Hi,
we are facing an issue with a standard redirect test using JS SDK. We have a page and 50% of our vistors are redirected to another page. We have implemented an async tracking with a delay of 1000ms, so that we are not missing any GA4 Events.
Everything seems to work as expected.
Now the issue we have: We have a very high Sample Ratio Mismatch (around 90% Base, 10% Variation). Our current assumption is that it has something todo with our Cookie Consent as most visitors on that page are new users, which have not given any consent yet. Here is what we can actually see:
Experiment is created here:
https://www.mvv.de/lp/maimarkt/photovoltaik
B1: Base Version (no redirect) - consent already given:
1. user is visiting the page
2. page view is sent
3. experiment is triggering
4. GA4 Event experiment viewed is sent
B2: Base Version (no redirect) - consent not yet given:
1. user is visiting the page
2. page view is stored in datalayer
3. cookie consent is opening
4. experiment is triggering
5a. User is giving consent
6a. GA4 Event: experiment viewed is sent
5b. User ist not giving consent
6b. no GA4 Tracking is sent
V1: Variation (redirect) - consent already given:
1. user is visiting the page
2. page view is sent
3. experiment is triggering
4. GA4 Event: experiment viewed is sent
5. Redirect to other page
V2: Variation (redirect) - consent not yet given:
1. user is visiting the page
2. page view is stored in datalayer
3. cookie consent is opening
4. experiment is triggering
5. Redirect to other page
6. cookie consent is opening on variation page
Our assumption is, that without the consent of the user (and no experiment viewed is sent to GA4 => B2b, V2), that user is not counted in any way. However it seems that V2 users are counted to our Variation-Sample (as they have been redirected), but not to the actual statistic (as there is no experiment viewed for them).
Our expectation would be that we have statistics with a 50/50 sampla ratio.
Is there anything we might do wrong? Are we allowed to trigger the experiment eventhough we are not tracking the results? Or are we only allowed to run the experiment after the user has given consent?
Thanks a lot for any hints.