This message was deleted.
# ask-questionsw
worried-planet-2191
01/14/2025, 2:03 PMThis message was deleted.
š 1
f
fresh-football-47124
01/14/2025, 5:25 PM
hi Jimena,
You are correct in the usage of Namespaces. If you go to 100% on one test - it depends on how you do this
fresh-football-47124
01/14/2025, 5:28 PM
if you remove the experiment and set a force rule, or just change the code, the namespace will be ignored and this winning variant will apply to all experiments running (including that other parallel test). If you keep the experiment running but change the split to 100% 0% and still have the namespace, then that will effectively set deploy the experiment and preserve the namespace.
fresh-football-47124
01/14/2025, 5:29 PM
you'll still have a problem when you finish the second experiment - if you ran both experiments in isolation, and you have two winners, are you sure that the combined variants will also win?
r
ripe-tiger-2213
01/14/2025, 5:47 PMThanks Graham!
100% clear on the first point, the idea is to maintain the namespace as long as the other test is running at least.
On the second part, Is a valid point the one you're raising. The issue is that we're launching some ui changes in parallel of a trial flow that will run specifically on some targeted users. Technically is happening on the same area but the changes implemented as well as the objective and expected impact are quite different from each other. In this sense, we might test both variants in a new experiment after they finished just to be on the safe side but we don't have concerns about both tests having negative interaction effects.
f
fresh-football-47124
01/14/2025, 5:48 PM
when both tests finish, you could deploy test 1 to 100% for all, and then start a new phase for test 2, incorporating the test 1 changes, to see if the results of test 2 continue when test 1 is applied
š” 1
r
ripe-tiger-2213
01/14/2025, 5:53 PMGood point, thank you for the suggestion!
g
green-church-13845
01/22/2025, 10:04 AMHey, I have a follow-up question on this. @ripe-tiger-2213 is in my team š
Our setup: we have 2 tests
ā¢ Test 1: Namespace 0-0.5 and the test only runs on platform A only
ā¢ Test 2: Namespace 0.5-1 and the test runs on two platforms (A and B)
My assumption here: 50% of the users on platform B are not taken into account. Is this correct?
Can we include the users without splitting test 2 into 2 separate tests?
Thank you for your help š
121 Views