Mastering Performance Max Experiments for Campaign Success

When Performance Max campaigns were launched, the entire equation of how ads could work on Google changed. With Google AI’s automation and the extensive reach of PMax campaigns, advertisers could boost their ad performance and get better business results without much manual intervention. 

However, that raised another important question: if most settings in PMax campaigns are automated, is there any way to test what’s really working? If you have the same question, then don’t worry, that’s where Performance Max Experiments can help. 

In this blog, we will cover PMax experiments in detail, the different types of experiments advertisers can create, how to set them up, their pros and cons, and many other important details. 

Are you ready? Let’s get started. 

What are Performance Max Experiments? 

Performance Max Experiments are built-in tools within Google Ads that you can use for A/B testing different settings and features of your PMax campaigns. These Experiments also allow you to measure the overall impact of using PMax campaigns:

1. Either as an addition to other existing campaign types such as Search, Shopping, etc., or;

2. As a possible replacement for those campaigns.

Through these experiments, you can compare a PMax campaign (or a PMax feature) against a control scenario to measure if there is any incremental lift in conversions or conversion values.

For example, you can use a PMax experiment to measure the incremental lift of a newly created PMax campaign running alongside an existing Standard Shopping campaign.

Types of Performance Max Experiments

There are different types of Performance Max Experiments: 

1. Uplift Experiment: These types of experiments allow you to easily measure the uplift in conversions or conversion value that occurs when you add PMax campaigns to complement other campaign types, such as Search, Display, Discovery, etc. 

Uplift Experiments are useful when you are not yet fully ready to switch to Performance Max campaigns, but are interested in evaluating their potential benefits. 

2. Upgrade Experiments: These experiments allow you to simulate a situation where you shift your budget from existing campaigns like Standard Shopping, Dynamic Search Ads (DSAs), or even Google Display Ads (GDA) to a PMax campaign. This will allow you to understand how your PMax campaign might perform (if you go on to create one) before you commit to it.

Currently, Google offers three templates for these experiments: 

• Standard Shopping campaign vs. Performance Max Campaign (with GMC)
• Dynamic Search Ads (DSA) vs. Performance Max 
• Google Display Ads Campaign (GDA) vs. Performance Max 

3. Optimization Experiment: With these, you can test different variations of your PMax campaigns. Primarily, these experiments are helpful for advertisers to verify if adding or modifying any PMax assets affects the campaign performance. 

For example, you can test how your Performance Max campaign will perform with Final URL expansion enabled or disabled. 

How to Set Up Performance Max Experiments?  

In general, there are two ways to create PMax experiments:

1. From the Experiments page.
2. And from the recommendations cards that you may see on the Recommendations or Experiments page. But please note that this option is only visible when you have an existing campaign that is eligible for A/B testing, when, let’s say, it has enough conversions.

Below, we will be discussing how to create PMax Experiments from the Experiments page. It is because this method is useful for all advertisers, including those who don’t yet have an existing campaign but wish to create one from scratch.

Creating Uplift Experiments for Performance Max  

1. In your Google Ads account, click on the ‘Campaigns’ menu on the left side. 
2. Click on the ‘Campaigns’ drop-down in the section menu. 

3. Then click on ‘Experiments’. 

4. Choose ‘Performance Max Campaign Experiments’

5. Click on the ‘+’ button and then choose ‘Test uplift of Performance Max.’  

6. Then you will have to select an existing (active) PMax campaign or create a new one before you can create the uplift experiment. 

Note: PMax campaigns with linked Merchant Center accounts are not eligible for uplift experiments.

7. Click Next. 

After that, in the “Performance Max (treatment)” section, the campaign you selected or just created will be tested against Google’s auto-generated list of comparable campaigns. However, you can update this list in the future.

Note: Comparable campaigns are automatically selected by Google (if you don’t select any) for the experiment you create. And Google selects these campaigns based on factors like the same domain, at least one conversion goal that overlaps (with your PMax campaign), and locations that overlap.   Google includes Comparable campaigns in the control and trial groups of your experiment.

8. Then, from the Traffic split section, you will have to split traffic between the control and treatment groups. By default, it is set to 50%. 
9. Click on Next. 
10. You can check which campaigns are added to each experiment arm in the “Test Groups” section. 
11. Add the Experiment Name, and select a start date. 
12. Schedule it. 

💡 What are the Control and Treatment groups in Performance Max Experiments? In a Performance Max experiment, Google splits your campaign traffic into two groups: 1. Control Group: This group is basically your original campaign setup and acts as the baseline for the experiment. It doesn’t include the changes or edits you want to apply. For example, if you want to test the impact of Final URL expansion on your PMax campaign, then the control group will include your existing PMax campaign without Final URL expansion turned on.  2. Treatment Group: Conversely, the treatment group is the campaign version that reflects the final change and extra features you want to add. Referring to our previous example, if you want to create an experiment to test the impact of the final URL expansion asset, then the treatment group will reflect the campaign version where the Final URL is enabled.

Creating an Upgrade Experiment For Performance Max

Steps 1 to 4 remain the same as those we followed above for creating uplift experiments. In step 5, after you click on the ‘+’ button, select ‘Test vs. Shopping campaign’ from the list. 

Note: We will use the Performance Max vs. Standard Shopping template for creating our upgrade experiment. However, you can also choose to test Performance Max against Dynamic Search or Google Display Ads campaigns.

1. After that, you will have to select a Shopping campaign that is currently not a part of any other experiment. 

2. Then decide the traffic split.

3. After that, you will have to select or create a Performance Max (treatment) campaign to test. 

Remember, you can either ask Google to create a campaign for you, create it yourself manually, or select an existing campaign. 

4. Click on Next. 
5. Then just check the campaigns that are in the experiment arm in the “Test groups” section.

6. Select the start and end dates of the experiment. Also, add its name. 

7. Click on Schedule. 

Creating Performance Max Optimization Experiments 

Before we learn how to create optimization experiments, you must note some important points. 

• After you create your optimization experiment, Google recommends not changing the settings of automatically created assets, Final URL expansion, or even the video enhancement feature. This is because any edits to these assets can affect your experiment results.
• Similarly, you should not change any other assets in the campaign once you create the experiment. 
• Your assets meet Google’s asset requirements. 

In addition, your campaigns must meet certain requirements to be eligible for optimization experiments in PMax. 

• Your campaign must not be a part of other experiments. And it should be an active campaign. 
• If your PMax campaigns are restricted by budget, Google recommends raising your budget. This will ensure that the experiment runs smoothly. 
• In case you are creating this experiment for automatically created assets (such as Final URL expansion), ensure that both “Final URL” and “Automatically Created Assets” are turned off for the campaign. 

Once you are clear with these requirements, follow the same steps from 1 to 4 highlighted above in the section where we created uplift experiments. 

Then, after clicking on ‘+’, select “Test a Performance Max Feature.” 

1. Under ‘What do you want to test?’, choose ‘Assets provided by you.’ If you want to test out text assets generated or enhanced by Google AI then you can select the other option.

2. Under ‘Select Experiment type’, click on ‘Assets for retail campaigns’.

3. Click ‘Select campaign’ and choose the Performance Max campaign you want to test.

4. Then, pick the ‘Asset Group’ for testing.

5. Define the traffic split between control and treatment arms.

6. Under the treatment arm, click Add additional assets and upload the supported asset types.

7. You can also rename your experiment by editing the “Experiment name” field.

8. Set the start date and end date using the dropdown options.

9. Once done, click Schedule to launch your experiment.

Note: 1. Currently, Google only allows you to analyze the impact of adding assets to a feed-only PMax campaign. 2. In the section above, we discussed how to set up a PMax optimization experiment, which can help you measure the uplift of your creative assets or their settings. But Google also allows you to create another type of optimization experiment, using which you can measure the uplift of Final URL expansion. Refer to this Google page to learn how to set up those experiments.

Benefits of Performance Max Experiments 

Let’s discuss the benefits of using Performance Max Experiments: 

1. Track Measurable Performance Gains: With these Experiments, you can easily measure the positive impact that Performance Max campaigns (or any of their assets) can have on your overall marketing effort. Thus, instead of making guesses or assuming, you get to see detailed statistical data that informs you if you should commit to creating a PMax campaign (or adding or modifying any assets).
2. Optimize PMax Campaigns: The optimization experiments can help you easily determine the best combination of assets for your PMax campaigns. That can help you maximize results from your ads. 
3. Find the Right Combination of Google Ad Campaigns: PMax Experiments, such as the uplift, help advertisers understand which types of campaigns complement PMax campaigns well. This allows them to allocate their budget more wisely. 
4. Auto-apply Experiment Results: By default, for some experiment types, this feature is enabled. With this feature, Google can automatically apply changes to your original campaign from the successful experiment. 
5. Controlled Testing: You can customize the testing parameters. For example, Google allows you to split the traffic between the control and treatment arms as per your choice. 

Best Practices for Performance Max Experiments 

You should follow these best practices for PMax Experiments: 

1. Run your experiments for an optimal duration: Google recommends that you run your PMax experiments for at least 2-3 weeks. This is to ensure that your experiments gather enough data, and you can get a final winner between the two campaigns. If your campaigns are not able to gather enough data, then you should raise the campaign budget. 
2. Make minimal edits to campaigns: When your experiments are in progress, Google recommends that you shouldn’t make or make as few changes as possible. 
3. Target the same products: If you are creating an upgrade experiment, let’s say between Standard Shopping vs. Performance Max, then Google recommends that you target the same set of products in both campaigns. This is to ensure that the performance of both campaigns can be measured in similar conditions. 

For such experiments, also ensure that the same set of products isn’t targeted in any other campaign. That will help avoid overlaps. 

4. Minimize other experiments: You must also ensure that you don’t create any other experiments (for Ad variations, etc.) when your PMax experiments are in progress. That is to ensure that the experiments don’t interfere with each other. 
5. Regularly monitor your experiments and their data: Lastly, check the performance reports of your PMax experiments. Look at metrics like conversion rates, CTR, etc., to understand how your test compares with the original campaign. 

Google also recommends using confidence intervals (which highlight margin of error) to determine whether the results of your experiments are statistically reliable. If they are, then you can apply those results to your original campaign. 

Conclusion 

We hope this blog helped you learn what PMax experiments are and how they work. Ensure that you use these experiments to avoid guesswork about how your PMax campaigns or their assets might impact your results. 

Good luck! 

FAQs

1. Can I edit comparable campaigns?

Yes, but only for 7 days after the experiment starts. You can add or remove campaigns from the list once Google shows you which ones were selected.

2. Will my existing campaigns be affected?

No. Experiments won’t hurt your ongoing campaigns. They’re just used to compare traffic and results between setups.

3. Can I change the traffic split during the experiment?

No, once you set the split (like 50/50), it cannot be changed later.

4. Can I run other experiments at the same time?

Yes, but it’s better to avoid running multiple tests together. That helps get clearer results.

5. Can I adjust the campaign budgets during the test?

Yes, but try not to. Changing budgets during the experiment can affect accuracy.

6. Are uplift experiments available for all campaign types?

No. They only work for Online Sales (non-feed), Store goals (offline), and lead-gen objectives. They don’t support PMax with a product feed.

Check Out These Related Articles:

• Performance Max Ad Specs
• Performance Max Vs Search Campaign
• Performance Max Negative Keywords
• Ultimate Guide to Google Ads A/B Testing
• How to Set Up Performance Max Campaigns
• Performance Max vs Smart Shopping