In the exploration of Google Search in 2023, our focus centres on the Performance Max (PMax) campaign type. Through the utilization of our proprietary testing database, Hippocampus, we examine the broader impact experienced by 18 Brainlabs clients across diverse sectors, aiming to assess PMax’s performance and significance.
The results reveal the strengths of PMax, demonstrating its ability to achieve superior CPA and ROAS outcomes compared to similar Google Ads campaigns. We also delve into the crucial inputs that contribute to success within the PMax environment, namely data, creative assets, and feeds.
We also provide valuable insights and recommendations for effectively harnessing PMax’s potential and navigating the ever-evolving landscape it presents.
Google Search in 2023: Is Performance Max a must?
Jan 2023
Andy Goodwin,
Head of Paid Search UK
at Brainlabs
Index
- Changes to Google as we know it 3
- What do the changes to Google Search mean for advertisers? 5
- So, is Performance Max any good? 6
- How do brands win at Performance Max? 8
- Our wishlist for Performance Max 13
- Final takeaways 15
Changes to Google as we know it
We’re on the cusp of a very exciting time in the search landscape – and for once, we’re not referring to automation, privacy or first-party data – but rather to an overarching facelift to Google Search as we know it.
For decades now, Google Search has relied on its familiar model of “ads at the top, and ten blue links”. Whilst SERP features have evolved over time with additions such as creator video content & infinite scroll, the overall way we interact with search has remained largely unchanged.
One glance at the announcements coming out of Search On 2022 will tell you we’re not far away from Google heading in a very different direction with their user-facing products, with the aim of making search more natural and intuitive.
- Ads at the top
- Ten blue links
Visual experiences were the predominant theme of Search On, and there are plenty of plans to bring that to life in the outputs we see – whether it’s using Maps to reconstruct a 3D visualisation of an area, or creating 360° spins of product results.
But the outputs are only part of the picture, and arguably more interesting are the advances in Google’s ability to process visual inputs, particularly when it comes to Multisearch. Combining text, image & voice inputs will allow users to search more intuitively, as well as to process queries that would be difficult to distill into a standalone text string in the search bar.
Many use cases were presented, and some of our stand-out favourites were:
- Information overlays onto realtime camera images, via “Live View” in Maps
- More seamless translation of languages embedded within images
- Combining image & text inputs to find products more tailored to your needs
Images courtesy of Google
What does this mean for advertisers?
Whilst we’re yet to see exactly how these changes play out, it doesn’t feel too bold a prediction to claim that Performance Max will be absolutely central as the buying-engine for brands to tap into these new user experiences.
Performance Max – often referred to as PMax – came onto the scene as a beta product at the tail-end of 2020, and has been rapidly picking up momentum ever since. Both its benefits & its shortcomings have been hotly debated over the last 2 years.
At the heart of its proposition, PMax strives to single-mindedly hit an advertiser’s performance objective, without constraining itself to any one media placement in Google’s ecosystem. However, this media-agnostic approach, along with the associated lack of visibility into what is actually driving campaign performance, has been a major talking point for the industry as it gets used to this new ad product.
Image courtesy of Google
So, is Performance Max any good?
To weigh in on the debate, we tasked ourselves with assessing whether PMax was actually delivering on the performance it promised.
As we’re often prone to do, we turned to our proprietary testing database, Hippocampus. Named after the memory centre of the brain, we use Hippocampus to assess the macro impact witnessed by our clients across a wide range of digital media tactics.
We aggregated test results from 18 Brainlabs clients in a range of sectors, and identified two topline findings:
- Most importantly, PMax drove stronger CPA & ROAS than similar Google Ads campaigns. This supports the proposition that, by having the freedom to optimise across multiple touchpoints, the algorithm is able to drive outcomes more efficiently for advertisers.
- What’s more, PMax is already making healthy contributions to overall conversion & revenue volumes. On average, our clients who used PMax netted out at spending 17% of their Google Ads budget through this campaign type. Note that this figure does vary considerably – for instance, many retailers invested in excess of the 17% average.
Median uplift metrics:
18% Conv.
-12% CPA +16% ROAS
30% Rev.
Data relates to the % of convs/revenue achieved by PMax vs. similar campaigns in Google Ads, and the % difference in CPA/ROAS vs. similar campaigns. Revenue & ROAS metrics are measured across the subset of clients that tested PMax & who also record conversion value within Google Ads.
“Google’s PMax proposition is centred around delivering results… and it’s fair to say that PMax achieves this pretty comprehensively”
Andy Goodwin, UK Head of Paid Search
How do brands win at PMax?
As well as figuring out if it could do what it said on the tin, a secondary aim of our testing was to identify exactly how brands could differentiate themselves in a World of PMax. With the nuts and bolts of media optimisation accounted for, what can agencies & advertisers do to drive further success through PMax?
We identified three main inputs that represent a substantial amount of the value that marketers can add to PMax campaigns. They represent quite a shift away from the more familiar skill-sets of Google Ads practitioners, so it’s important to start thinking today about how you can bring this to life for your brand.
1 Data 2 Creative 3 Feeds
Data
In this context, we’re referring to the data that we feed into Google to direct its optimisations. PMax offers less control over other aspects of your campaign, so there’s now greater emphasis on defining your objectives.
In practice, this is achieved through first-party data integrations, paired with value-based bidding. It’s no coincidence that on average we saw PMax deliver +2% better efficiencies when paired with value-based bidding.
In our eyes, thinking more strategically about the challenges our media is solving can only be a good thing. It encourages us to get closer to customers, and to really contemplate what drives value for the business.
We’ve seen that PMax will achieve what you ask it to, but have you told it to do the right thing?
Margin
First-time buyer Returns
Account creation
Client Database Brainlabs Offline Upload Tool Brainlabs Audience Sync Tool
High value customers
Repeat purchasers
Margin based bidding
Segmented audience messaging
Creative
When it comes to creative, the number one rule is don’t use the auto-generated assets. Whilst anyone who’s played around in a PMax campaign for 5 minutes could probably tell you this already, it’s interesting to note that Google themselves make point of stressing this; they acknowledge that whilst auto-generated creative is getting better, it does not compete with advertiser assets. If you’re going to test PMax, make the effort of building all the assets.
More generally, learning to optimise image & video creative for performance purposes will be a new challenge for a lot of specialists who’re most familiar with the trusted text ad format. Whilst creative insights in PMax aren’t hugely forthcoming right now (more on that later), we’re keeping our fingers crossed for greater creative testing capabilities being introduced into the PMax roadmap.
A B
Feeds
For eCommerce brands, feeds represent the 3rd key pillar for success. Shopping placements are crucial for many retailers, and the pivotal role feeds play in those auctions mean they’re an essential lever to test & optimise.
For non-retailers, feeds aren’t to be ignored either – they may not play much of a role today, but indications from Google suggest that feeds will come into use more broadly to help connect PMax with the more visually rich SERP experiences that are being developed. This will touch much more than just the retail vertical, so it’s worth keeping an eye on any developments in this space.
Exactly how this will looks for each sector remains to be seen – but it doesn’t take too much imagination to envisage, for example, how feeds would be incorporated to help promote the inventory of travel brands.
Feeds Determine:
- How much we need to pay to win the auction
- How competitive we are in the ad auction
- Whether our ads are eligible to show
- The information a user sees
Agency model
Bonus insight
And finally, whilst it didn’t make it into our top 3, it’s worth also giving some thought to the structure of the agency doing the media buying. It doesn’t sound like the most exciting topic, but it is important: the objective-focussed approach of PMax is at odds with the more traditional, channel-siloed agency models. Internal challenges such as “which department should run this campaign” can throw up blockers to effective activation on PMax.
Agencies with a less siloed structure have greater freedom when it comes to budget allocation, and so don’t face these same difficulties. For a number of years now we’ve designed our biddable teams around Client Partners, who are tasked with taking a client brief and translating it into a multi-channel plan, by tapping into the various channel specialists within their immediate team.
Client Partner Client Brief
Our wishlist for PMax
No article on PMax would be complete without acknowledging its limitations, and this one is no different. There are elephants in the room, and they need to be addressed.
We know that Google themselves would want feedback to improve on what is already an incredibly valuable product in the industry. It’s our belief that the vast majority of industry concerns stem from one of two places – and so, to make PMax work even harder for our clients, we’d hope to see Google’s product roadmap focus on two key areas..
1. Incrementality
Firstly, we can all agree that any performance media product should be all about growing the business in the short-term. This necessitates a focus on incremental conversions & revenue.
Whilst PMax places more emphasis on defining objectives, it feels like a bit of a miss that you can’t actually set a PMax campaign to the “correct” objective – ie one of incremental outcomes. This isn’t such an issue when media channels are kept in silo, because we can isolate the incremental contribution of each channel through tests, and subsequently refine our spend allocation.
However, when you can’t optimise for incremental outcomes, and everything is housed together under one roof, this leads to some of the very valid concerns that have cropped up – for instance, around the identical treatment of branded & generic queries.
And whilst Google are correct in pointing out that the DDA model used in PMax is weighted towards incrementality, that can only ever mean incrementality within Google’s ecosystem – which doesn’t capture the whole picture.
2. Reporting
Google engineers look away now, you’ve heard this one before – but item #2 on our wishlist is to have greater access to raw data breakdowns within PMax.
When raised in Google forums previously, there’s been a tendency to deflect towards the Insights page. The conversation tends to go something like this:
- We want more data!
- What do you need it for?
- We don’t know… but we know we’ll need it for something!
On the face of it, we’re not putting forward a very compelling case. But the point that’s being missed is that, when it comes to answering marketers’ most pressing, strategic challenges, it isn’t possible to pre-empt the questions we’ll have. The process of building a marketing strategy throws up questions that are bespoke to each individual brand. Google Ads’ raw data has historically been a fantastic source to answer some of these questions – but we’re doubtful this can ever be done through a “generic” suite of insights.
On the other hand, it’s not hard to see why Google are going down this route. Quite frankly, many practitioners haven’t kept pace with the changes in machine learning over the last few years, leading them to misinterpret data and make suboptimal decisions in their campaigns – understandably something that Google want to avoid. Our view is that whilst these more directional insights might provide a helping hand for the less skilled practitioners, they actually curtail the potential of those who are most fluent in automation & machine learning.
Final takeaways
To wrap up
Just over 2 years on since its beta launch, and Performance Max is already driving undeniably impressive results. For performance marketers, it’s become a topic can’t be ignored; results have to be our number 1 criteria for success.
If anything, the evolution of Google’s user-facing products will only lead to PMax becoming a more integral part of their Ads offering. We believe there’s still huge headroom for PMax to get better – and we’ll continue to agitate for the changes that we want to see. Whilst we can’t disclose the specifics in this report, the good news is that we are now starting to see some more exciting, additive features materialise in the PMax product roadmap.
For now, start to think about the capabilities you can build into your team to make the most of this format, across data, creative, feeds – and potentially even your media agency. It’s never too early to start testing.