How Does Google’s PMax Stack Up Against Standard Shopping?
Clay Cohen, Sr. Director of Product Marketing
Published 03/22/2023
Optimizing your PMax campaigns for maximum incremental conversions takes a bit of trial and error. In this post we’re answering the top 10 FAQs from our recent Google PMAX webinar.
Google’s Performance Max (PMax) is all the rage. Learning how to allocate spend across all the Google ad properties is critical for brands leaning into this new channel. Although PMAX promises to remove the burden on brands to self-allocate their spend, there are some trade offs to be aware of. For advertisers, more convenience means less control and visibility into campaign performance. By testing for incremental lift over the past year, we’ve started to uncover some interesting insights.
Following up on the Webinar
During our latest webinar, our expert analysts teamed up with SVP of Ecommerce for RealTruck, Luke Coltrin, to discuss our findings from over 50 incremental lift tests, and help brands learn how to approach PMax more strategically, and make the algorithm work for them.
Our latest webinar revealed some exciting new insights for brands looking for guidance
About 61% of Measured brands invested in PMax in the second half of 2022, with a large percentage allocating upwards of 50% of total Google spend towards the channel
What Is PMax?
PMax operates as a Google aggregator, and serves across all of Google’s inventory, such as YouTube, Search, Shopping, Display, etc. With it allocating your media budget for you, it makes sense to why our brands wanted to learn how well it actually performed.
Results were certainly mixed from our initial testing period. Watch the video from last week for a more in-depth overview of the results.
We also had a lively Q&A during the session, which is included below if you’re curious about what your peers are asking.
Top 10 FAQs from our Webinar
We had over 70 questions during this webinar! We have aggregated the questions into the most popular questions, and answered them below.
Can you please explain how your geo testing works?
Measured uses matched market testing in which our product will identify the optimal test markets (cells) representative of the country overall. Within each test cell, only one factor is changed (i.e. turning off a specific channel’s spend) and then compare actual conversions observed during the test vs. the predicted normal level of conversions without the change made (called the counterfactual).
The difference between these two is the incremental contribution (resulting drop in conversions) from the cell in question. That amount is compared against what the vendor reports to calculate an incrementality coefficient.
For example, if we calculated 50 incremental conversions (sales lost in test cell) and a vendor reported 100 in that same time period, incrementality would be 50 / 100 = 50%. For most vendors, the geo targeting parameters are managed via API with no manual intervention required from the client/agency.
What are some of the specific considerations in testing? PMax?
Given the dynamic nature of PMax, the Measured team will work closely with clients to ensure tests are set up to maximize the chances of statistically significant reads.
We primarily want to ensure PMax and other Google tactics will not unexpectedly increase spend during the test period, which would cause contamination of the results.
You will also want to ensure PMax has been turned on long enough to have stabilized and exited the learning phase before commencing testing.
What recommendations do you have when setting up campaigns to deliver the most incremental conversions possible?
Generally, Measured recommends our clients do the following:
- Work with Google to exclude branded keywords from PMAX campaigns
- Optimize for new customers and maintain updated customer lists on Google
- Manage ROAS targets in the platform appropriately (very aggressive targets can push PMax to deliver lower funnel/less incremental inventory).
Ultimately, it's important to consider both incrementality AND in-platform (PMax) ROAS to arrive at incremental ROAS, or ROAS(i) that meets your business objectives.
In other words: Incrementality x Pmax ROAS = Incremental ROAS
How would you recommend implementing PMax if higher ROAS targets result in less incremental inventory?
Testing PMax initially delivers baseline incrementality. Using this to modify in-platform targets accordingly and then retesting can confirm how much of a concern this is.
Our philosophy at Measured is to always test, learn, and act on the information and optimize media accordingly.
How should I evaluate PMax in the context of other tactics? Is it fair to compare a full funnel tactic like PMax to something that is upper or lower funnel?
The beauty of using Measured is we normalize this performance for you across the entire funnel.
Typically, based on in-platform conversions from vendors, upper funnel campaigns receive less credit than they should and lower funnel campaigns receive more than they should.
By implementing Measured and further testing your specific campaigns you can quantify the exact % adjustments you need to make to compare upper vs. lower-funnel vs. a full-funnel tactic like PMax. The Measured team can work with you to understand the implications for cross channel media allocations and how you can use the information from our Media Plan Optimizer to set the appropriate ROAS(i) targets for each tactic.
We have concerns that PMax is taking away from our branded searches. Have you seen direct correlations with this?
Our goal is to help you understand how to maximize the incremental ROAS of your media dollars.
Generally, we've found that PMax behaves somewhere in the middle of the incrementality values we've seen across the other Google tactics.
How should I be thinking about managing multiple KPIs (i.e. new customers and sales)?
Measured can be set up to test different experiments and you can understand the incremental effect on multiple KPIs by testing various times.
Our incrementality model has also been calibrated against different conversion types. Ultimately, you can evaluate the performance across the multiple outcomes versus your business objectives to determine what's right for your brand.
How do you normalize incrementality values for different attribution windows in Google (last click, DDA, etc.)?
Our geo test results are reflective of the brand-specific settings. The results discussed in our webinar reflect our general findings across windows and, in our incrementality model for brands who aren't actively geo testing, we account for the windows in use when delivering the Measured Incrementality Model value.
Have you done analysis on how PMax specifically compares with Standard Shopping?
Google seems to push PMax as a replacement/improvement. Is it actually?
Our findings indicate incrementality % on PMax (ie: the accuracy of it’s reporting) is on average higher than traditional Brand Shopping but lower than Non-Brand Shopping. However, you have to also consider conversion rate in order to evaluate overall effectiveness.
For example, Tactic A has $5 Google-reported ROAS with an incrementality of 50% = $2.50 ROAS(i).
Tactic B has $10 Google-reported ROAS with an incrementality of 25% = $2.50 ROAS(i).
In this case, both tactics performed equally well despite incrementality differences. It's important to consider the tradeoff between incrementality and in-platform ROAS when determining effectiveness. With a 50% reduction in incrementality, you would need to see in platform ROAS improve by 2x to net out at the same place.
Were these tests done alongside other campaign types within Google or were they the only ones live?
Our tests were run alongside a variety of conditions - some clients are just trying out PMax, some have allocated over 50% of their Google budget to the tactic.
The PMAX algorithm is very complex and it is difficult to estimate how the overall spend on PMax vs the rest of Google impacts incrementality. There are likely many dynamics at play - industry, brand awareness, campaign setup, retailer presence, etc.
I know this webinar is specific to PMAX, but it seems so many of these findings could probably be said about Meta’s ASC too.Anyone at Measured able to speak to that high level?
ASC is a newer tactic than PMax and Facebook has not yet allowed geo targeting that would be required to facilitate these types of tests.
We've heard this capability is coming soon and it will be a high priority for us to test it once available.
Measured's assumption is that many of these same principles would hold, though campaign setup is different between the two and ASC gives you some more control by excluding certain audiences or constraining the amount of spend on past customers. Please keep your eyes peeled for more information as we have it.
Should an advertiser just run all PMax? And allow Google to allocate the whole budget e.g. like a mutual fund.
As performance results varied widely across different brands, there is no one-size-fits-all answer for this. Instead, we advocate a test-and-learn approach where advertisers can understand the tradeoff between conversion rate and incrementality.
How those two things interact, along with your comfort with less control and the "black box," will inform how much of your budget to shift to PMAX. It's designed to take in more data points and maximize conversion efficiency, but it's critical to understand the incrementality of those conversions before you trust it blindly
Interested in testing PMAX for your brand?
The Measured Incrementality Platform is purpose-built to help consumer brands measure, analyze, and optimize their media spend to maximize incremental business impact. Schedule a personalized demo today to take the guesswork out of black box reporting of PMax and other hard to measure channels.