Hi!
We have 2 A/B tests running on our homepage. They target different audiences but some users fall into both audiences and are therefore currently being counted in both tests which is causing contamination in the data. How can we make these 2 tests mutually exclusive and randomly assign users who fall into both audiences into a test?
I've found here https://experienceleague.adobe.com/docs/target/using/audiences/visitor-profiles/use-profile-scripts-... that we would need to create a profile script as below. What would we then need to do?
Thanks!
Solved! Go to Solution.
Topics help categorize Community content and increase your ability to discover relevant content.
Views
Replies
Total Likes
Update:
I've now set up the profile script and audiences but:
If a customer falls into 'Group A' which I set for Test A, but they actually would've only qualified for Test B, they'll then not qualify for either test? So we would see a reduction in traffic to both tests. Is there any way of avoiding this?
Views
Replies
Total Likes
Update:
I've now set up the profile script and audiences but:
If a customer falls into 'Group A' which I set for Test A, but they actually would've only qualified for Test B, they'll then not qualify for either test? So we would see a reduction in traffic to both tests. Is there any way of avoiding this?
Views
Replies
Total Likes
Wouldn't you first set up audience who qualified for B? then split the traffic and then exclude that from A and split again. By default, you are already artificially separating the segments. So it's not 50%/50% for A and B, it is B, then 50/50, and then exclude B, then 50/50 again.
Views
Replies
Total Likes
Given both tests are running on the same page, have you tried excluding test B audience from test A? When editing audience for activities, there's an option called "visitor profile" and one of the drop-down options is "not from other activities" Haven't personally tested this before, but maybe you can create some kind of a rule combination of "not from other activities" + "page name where you are running the tests"
Another possible solution is to add additional query parameters in the URL where you are running the test so Target can differentiate and avoid audience cross-pollination.
For example:
Test A: example.com/?control
Test B: example.com/?experience_B
Hope this helps.
Views
Replies
Total Likes
Views
Like
Replies
Views
Like
Replies