Rich-text Reply

Traffic Allocation and Visits Descrepancy

Anuragj 02-09-16

Traffic Allocation and Visits Descrepancy

We have launched A/B test with two pages variants where we allocated 50-50 Traffic to both the pages. But when we are checking Visits on these two variants using Adobe analytics, we are seeing one variant has double visits than other. ie. A has 4000 visits and B has 2000 visits each day.

 

Why is it happening?

 

Level 1

JDahlinANF 02-09-16
 

Re: Traffic Allocation and Visits Descrepancy

The split will show up in "Visitors" no "Visits".

 

On the surface, this means that users in one group return to the site (and hit that page) twice as much as the other one.

Anuragj 02-19-16
 

Re: Traffic Allocation and Visits Descrepancy

Thank you Napolean. I have checked Unique Visitors as well and they are also double in case of Control.
What we found that, we have applied a redirection rule in Optimizely, so both the control and test pages loads for particular user and hence same particular user is counted for control variant which is supposed to be only part of test variant. That is why Omniture giving us wrong numbers. I believe we do need to save experiment ids in evar/prop for measuring performance.
Level 1
JDahlinANF 02-19-16
 

Re: Traffic Allocation and Visits Descrepancy

[ Edited ]

If you are letting Site Catalyst run before Optimizely has a chance to redirect the users, then, yeah, you're going to have issues with your metrics.  If you don't want to use a prop or eVar but you record the URL already, you could build your segments like:

"Visits with Hit containing URL1 but not URL2""

"Visits with a Hit containing URL2"

 

If your implementation is set up as they recommend it, (Optimizely loading at the top of the page - so it can change things before the user sees them), this ought not happen.  We have Optimizely loading at the top and SiteCat loading at the bottom and never ever see this happen.  If this is how you have yours set up, share the redirect variation's code here and we'll figure out why it's not redirecting fast enough.