo: https://analytics.amplitude.com/demo/chart/new/av8cflw?source=copy+url

I concluded by thinking that the amount of “rolling time window” periods within your date range will equal the amount of consecutive decreases at the end (second to last) of your retention graph (example: with a 5 weeks date range on a 7 day-window, you will have 5 decreases on the 5 weeks prior to the last one). Hence, I would not take a look on the latest weeks given depending on your chosen date range. 

Once again, I hope I’m right.