Performance Max ads have been touted as an efficient way to manage Google Ads across multiple Google-owned channels, making campaign management simpler for PPC advertisers.
It combines all of Google Ads’ channel options together in one campaign and uses Google’s AI (Gemini) to optimize budgets and ad serving across all channels based on performance.
On the surface, I’ve seen great results from our clients’ Performance Max campaigns, but it’s certainly left me wondering which channels performed best.
Imagine the learnings you could glean from the rapid testing and data collection of AI applied to your Google Ads!
It could easily help you tweak and improve channel-specific campaigns and even utilize learnings from Google Ads across other ad platforms.
Exploring Performance Max data
Delving down into Performance Max data is easier said than done. Google is careful in providing only certain insights into the campaigns.
For example, within Google Ads (and when pulling the data from Google Ads via the API), advertisers can’t parse out performance data, such as conversions or cost per conversion, by channel. And why not?
I believe Google is using Performance Max for two main goals.
First, Google will likely realize that certain ad channels, such as Google Discover, act more as a branding interaction than a bottom-of-funnel (BOFU) interaction.
With marketers mostly being measured by performance, such as clicks and conversions, I suspect advertisers weren’t biting on these ad formats.
Performance Max allows Google a conduit to force advertisers to use channels they may not have originally chosen under the guise of overall optimized performance.
However, marketers benefit from AI-optimized ad-serving, but at a cost. We can’t remove the channels from Performance Max, which is affecting overall conversion performance.
So if Google Discover is negatively affecting performance, will Google stop showing your Performance Max ads there, or will they continue to show the ad on Google Discover to boost that ad channel’s usage at the expense of optimal click and conversion performance for your campaign?
Unfortunately, advertisers aren’t able to adjust Performance Max channels and have relatively no visibility into ad channel performance.
And therein lies the conundrum with Performance Max. Love it for the budget optimization or hate it for the lack of performance control.
If you, too, have a love/hate relationship with Performance Max, let me share some ways you can bend Performance Max to your marketing will through data from both Google Ads and GA4, even despite its data limitations.
As Menachem Ani shares in his 2022 article, focus your effort on guiding the AI machine.
Create broader campaigns and learn from them
When we first started testing Performance Max for a fine wine retailer we work with, at the advice of other Performance Max experts, we created very specific wine variety campaigns, such as champagne, merlot, white zinfandel, etc. However, we didn’t find this necessarily successful.
While some wine varieties performed well, less popular varieties struggled. Even after garnering high CTR, they often didn’t result in conversion and revenue.
We shifted strategies and decided to go a bit broader. With this strategy, we determined which varieties performed strongly through revenue data. Those that performed well earned their own breakout Performance Max campaign.
When testing Performance Max, consider starting with a broader category and then learning from your data to narrow it down.
You can glean this information from GA4 using an Explorations report (or pull the data via the GA4 API) and review item sales by campaign.
This will tell you the specific categories performing well through Performance Max and could likely support their specific Performance Max campaign.
Dig deeper: How to combine GA4 and Google Ads for powerful paid search results
Create multiple asset groups per campaign with a specific focus
Think of asset groups like you would ad groups in search campaigns. In search ad groups, it’s a best practice to group common keywords together in an ad group. Do the same with asset groups and the asset focus.
We recently ran two asset groups in a test in a Performance Max campaign for the same retailer. The asset groups had the same ad copy and extensions but different images and videos by group.
Group A contained images and videos featuring beautiful scenery of vineyards, luscious glasses of wine pouring or a group of friends celebrating with wine.
Group B featured images of specific wine bottles from our inventory, using a variety of brands based on our most popular wines in that category.
While Google doesn’t make it intuitive, you can compare the results of two asset groups. Asset groups in Performance Max campaigns are somewhat akin to ad groups in search campaigns, but there is limited reporting in Google Ads around asset groups.
To see how one asset group performs against another in a Performance Max campaign, navigate to the campaign in Google Ads, and instead of the default asset group Summary view, switch to the Table view:
The table view provides greater detail of performance by asset group, and you can customize the columns as well, allowing you to see conversion data by asset group:
Use different destination URLs for each asset group
However, seeing asset group performance in GA4 is a bit more difficult. Unlike ad groups, GA4 does not pass asset group information through to GA4. So, if you want to pull this data together in GA4, how can you do it?
The simplest way is to add a tracking parameter to the asset group destination URL. For example, you could create a separate destination URL for each asset group using a parameter, such as utm_creative_format, then capture that parameter in a custom dimension in GA4.
This would allow you to break down actual conversion and sales data by asset group in GA4 in addition to campaign.
Ad strength isn’t always indicative of possible success
In Performance Max asset groups, more asset items (such as images, videos, etc.) typically increase your ad strength. However, Google’s definition of ad strength isn’t always indicative of likely success.
In the example below, both asset groups have identical assets except for the visual creatives. The only differences between the two groups are:
The focus of the creative asset: The “Good” asset group focuses on scenes of the product, while the “Average” asset group focuses on actual product images.
The types of creative assets: The “Good” asset group contains the maximum number of images and five videos, while the “Average” asset group contains the maximum number of images and no videos.
We tested this same asset group across all campaigns for over 30 days, using a scenic asset group versus a product image asset group.
In every case, the product image asset group featuring product bottles outperformed the scenic asset group every time, even though Google’s defined Ad Strength method implies that the scenic asset group will perform better.
The scenic asset group often garnered more clicks and a higher CTR, but it never generated revenue over that test period – only the product image asset group did.
While ad strength as a metric may appear to be helpful, use your actual conversion data from
Set your ROAS high, then go higher
If you set your ROAS goal at 200%, then you’ll get around 200%. But what if you set it higher?
One of Performance Max’s greatest strengths is its ability to maximize budget to achieve your desired ROAS. Push it to the maximum and use Performance Max for what it’s good at.
Check your outcomes in GA4 and Google Ads to see how the change impacts performance.
Beware of the cross-network default channel grouping
Performance Max campaigns appear as “cross-network” in GA4’s default channel groupings. Rolling all campaigns together under this aggregate grouping to evaluate performance can be tempting.
However, because Performance Max is AI-assisted, combining new campaigns with existing ones won’t provide the true clarity that individual campaign-level reporting will. Avoid the temptation to evaluate Performance Max on an aggregate level in GA4.
