Google Ads Experiments can help you discover whether your campaign would perform better if you were using a different bid strategy or maybe a new set of creatives. In the fast-evolving world of Google Ads this in-built testing feature is one of the easiest ways to test campaign changes and improve your campaigns over the long run.

In this article we will look at the experiments feature in Google Ads and how to utilize it to maximize your campaign performance.

Google Ads Experiments allow you to test different campaigns and ad settings so you can learn what helps you drive better and more efficient results whilst splitting your budget evenly. You can test different bid strategies, ad copy, different landing pages, location/targeting settings, match types, and more.

The most common use cases for running experiments now are:

  • A/B testing ad copy and creative assets, especially with Responsive Search Ads you can test assets such as text, images and videos.
  • Testing which bid strategy gets you the best results (Target ROAS vs. Maximize Conversions).
  • Measuring the impact of Performance Max over your existing Search or Shopping campaigns.
  • Testing which landing page performs better for your campaigns.
  • Testing Broad Match keywords in your search campaigns
  • Testing AI Max for Search Campaigns.

Why run experiments?

Experiments let you split the traffic and budget on your test and control campaigns equally, giving you a statistically sound A/B test. You can specify the duration of the experiment, and if your theory doesn’t work out, you can simply end your experiment at any time. You can compare the results of your control and test campaign, and if the test campaign is successful, you can apply those changes to your original campaign with a single click.

You can see below a recent example of testing target ROAS bidding on a Shopping Campaign which was originally set to manual CPC.

Positive Google Experiment Treatment Arm Results

How to run experiments?

Google Ads has streamlined its Experiments tool to make testing easier, especially as new features and campaign types are introduced that Google want you to adopt.

On the left-side menu of your Google Ads account, click on ‘Campaigns’ > ‘Experiments’ and then click the blue plus sign to create a new experiment.

  1. Choose the experiment type (Campaign features, Assets, Custom etc.)
  2. Select the campaign in which you want to apply this change to.
  3. Follow the instructions for the specific experiment type you selected.
  4. Name your experiment, set a start and end date, and choose your experiment split (the percentage of your campaign’s budget that will be allocated to this test).
  5. Click on ‘Create Experiment’ and you’re done.

Setting up a Custom Search Campaign Experiment

  1. In the left-side menu, click on ‘Experiments’ and then ‘Custom experiments’. Click the plus sign again.
  2. Choose ‘Search’ as your campaign type.
  3. Next, name your experiment and select the campaign you want to experiment on.
  4. You will be taken to a campaign-like interface where you can make your changes. Go to ‘Settings’ and change the bid strategy, add new broad match keywords, change the landing pages or whatever other change you wanted to test.
  5. After you are done making your changes, click on ‘Schedule’.
  6. Set your experiment split (a 50/50 split is recommended for a clear comparison) and in ‘Advanced options’ pick whether you want a search-based or cookie-based split. Cookie-based is the recommended option as it keeps the user experience consistent by showing the same user the same campaign version.
  7. Set your start and end dates and choose whether to enable sync. Keeping this option on is a best practice, as it ensures any changes you make to your original campaign (like adding new negative keywords) are also applied to your test.
  8. Once you are done, click on ‘Schedule’ and your experiment is set up.

Setting up a Performance Max Experiment

  1. From the ‘Experiments’ menu, select ‘Performance Max Uplift’ under Campaign types.
  2. Choose your existing Performance Max campaign to be the “Test” arm of your experiment.
  3. Google Ads will automatically select a list of comparable Search and/or Shopping campaigns to serve as the “Control” group. You can manually adjust this list if needed.
  4. Set your experiment split (a 50/50 split is the default) and schedule your start and end dates.
  5. Launch the experiment. This test will show a portion of your audience your new PMax campaign, while the other portion will continue to see your original Search and Shopping ads. This helps you measure the true incremental value that PMax adds to your account.

What to do after launching your experiment?

Keep a close eye on the results to make sure your experiment is running successfully. In the ‘Experiments’ tab, you can find the results on the main page. You will be able to see the numbers of both your control and test campaigns and the difference between them. Google Ads’ reporting will also indicate whether the difference is statistically significant, helping you avoid making decisions based on random fluctuations.

If you find the results of the experiment campaign better, you can ‘Apply’ the changes to your original campaign. If the results are poor, simply click on ‘End’ experiment to discard it.

We recommend running your experiment for at least two weeks but ideally a month before you decide whether to apply or end it, as this gives Google enough time to learn and gather statistically significant data.