URL test - traffic sources
I'm just getting started with Optimizely, and as a first test I'm running a page on our site against a page on a different URL (created in unbounce). I'd like to ask about how the traffic sources are being recorded by our Google Analytics.
The test is a 50-50 split between the two pages, and while numbers of traffic that are hitting both are consistent (the 50-50 split is working fine), there does seem to be a large difference in where the traffic that is hitting each varient is from.
For example, the original page is receiving normal traffic with a healthy split between direct, organic, social, referrals etc. but the unbounce page is receiving almost all of it's traffic from direct - and a small chunk from organic and hardly any from social and referrals.
There are no filters on source or anything else except for excluding mobile traffic. This is filtered out on Google Analytics too, so won't be causing the discrepancy.
Why might this be? the traffic being routed to the unbounce being classed as direct when it's redirected by optimizely, hence losing it's original source? Or is this something else happening here?
Any help would be greatly apprecieated!
As I understand, Optimizely uses the JS function window.location.replace() to redirect to the variation page. This function *should* preserve the referrer information, except on some older browsers. I don't suspect this is the issue.
Does the original site use HTTPS? And if so, does the Unbounce page use HTTP? If the answer is "yes" to both, then that's what causing the referrer info to be dropped. HTTPS does not pass referrer info to HTTP, so you will always see "Direct" as the source for HTTPS->HTTP traffic.
Hi there Greg,
Thanks for taking the time to reply, and sorry for my delay - only just got the notification that you had.
The both the original and test page use http, so unfortunately it's not that - good to know for the future though. What is strange is that it's not losing all the referral data, just a large portion of it. So there are some organic (and other types of traffic) coming through, but they are much to low to be normal.
Thanks again for your help, I wonder if you have any other suggestions?
Not too sure what else it might be. I would look into what browser (and version) your visitors are using, because older browsers (especially IE) might not pass full information with the location.replace() function. Also, you should be able to view test results segmented by channel from Optimizely's results page. At least that might give you some additional information about users in the experiment.