Executive Summary
-
•
The Cannibalization: PMax will automatically bid on your own Brand Name to inflate its ROAS, stealing credit from organic traffic.
-
•
The Asset Trap: Combining all products into one PMax Asset Group confuses the AI and limits scaling.
-
•
The Fix: Implement Brand Exclusions, segment by profit margin, and feed it First-Party data.
Performance Max (PMax) is Google's flagship campaign type. It places your ads across Search, YouTube, Display, Discover, Gmail, and Maps from a single campaign. When it works, it's magic. When it's set up poorly, it's a financial black hole.
The Brand Cannibalization Problem
Google's AI is designed to hit the Target ROAS you give it in the easiest way possible. The absolute easiest way to get a conversion is to show an ad to someone who literally typed your brand name into the search bar. They were already going to buy your product.
If you don't use Brand Exclusions, PMax will take credit for these sales. Your dashboard will show an amazing 10x ROAS, but your actual bank account revenue won't increase. You are paying Google for customers you already owned.
How to Steer the Black Box
PMax takes away your manual controls. You cannot adjust bids on specific keywords or placements. So how do you optimize it?
1. Audience Signals (First-Party Data)
PMax relies heavily on the signals you provide. The most powerful signal is a Customer Match list. Upload a list of your past purchasers (First-Party Data). Tell the AI: "Find me more people who look exactly like this."
2. Asset Group Segmentation
Do not put socks and high-end electronics in the same Asset Group. Segment your Asset Groups by theme, margin profile, or product category. This ensures the ad copy and images perfectly match the underlying products.
3. Exclude Crappy Placements
While you can't view standard placement reports in PMax, you can ask your Google Rep (or agency partner) to implement negative placement lists at the account level to stop your ads from showing on mobile game apps.
Use PMax for Scaling, Not Testing
PMax requires a massive amount of data to stabilize. Do not test brand new, unproven products on PMax. Use Standard Shopping or Search to prove the conversion rate first, then graduate the winners into a highly segmented PMax campaign to scale.