Campaign Experiments in Google AdWords: The Easy Way to Test
Need a quick and easy way to conduct a test in one of your online campaigns – a landing page, bidding strategy or keywords? AdWords campaign experiments are simple to set up, fast, and best of all — free. No need to pay for external software or services.
Here’s how it works – select the campaign you want to run an experiment on. On the top right, select “Create new” from the “Drafts” drop down menu.
Create a draft, setting up the variables you want to test. For example, one of our clients is currently testing a landing page design, to try to boost conversions. In this case (as in most testing), the ads, keywords, bidding strategy, basically everything in the draft stays the same as the control. The only variable is the link to the new landing page.
Name your draft:
This creates a copy of your original campaign. You can then make changes in the draft to the variables that you want to test.
Once you’ve made changes to your draft and everything’s set up the way you want, you can then create your campaign experiment. Click on the blue “Apply” button and then select “Run an experiment.”
On the next screen, you decide how long you want it to run – start and stop dates. You can also decide what percentage of traffic you want to go to the experiment. In our case, we’re splitting the traffic 50/50 between the experiment and the control. But there may be cases where you want a higher percentage to go to the control or test. You can decide.
Creating an AdWords campaign experiment
Once you have your settings, click “Create” and let the experiment begin!
To view the results of your campaign experiment in adwords, select it from the campaign menu on the left side bar:
Here’s what your results will look like:
After you’ve decided on a winner of your campaign experiment, it’s simple to apply the result to the entire campaign. If your control is the winner, you simply turn off the experiment, pushing all traffic back to the control. If the test wins, you can click the “Apply” button and the experiment settings will be applied to the entire campaign. Voila!
Side note – you may not see the traffic split exactly the way you designate. For example, in our landing page test, even though we set it up to split traffic evenly, the control portion received a higher percentage of impressions. Here’s the explanation from our trusty Google reps:
“When you setup a 50/50 split, or ‘even ad rotation’ in a Campaign settings, this only assures that you are going to enter the same number of auctions. If you run a campaign experiment in adwords and are not seeing even serving between the control and experiment, that means that the one with the higher volume is simply doing better. Typically this is reflected by a higher CTR in the portion that is winning more auctions (as we see here). What we can assure you is that we are entering the experiment and control into the same number of auctions and what happens after that shows them how effective their test was.”
So keep that in mind as you’re evaluating your campaign experiment results. For help setting up campaign experiments in AdWords, or any other online marketing testing, feel free to get in touch with me, or any other AIM team member.
P.S. if you’ve found this article useful, check out Natalie’s article on AdWords scripts, and automating your acquisitions at scale.