AdWords is now Google Ads. Our new name reflects the full range of advertising options we offer across Search, Display, YouTube, and more. Learn more

4.3K members online now
4.3K members online now
Welcome to the Official Google Ads Community
Find optimization tips and how-to guides and videos by the Google team, Google Ads Experts and other industry experts
Collaborator ✭ ☆ ☆

Get a more accurate 50/50 ad split test

I was recently having some issues when trying to do an A/B ad test because the impressions weren't rotating as evenly as I wanted. I was running 2 ads and the settings were set to rotate evenly for 90 days then optimize for clicks (i did reset the 90 days by changing it and putting it back again) so it was set up correctly for this test.


My impressions were more like a 70/30 split. The system just wasn't doing what I had planned for it.


This is when I decided to try using AdWords campaign experiments to see if I could make it more accurate.




I started an experiment with a 50/50 split and changed one ad to control and the other to experiment (you can set it to run for about 90 days as well so it's similar to the rotation settings)


In the next few days my stats were looking much more even. The ads were now showing impressions with a 52/48 split.


If you have some time sensitive ad copy testing to do, it could be worth trying out this method as it really did help get more accurate data quicker.

about Adam Briggs

Adam is the founder of assistAd, an online tool developed by a group of search specialists to make it easy for you to plan, manage and optimize your search engine marketing campaigns. We're still under development but will be launching the BETA version very soon. You can contact adam by emailing or on skype: briggidere Or visit

AdWiser Top Contributor
September 2012

Hey, Adam.


The new A/B experiments in Analytics are designed to work like that ( instead of splitting the traffic 50/50, as the Website Optimizer used to do. And I'm not sure I like it, to be honest.


Indeed, it should minimize losses, but it becomes less significant this way.

September 2012

Hi Calin,


Yeah, I've been using the analytics experiments a bit now and still learning. I think it's a better system than the website optimizer.


This is more of a way to do ad copy testing for the same landing page. I found that the standard rotate evenly settings just didin't give me a close enough split when testing 2 ads so tried this out and it seemed to work much better.

stickleback Top Contributor
September 2012

I've been using the A/B Content Experiemtns in Analytics and the biggest difference I've realised is that it is experiementing with ALL your traffic - not just the traffic from AdWords - in fact one of my current experiments is for a client that has NO  AdWords campaigns....


This should be taken into consideration if, for example, you have a brand that is well known and a lot of your traffic is organic and from brand based searches. In this case you will need to run the experiment on a landing page (could simply be a dupe of the regular home page) and send your AdWords traffic there..... this will ensure that your test is only on your AdWords traffic - or any other segment you can control the destination url for.


But the original OP here was talknig about the ads in Adwords and how to spolit them . this does require the experiment in AdWords.


But a word of warning - don't get too many experiments all running at once - make sure that if there is a difference in performance between one ad and another, you can actually allocate that to the ads . and that hter eisnot another variable involved - more than one variable and you've got an issue - you can't readily make any claims about the perofrmance - much as you would like to.


It's the old "experimental method" they taught us in my Pysch degree course!