The campaign Google loves but advertisers love to question. What is it?
You guessed it—it’s Performance Max. Since we’re increasingly required to use this ad format, I want to discuss ways to work around the inability to control the position of search headlines and the challenge of tracking individual asset performance.
One thing is certain, when Google launches products, it tends to release an MVP and then incrementally add features it deems necessary. In a world where Google seeks to maximize its revenue from every possible aspect of its ad product, it often launches ad products with limited control and maximum targeting, only gradually reintroducing control, transparency, or autonomy if it sees resistance to adoption of the product.
Performance Max exemplifies this strategy. Launched in beta in October 2020 and rolled out widely in November 2021, it has since gained features designed to appease its core customers (Google Ads professionals & agencies), though they don’t quite match traditional search capabilities. In 2023, Google introduced search themes, a lite version of the search terms report, and long-requested brand exclusions. By 2024, it added asset-level reporting, allowing advertisers to gain visibility into individual asset performance and more features continue to be introduced over time.
This brings us to the present and what we can do right now with Performance Max for copy testing. There are two elements of the ad copy that we’d love to understand with our tests:
- What is the performance and strength of each individual ad asset in the copy?
- What is the best performing combination of the individual ad asset?
As of February 2025, here are the best ways to test these ad asset elements in Performance Max:
Use Asset Groups as Ongoing Testing Buckets
How to test this: Create multiple asset groups within the same Performance Max campaign. Each asset group should target the same audience signals but feature different copy variations. Over time, observe which asset group outperforms the others in terms of conversions, CPA, or ROAS.
Where to analyze: Navigate to Campaigns > Asset Groups > Table View for a detailed overview of each asset group. Google allows you to use the same reporting columns as in search, enabling granular analysis of impressions, clicks, conversions, and more.
The result: While this method doesn’t allow for analyzing individual headlines, it provides insights into the performance of broader combinations of headlines and descriptions which is half the battle of understanding ad copy performance within Performance Max. I believe this test is also a good fit for companies with a low volume of traffic or conversions, perhaps due to their low ad budget or small market opportunity with Google Ads.
Rotate Copy in Asset Groups for a Pre-Post Test
How to test this: Swap out headlines and descriptions every two weeks and conduct a pre- and post-performance analysis of a specific asset group.
Where to analyze: Similar to the first method, use the Asset Group Table View to track key performance metrics.
The result: Unlike the first method, which relies on Google’s algorithm to distribute the traffic, this approach offers a more controlled way to analyze copy effectiveness by directly replacing old copy with new variations. If the testing periods are kept pretty short, we can presume that the auction environment stays mostly the same and thus we can get as close as possible to an objective A/B test. This testing method is still better for testing combinations than it is for testing individual copy.
Analyze Asset Performance Labels & Conversion Data
How to test this: Google provides performance labels like “Best,” “Good,” or “Low” for individual assets. Recently, it also introduced conversion and conversion value data at the asset level, making this method the closest thing to direct performance tracking for copy!
Where to analyze: Go to Assets > Assets and add the conversion and conversion value columns to evaluate the performance of each asset. Currently, these are the only two performance metrics available at the asset level. While it doesn’t provide the impressions, clicks, CTR data we’d all probably love to see, conversion data is a great first step towards ad copy transparency.
The result: Use this data to double down on “Best” performers by replacing the underperforming headlines and incorporating additional copy variations of the best performers. This is as close as we get to individual copy tests and analysis!
In order to do more objective analysis of your copy, you should pick one of these methods and stick to it. However, for imperfect but more holistic testing, I think a combination of 1 or 2 with number 3 will allow you to test both the individual copy and the combinations of the assets at the same time.
Bonus: Test Copy in Other Channels First
I know, this is basically the opposite of testing copy in Performance Max but hear me out! If you want more control, consider testing the same headlines and descriptions in Google Search Ads first, where reporting is more granular and testing is easier to control. Standard search ads allow for true A/B headline experiments, while responsive search ads let you download copy combination data from the past 90 days. Though this data only includes impression metrics, it can serve as a valuable proxy for final conversions.
As Performance Max continues to evolve, we should monitor Google’s updates for more granular reporting features. If Google ever phases out standard search ads in favor of Performance Max, it will likely introduce more detailed performance insights to encourage wider adoption. Given the trends over the past 4.5 years, we expect further improvements—and we’ll update this article as they roll out.
