Hi Amplitude community,
So we pushed both user_id and device_id on our Amplitude. In Event Segmentation, I know that the current unique users is counted based on user_id (if exist).
I wonder is there is a way to show the number unique devices (always based on device_id), instead of unique users in Event Segmentation?
Here is the visual setting:
Best answer by CCrowleyView original
As you already know, Amplitude runs its Uniques analysis by identifying unique user rather than unique device_ids. Afaik, showing unique devices based on device_ids isn’t supported yet natively.
You can read more on how Amplitude identifies uniques if it encounters multiple device ids for a user in this doc
If you really want to track use cases like devices per user,etc then I would suggest instrumenting device_id as a custom user property/ event property according to your use case.
Hope this helps.
Thanks for the answer
@Saish Redkar !
One additional question about merging user. As I read in the documentation, I know it will effect subsequent events, as it said :
But how about past events? Let say we have these event from the same user :
So in this case, will the report after merging change the past data? Here is more explanation of Andy data and the question (highlighted yellow)
Report before merging (i.e., get on Tue)
(will this remain 2 or changed into 1)
Lau, to answer your first question, you can create a chart that shows the count of device IDs instead of the native unique users chart. Just group your event by device ID and useing the formula chart do a PropCount. That will give you the count of unique deviceIDs for the given timeframe.
Still pondering your second question.
@Lau, It’s great to see folks helping one another out here. I wanted to check in. Were you able to make the progress you needed? Please let me know if you need any additional support and I can connect with the team. 😀
One trivial questions, is it possible to change the “Best Answer” ? I think the
@CCrowley ‘s answer is more informative at this point.
@Lau 😃 Thanks for closing the loop here. I’ve just updated the best answer.