Performance Max (or PMax) campaigns are gaining traction among search advertisers.

Google’s machine learning resources optimize ad placements across Google’s entire network of inventory. This lets advertisers maximize their efforts and reach their target audiences more efficiently. 

However, there are always trade-offs when taking advantage of these proprietary capabilities, especially as we lose more visibility and levers to pull. 

In general, we have limited visibility into the data for these campaigns, making it difficult to make informed decisions about how to best optimize campaigns and allocate budget. 

Additionally, advertisers often need to analyze their aggregated data across all campaign types and platforms. This can require:

  • Fetching data through the Google Ads API.
  • Loading it into a larger data warehouse for further manipulation and analysis.
  • Joining it with Google Analytics data to create a fuller picture of the user journey. 

Here are some limitations to be mindful of when extracting and analyzing PMax performance data within and outside the Google Ads interface.

1. Limited granularity of PMax data

PMax campaigns offer limited reporting options than other Google Ads campaigns, which can make it difficult to analyze performance in the ways we are used to. 

Typically, Google Ads campaign data can be fetched by accessing the standard report through the API. You can define the level at which you would like to segment the data, even down to the keyword level. 

Since PMax campaigns use machine learning to determine the best placements for ad delivery, no ad groups or keywords are associated with these campaigns. 

Therefore, a standard report generated at any level that is more granular than the campaign will contain several irrelevant fields to PMax and exclude all data from these campaigns entirely rather than simply nullifying the irrelevant fields. 

To capture your standard and PMax campaigns, you must call the API multiple times and retrieve two separate data connections that can later be loaded and unioned within your data warehouse. 

  • The first should be a standard report at the desired level of granularity, which will contain no PMax campaign data. 
  • The second should also be a standard report at the campaign level, but this time should exclude all campaigns that are not PMax to avoid duplicate data. 

Also, be aware that many custom reports and segmentations can be helpful for campaign analysis, such as Performance Max Placement. 

They cannot be retrieved through the API and can only be viewed in an isolated environment within the Google Ads interface. 

Get the daily newsletter search marketers rely on.

<![CDATA[ function getCookie(cname) { let name = cname + "="; let decodedCookie = decodeURIComponent(document.cookie); let ca = decodedCookie.split(';'); for(let i = 0; i

2. Google Analytics insight requires thoughtful navigation

With the rollout of Google Analytics 4 and the subsequent deprecation of Universal Analytics, advertisers can use website and app data to understand the customer journey and post-click website engagement activity. 

Any website traffic generated by PMax campaigns should be carefully viewed and analyzed.

For starters, you will not see PMax data fall under the default Paid Search channel grouping but a separate grouping called Cross Channel containing PMax and Smart Shopping campaign data. 

Be wary of placing any dimension filters incompatible with PMax campaigns.

Unlike the API issues noted above, where the data will not show up, these filters will cause data to display incorrectly within the GA4 interface and cannot be relied upon.

For this reason, gaining cross-channel insights that include PMax campaigns within GA4 can be challenging.

Additionally, PMax campaigns count engaged view conversions.

This type of conversion is highly valuable, as it is more specifically tailored to video advertising and the user behavior that follows a video ad as opposed to other ad types and is a strong indicator of engagement.

Just be aware that Google Analytics, by default, doesn’t count these conversions and will need to be intentionally configured to do so. 

3. Traditional analysis methods may not apply

Given the above issues, using the Google platforms to generate reporting and insights on PMax campaigns in isolation is always an option.

When viewing the available data within the platform, it is essential to be mindful of all the various limitations surrounding this data and know that traditional analysis tactics may be neither effective nor possible. 

For example, while some basic reporting templates are within the platform for PMax campaigns, advertisers cannot customize any reports or create custom metrics. 

Another factor to consider is that since PMax campaigns are optimized on real-time data, campaign performance should be analyzed closer to real-time and rely less on historical data and trends, as the algorithms are constantly adjusting to maximize optimizations.

This reliance on real-time data also makes it difficult to conduct traditional A/B tests, especially because we don’t have control over things like ad placements, formats, creative elements, or audiences we can isolate to test hypotheses. 

Instead, you can only run tests comparing your PMax campaigns to standard shopping campaigns or run an uplift experiment demonstrating how adding a PMax campaign to your existing campaign mix can increase conversion volume.

Other examples of insight that we lose out on with PMax campaigns include audience targeting, ad placement and budget control. 

While this is all by design, it can be a difficult adjustment for advertisers to lose the ability to have a say in where their dollars are allocated.

They may have neither the time nor the budget to allow the campaign to run long enough to gather sufficient data to maximize efficiency. 

And while PMax does optimize based on audience behavior and ad creatives, they do not provide detailed data on these behaviors or how individual headlines or images may perform. 

Opinions expressed in this article are those of the guest author and not necessarily IXLCenter.io. Staff authors are listed here.

Share This