Performance Max: Over a Year in, Are We Prepared for a Keyword-less Future?

While I believe search will continue to thrive and drive results for advertisers, how we buy search is undoubtedly changing as is the search engine results page (SERP) itself. One of the largest shifts in how we buy across the Google space has been Performance Max (PMax). It’s a keyword-less, AI-powered campaign type that uses machine learning models to optimize bids and placements across Google (including search) to hit a core objective. Advertisers don’t bid on specific keywords; instead, they rather rely on AI to handle bidding and targeting through audience signals, the advertiser's own website/URLs, and creative assets. This serves ads across the Google network to match search queries and browsing behavior to those most likely to convert to the desired action. So, while the search experience is similar to the user, how we as digital marketers buy search is changing rapidly.

Early adoption of Performance Max was mixed and vertical-specific—now the tide is shifting.

Advertisers have embraced PMax with mixed readiness, many fighting the loss of control they have come to expect from Google ads over the years. In the earlier days of PMax, I felt that hesitance as well, especially for non-retail verticals and for very complex advertisers. I also knew from years of navigating similar evolutions in Google Ads that there would inevitably be new features and modifications based on feedback from agencies and advertisers. Sure enough, our testing has shown that while we may not be ready to say goodbye to keywords, PMax does get us one step closer to the option of scalable keyword-less targeting.

Google is now transitioning to a more visual SERP with a reliance on visual formats and a generative search experience. Google has bet big on Performance Max, Broad Match, and AI-driven products in general. Google has also released more insights, creative tools, targeting, and testing levers in the platform to improve the product and allow for more insightful ways to leverage data across marketing efforts, including informing audience selection and creative. Additionally, Google is testing and launching automated and generative asset creation to help advertisers with the hurdle of costly and time-consuming creative iterations.

Let’s recap how PMax has evolved in the last year.

There continue to be new features launched to improve this product and improve performance for campaigns. I’ve detailed out a few below that are more significant in terms of our utilization.

• Uplift experiments and PMax vs. rSC (standard shopping) testing: These tests allow advertisers to compare the performance of PMax campaigns against other types of campaigns, such as standard shopping campaigns, and against other PMax campaigns with different settings or strategies.
• Brand exclusions and account level negatives: These features allow advertisers to prevent their PMax campaigns from showing ads for certain brands or keywords. This can be helpful for preventing ads from showing for competitors or for keywords that are not relevant to the advertiser's business.
• Video builder upgrades and improved asset-level group reporting: These features make it easier for advertisers to create and manage their PMax campaigns. The video builder allows advertisers to create videos for their PMax campaigns without the need for any video editing experience.
• The improved asset-level group reporting provides advertisers with more insights into the performance of their PMax campaigns.
• Final URL expansion helps you optimize your Pmax campaigns' performance. This will help find a more relevant landing page based on the user's search query and intent and allow for a customized dynamic ad headline that matches the landing page content.
• SA360 floodlight bidding support which allows advertisers to use their own conversion data to bid on PMax campaigns. 
• The ability to not input assets and run shopping/feed ads through the PMax ad type. While it’s not 100% guaranteed that auto-generated assets won’t run, our testing proves that the majority of spend goes to shopping placements.
• Allowing for scripts to be leveraged to showcase spend by tactic (video views, etc.).

We were an early adopter and rolled out a very systematic approach to testing.

At Media.Monks, the mid-market team has been highly committed to testing PMax since the initial product launch announcement, working to incorporate all its new functionalities and experiment options. We’ve tested PMax against standard shopping and Dynamic Search Ads (DSA). We’ve also tested new customer bidding, and even multiple asset groups with distinctly different creative versus just a single asset group and more. While we tend to see higher performance across our retail clients utilizing the feed-based and local solutions, we are now seeing growth in other verticals, even with sensitive lead-gen advertisers such as healthcare, as a result of new features.

Here’s an example of our testing. A healthcare client recently did a head-to-head test with DSA and was able to scale 10x in investment and lead volume by tapping into the power of PMax. In a few short months, the shift to PMax drove a 13% increase in leads with an 8% reduction in CPA. Setting thoughtful, flexible targets; breaking out campaigns by business need; having objective targeted creative; and giving machine learning the space to find the right people at the right time with first-party data allowed the brand to scale.

Here is what we anticipate will come next:

• Visual/image site links will be favored over text site links.
• There is a possibility that PMax is the end goal for Google Ads, with limited to no ability to bid on selected keywords (i.e., a keyword-less future).
• 2024 could be the year we lose other match types. I’m hoping we keep exact match, but we know broad match will stay for some time given the investment in the new broad match.

A look ahead at the road ahead for keyword bidding.

Over the last year, Google has been an excellent partner, listening to feedback from experts in the field and investing in more data insights, testing ability, training, and targeting controls. In addition, the new generative creative levers in the platform can arm us as search marketers to harness the power of AI while utilizing our knowledge of the Google network and specific brand needs to drive client wins.
Are we ready for a fully keyword-less search buying experience today? No, but given the progress of Performance Max over the last year, it’s becoming more likely that pairing broad match and Performance Max will be in every best practice deck from Google partners for the foreseeable future.