How to differentiate between Mobile Web and Desktop Web (and Tablet as a bonus)

  • 4 October 2022
  • 3 replies

Hi all, 

Is there an equivalent of Google Analytics Device Category in Amplitude? Here’s a post about how GA does it.

I’m looking to segment our Web platform events and user properties further by Mobile vs Desktop (and Tablet as a bonus). 

Here's some Amplitude community topics, that I've found so far, and I think it requires segmenting by a combination of OS and Device type, but I can’t find a clean way of doing it.


Many thanks in advance. 



Similar topics.


Best answer by Maya 10 October 2022, 11:28

View original

3 replies

Userlevel 1

Hi Raul,


At this time, our SDK does not have a default user property that distinguishes events from web and mobile traffic. In case you can track these differences on your end, you can create a custom user property that will indicate whether an event is from a web or mobile source.

While there isn’t straight forward possibility to do so in Amplitude, we would indeed suggest that you use 'OS' and ‘Device Family’ user properties to come closer to these results without additional instrumentation (only captured if you are using the Amplitude SDKs to send data). Please note I would not position this as a solution but as an attempt to get close to these results.


Many of your mobile users have the word "mobile" in the name of their OS, and you can segment for events where with "OS contains Mobile" and "OS does not contain Mobile". Additionally, you can use “Device Family” to check if it contains “tablet”. 



Things to be aware of:

  • The OS = "Chrome 105" would apply for mobile and desktop
  • The '[Amplitude] Device family' and '[Amplitude] Device type' values based on a custom mapping that we do whenever we encounter new devices. Sometimes, these might not be added to the existing mappings. You can request a new device to be added as mentioned here

Hope this helps! If you have anymore questions, please let me know!

Hi guys,

Just chiming in here as I think the solution you have offered is a real hacky/bandaid workaround. I often find myself trying to achieve what Raul has mentioned in Amplitude. 

I think this would be a very worthwhile default library user trait to have!


I agree with @toby_c5k , this solution is too much of a bandaid that’s prone to error, and does not give us an easy solution on par with other standard tools in the field (i.e. Google Analytics).