Redirect Test cause huge spike of new visitors in Google Analytics
We created a redirect test on our homepage with 2 redirect variations, e.g. example.com/v2 and example.com/v3 and this morning we noticed a huge spike of new visitors coming in from direct traffic. A quick check on Google Analytics showed huge portion of new visitors landed on the 2 redirect variations which lead us to believe that it is possible some new visitors are recorded twice in GA, one in the original, one in the variation.
I would like to know if this is a possible theory. Also, I checked our page source code that the optimizely script is located at the top of the head, before GA's script. I've attached a screenshot from GA just in case.
Solved! Go to Solution.
Do you also see a spike in overall visitors or just new visitors? Does the / to /v2 jump happen across domains? Optimizely does a client side redirect which means your referrer properties etc won't carry through to the subsequent landing page (so you won't be able to tell if the /v2 or /v3 visitors are organic etc.. unless they are tagged w/ utm_ parameters)
That said, you should still see an overall steady "total traffic" number that doesn't change. What's strange to me from your screenshot is that the landing page url contains a ".com" whereas most suites view of that report shows just "/" or "/v2". To debug further I'd suggest:
- check your ga profile filters
- confirm your /v2 or /v3 is on the same domain and/or the ga cookies are shareable across them if they are subdomains
- utm_ tag what you can and/or use optimizely's segmentation in their reporting suite for your marketing source attribution goals!
TLDR: Yes this could be caused by optimizely b/c the referrer is lost in the redirect so you don't have that information on the subsequent landing page for new visitors so some of those "new organic" might now be counted as "new direct"
Hope that helps!
Edit: Wanted to setup a fact checking playground for this so we have a couple of things to try:
- Tested the resulting redirect and referrer for each of the pages and found that *no pattern for client side redirecting carries the referrer *in chrome**.
- The only pages that keeps the referrer intact (which is what GA needs to attribute "not direct" to incoming traffic that's not utm_ tagged) are landing pages with no client side redirect that don't get less secure by going from https to http.
- Code for the page is here and the secret sauce essentially just redirects based on a query param: http://git.io/vJH4L
- Here's the final Data Table: http://git.io/vJH0Q
- Tom Fuertes | CTO @ CROmetrics / LinkedIn
"Most Impactful Use of Personalization" and "Experience of the Year" Optie award winner.
Thanks for your reply. I've checked our overall traffic since Apr 1st and found no spikes in total sessions. There is a spike in direct traffic (total sessions and new visitors) since last week, and a drop in organic search traffic, so you're right that referrer is lost in the redirect, and huge portion of organic search traffic have been converted to direct traffic. I've also attach some screenshots just to make sure we're on the right page.
I guess I should have wait few more days later to confirm the GA reports then report it here.
Regarding the extra domain name in /, /v2, and /v3, I've prepend the hostname into the request URI since we are using this tracking code on multple domains, so I can confirm that /, /v2, and /v3 are from the same domain.
Thanks for your help.