AI Max for Search campaigns (also known as “AI Max” for short) is a comprehensive suite of targeting and creative enhancements for Search campaigns. Two main features, search term matching and asset optimization, leverage Google AI to optimize your ads in real time and tailor your creative message to reach your customers and drive more value for your business.
This article describes how AI Max for Search experiments work. This experiment approach is aimed at delivering faster results and reducing some of the common experimentation errors, by diverting traffic and budget within the existing campaign (instead of creating a campaign copy for the experiment treatment arm).
On this page
- Benefits
- How it works
- Steps to implement
- Setup limitations
- Reporting
- Applying the feature
- Differences between new AI Max experiments and AI Max experiments through custom experiments
Benefits
This new approach offers several advantages (versus AI Max experiments through custom experiments):
- Faster results: By using a single campaign, you can likely get results and insights more quickly.
- Minimal setup errors: With changes applied to both the control and treatment arms (since they are both within the same campaign), there are fewer potential chances of setup errors and sync issues.
- Shorter learning period: This approach is likely to have a reduced learning period, since the traffic remains in the same campaign.
How it works
Search feature experiments are designed to enable advertisers to easily test key Search Ads features like Broad Match keywords and AI Max settings.
- Control: A % of your existing Search campaign with AI Max toggle turned off.
- Trial: The remaining % of your existing Search campaign with AI Max toggle on.
Once an experiment is complete, advertisers can evaluate performance and choose if they want to “apply the experiment”, which turns on the AI Max setting for the campaign.
Steps to implement
Set up AI Max experiments by selecting the on the Experiments page.
- Click the plus button
, and select Campaign features and settings.
- Select AI Max for Search campaigns.
The experiment, by default, activates both AI Max features, Search Term Matching and Asset Optimization, at the campaign level. However, you do have the flexibility to disable Search Term Matching at the ad group level and disable Asset Optimization at the campaign level, if you prefer. Your chosen settings will only apply to the treatment arm of the experiment.
Similarly, you'll have the flexibility of using other features that are available with AI Max, like adding URL exclusions at the campaign level and URL inclusions at the ad group level.
Setup limitations
You cannot create an AI Max experiment through this flow if:
- The campaign has enabled Legacy features like text customization (formerly known as automatically created assets), brand inclusions and exclusions, ad group location inclusion
- The campaign targets Display network
- The campaign is a part of a Portfolio Bidding Strategy
- The campaign uses Shared Budgets
Reporting
Experiment results can be found under the “Experiments” tab, with an expanded view of the “Experiment summary” page when you select the experiment campaign.
Applying the feature
You have a few different options of how to apply results:
- When creating the experiment, there is an option to “Apply your experiment changes if results are favorable”. You can select this during setup if you want the experiment to automatically apply if the results are favorable at the end of the experiment.
- After the experiment ends, you can select “Apply” from the Experiments table.
- After the experiment ends, you can enable AI Max from the Campaign settings page.
Differences between new AI Max experiments and AI Max experiments through Custom experiments
New AI Max Experiments |
Custom AI Max Experiments |
|
---|---|---|
Setup |
Automatic | Manual |
Availability for Search Campaigns targeting Display network |
No |
Yes |
Reporting surfaces |
|
|
Ramp up time |
Reduced ramp up time, therefore results can be obtained faster |
Ramp up time could be needed for the trial campaign, therefore experiments might need to run longer |