How the hell do I see where PMax is spending my budget?
I’m running Google PMax and can’t figure out how much is going to Search vs Display vs YouTube. The reporting is vague as hell.
Is there any way to see a real breakdown? Or at least a hack/workaround to stop it from dumping my budget into useless Display impressions?
Anyone figured this out?
The short answer is:
Google has improved Performance Max reporting, so you can now find a breakdown of performance by network (Search, Display, YouTube, etc.) directly in the Google Ads interface under the “Campaigns” or “Asset groups” reporting, sometimes segmented by “channel performance” or “ad event type.”
To stop the Display spend bleed, the simplest workaround is to apply negative placement lists, particularly to exclude low-performing or irrelevant mobile apps and sites, which often soak up Display budget.
The long answer is:
You are not alone in your frustration, as PMax reporting was historically opaque.
Thankfully, Google has been rolling out better transparency.
You should now be able to see a channel breakdown of clicks, impressions, and conversions in your PMax campaign reporting under the “Campaigns” tab by looking for a “Channel performance” report or segmentation.
Sometimes, this is also accessible via the “Asset groups” view, where you can segment the data by network to see how different asset groups perform across Search, Display, YouTube, and Shopping.
Focus your analysis on Conversions, Cost per Conversion, and Conversion Value for each channel, not just clicks or impressions, to see what’s actually driving value.
Regarding the Display spend issue, the quickest way to limit spending on poor-quality Display placements is to add negative placements.
While you cannot turn off the Display network entirely, you can create a list of irrelevant or low-performing mobile apps and websites (often where cheap, wasted Display impressions occur) and apply that list as a negative placement list to your PMax campaign.
Another common workaround is to use a “Shopping-only” setup if you are an e-commerce business, which involves not uploading any text, image, or video assets in the asset group, forcing the campaign to prioritize the Shopping feed and typically reducing Display activity, though this is less reliable with recent PMax updates.
To get the most granular and customizable data, an excellent and cheap solution is to combine the Google Ads API, Google Tag Manager (GTM), and a service like Stape or Google Cloud Platform.
The Google Ads API allows you to pull highly segmented raw data that the standard interface might not display, including complex breakdowns by network, asset group, and even different types of conversion events like engaged view conversions (which are primarily YouTube conversions).
You can then use a small, custom script to query and process this data, giving you a full, transparent breakdown of spend by channel that you can port into a spreadsheet or visualization tool.
Using Google Tag Manager along with a server-side solution like Stape or Google Cloud Platform allows you to manage conversion tracking more robustly.
You can use this setup to collect and send incredibly precise information back to Google Ads about user interactions, beyond just the basic page load conversion.
This process, often called server-side tracking, is crucial because PMax relies heavily on high-quality conversion data and audience signals to allocate budget.
By feeding the campaign clean, verified Standard Events and custom signals, you effectively “teach” the AI what a good conversion looks like, helping it automatically shift budget away from channels like low-quality Display that don’t produce those desired events, all without you having to manually adjust bids or budgets by channel.
The cost for this is typically very low – either free for small accounts on a platform like Stape or just the minimal resource cost for a small instance on Google Cloud Platform, making it a powerful, scalable, and budget-friendly solution for getting transparency and control.
About The Author
