One of the less known, and thus less used, components of Google AdSense (aff) is the Experiments option.
Start with AdSense Experiments
To get started with an AdSense experiment, go to My Ads and then choose BETA Experiments
In this screen, choose + New Experiment
The first thing you do is give a good name to your experiment so that you know which site you’re going to test or with which ad format.
In the field below you write what the component you want to test and what test you want to do.
Tip! Test one change per experiment, if you make more changes, you still do not know what exactly makes a better score.
Set up an AdSense experiment
Setting up your experiment is actually very simple.
First, select the ad unit you want to experiment and then what you want to change.
The upper part is your original and you also see what it looks like.
The variant is what you want to test. The test you want to run can consist of a variety of things, such as text ads only, instead of image and text, or just a color test.
Here is a color test setup to see if the Title in Red and the URL in default blue will give more clicks than the blue title.
Do those types of tests on a site where you get quite a lot of visitors, which not only generates more money but also ensures that the experiment runs fast.
After the setup, you confirm that Google can handle the rest. Google will divide the ad impressions and keep track of the results.
In addition, they will show more and more of the best option until a good comparison is possible and the “winner” is known.
You can see what’s happening during the experiment by clicking on the experiment you created.
You will then get an overview of the distribution and the consequences of your change.
In the example above, the experiment is only busy for a day and with only 63 impressions. For the end result, you will have to wait for sufficient information to be collected by Google.
Take note of the “Trust” column, only if you see a value above 90%, a good comparison is possible. At 95% the test will be completed. According to Google, you should definitely give 24 hours to see good reliable results
Seeing the reports of this experiment is not yet meaningful, but by the end of your test, it is a great measure of your changed income.
If your AdSense experiments go well, you may perform the best, but it may be that the variant does it better. In short, the experimental option is definitely worth trying out! Finally, here too, the measure is known …