Experiment using redirect to split traffic

[DougBr]

Hi - We are running an experiment to determine the best option to land campaign traffic. The Target Activity is set for all traffic to go to Page A, and 50% of the traffic will be redirected to Page B. 

In Analytics, the traffic split is not 50/50. There is an appreciably higher visit count to Experience A. My understanding is that this can happen if the analytics call fires before the redirect occurs at Page A. The customer experience is fine because the visitor still does not see Page A. 

My question is whether the traffic for Experience A needs to be adjusted to match Experience B. In the example below, the premise would be that Experience A, showing 4,000 visitors, is inflated by 1,000 visits when the analytics fired before the redirect. If this is how it works, should I reduce the visitor count for Experience A by 1,000 so that both experiences are at 3,000. Doing so would have significant effect on conversion rates, so would like to rationalize this.

Experience	Visitors	Conversions
Default	4000	2000
Redirect	3000	2000

Thanks!

[alexbishop · Accepted solution]

I wouldn't recommend trying to re-balance out the traffic distribution manually. There's potentially a few different things at play here, have you looked through this FAQ regarding re-direct offers? There are a few suggestions there around how to resolve/mitigate this issue.

[DougBr]

Thank you for the notes. This is helpful.

What I have learned since posting this is that our site architecture may be involved. Most likely the way these calls are sequenced we are getting more traffic into ExpA because the analytics call triggers before the target call. Through a segment view we can see that we are not getting any overlap in the experiences, so we feel good about that. The only downside is that we have more traffic into one of the experiences. In this case, though, it's not a problem as we are focused on conversion rates. We can look past the traffic discrepancy and just focus on the rates.

Thanks again!