Skip to main content

I want to run an adoption analysis fof a feature. The framework I’m using for it is called TARS

T = target population

A = adoptied

R = retained

S = satisfied

Target population means all active users x the percentage of whom we believe should adopt the feature. e.g. all active users and 60% of them should adopt it.

Then we define an adoption event and a frequency, e.g. weekly.

Followed by the same event as retention event.

Satisfaction is measured by, for example, success or failure of the retention event (no failure = satisfaction).

I can do the adoption part, and I also could pull in active users as a defined segment. But I couldn’t figure out a way to display all active users, independent of whether they performed any event or not.

Also, I cannot figure out a way to filter the segment down, e.g. by multiplying it by 60% or so because we believe at least 60% of all active users should adopt this feature.


Hey @lars 

I came across the TARS framework recently and was quite intrigued by it.

I’m not quite sure if I understood the “figure out a way to display all active users” part.

In Amplitude you can only include users who have performed an event to calculate active users and not the ones that are inactive during your selected time frame. You might want to expand the chart interval to as far as possible to ensure you get most coverage for the population number.

If you are trying to replicate the 60% of your active users number in Amplitude, then this might be a tough ask since this is only possible mathematically.  For your target population, you could use the active % metric of the event segmentation chart to determine % of users who have interacted with an existing or similar feature during the time period. This % would roughly give you the target audience if you didn’t have the 60% number.

You will have to tweak your segment accordingly to get the correct numbers at each step.

Hope this helps.

 


Hey @Saish Redkar , sounds like Ampltiude is actually not the right tool for a TARS analysis. We will get it done in Looker.


Reply