Solved

Remove internal users from analysis

  • 1 March 2024
  • 4 replies
  • 74 views

Hi, 

I’m looking to exclude internal users from my A/B testing analyses. When we all worked in the office it was easy to block the IP, but since our company is primarily remote workers I’m looking for a solution where I can create a segment that excludes everyone on our team from my test data. I don’t believe that asking individual users to somehow block themselves will be successful, so I was thinking about creating a segment and blocking user ids. It would be anywhere from 25-100 people so any suggestions are welcome. 

icon

Best answer by JennRu 11 March 2024, 15:48

View original

4 replies

Userlevel 7
Badge +10

Hi @mandychatigny 

Apart from using IP, you can define a user segment/cohort of your internal users and exclude them out from your charts.

You can follow some of the steps listed in this doc - https://help.amplitude.com/hc/en-us/articles/360016338212-FAQ-How-to-block-and-filter-out-internal-users

 

Hope this helps.

Hi @Saish Redkar thanks so much for the recco. We are usually auditing our site in incognito mode and that seems to be a barrier for my ability to employ the steps listed in that doc. Wondering if anyone else has had this same issue. 

Userlevel 4
Badge +8

Hi @mandychatigny is this pre-login site audits from employees in incognito? i.e. User ID is still anonymous?

 

If majority of these audits are occurring on the same devices, it might be helpful to filter out these users based on known device ID’s, though this would be a relatively static filter that would require updates when a new device begins auditing.

 

You can build an event segmentation chart on all employees post-login, group by User ID, group by Device ID, and then you can upload the device IDs as a comma separated value list into an Amplitude cohort to exclude based on devices. 

Hi Jenny, 

It is a pre-login audit from employees in incognito, which means that yes the User ID is still anonymous, and is the reason this has been a tricky situation for us. The dynamic cohort might be the best option since that seems to update more regularly, as opposed to a static cohort or segment. 

Reply