How To Create A Google Ads Experiment

To find the Experiments menu, scroll down in the middle column of your Google Ads account, until you see “Experiments”:

 There, you’ll see an overview of all the past and current experiments.

To create a new experiment, just hit big blue plus button:

Next, we’ll create a Custom experiment for a Search Ads campaign, since that allows for the most flexibility.

Experiment name

Next, you need to come up with an Experiment name & description.

Try to come up with a name that will make sense if you look at the experiment overview page 6 months from now.

If we’re testing the bidding strategy for your brand campaign, it might be named something like:

Next, select the Base campaign:

By default, Google will add your experiment name to the end of the Trial campaign name. So with a small tweak, you can make it look better:

If you click “Save and continue”, you’ll start to create your actual experiment:

From here, you can click on any of the areas of your campaign that you want to test.

In this case, we want to change the bidding strategy from enhanced CPC to Maximize Conversions.

So to do that, we need to open the Bidding options in Settings:

When you’re done making changes, click Schedule in the top menu:

Experiment goals

Before the experiment can go live, we’ll select 2 goals that work as the goals.

Which ones you select will depend on the experiment you’re doing, but in this case, we’re mainly interested in conversions and cost per conversion:

Once you start the test, you’ll have access to all metrics, so don’t worry too much about which ones you select.

Experiment split

Next, we need to define the experiment split. This boils down to how much of your traffic is allocated to each variation.

Usually this will be 50-50, to ensure you get enough data to reach a conclusion.

If it’s a risky change and you have tons of conversion data (> 100 conversion in the time frame you’re going to run the test in), you can go with a smaller traffic split like 70/30 or 90/10.

Under the “Advanced options”, you can get into the nitty gritty details of the actual test:

• Search based split: every time someone searches, they’ll see one of the 2 variations

• Cookie based split: once a person has seen one campaign, they will continue to see that campaign

By default it’s set to search based and that’s what I would keep it at in most cases.

Experiment dates

Next you need to define the time you want to experiment to run. You can either choose to leave it open indefinitely, and then end the experiment when you want to.

Or you can choose to end it after a certain number of days or on a certain date.

The right answer here really depends on what you’re testing. But if it’s your first experiment, you can just leave it at none and end it whenever you are ready to.

Note that an experiment will always start the next day, since it wants to get a full 24 hours of data!

Enable Sync

While your experiment is running, you might need to intervene in the Base campaign on something that’s unrelated to the thing you’re testing.

Let’s say a promotion has ended and you want to remove a certain headline from all your ads.

With the “Enable Sync” option, your Trial campaign will sync to the base campaign to make sure that all changes automatically sync.

This really is a great new addition to the experiments feature, as before you were afraid to touch the existing campaign in order to not screw up your test results.

When you’re all done, hit “Create experiment” and your good to go.