Performance Max asset group structure and hygiene

Performance Max Asset Group Structure and Hygiene

Applies when a campaign is serving across the automated inventory Performance Max covers, and delivery quality depends on the inputs you gave it, not just its own targeting.

The play

1. Split asset groups by product line or audience segment before doing anything else. One asset group trying to speak to every product or every buyer can't be coherent to any of them — that's a structural problem no amount of extra copy fixes.
2. Fill every slot the platform gives you, not just the minimum required — headlines, long headlines, descriptions, and a real image and video mix across horizontal, square, and vertical orientations. Ad strength is a direct readout of how much of that room you've actually used, and a group short on assets or asset variety caps its own performance ceiling regardless of targeting quality.
3. Upload real video. If a group has none, the platform will assemble one from your other assets automatically — treat that as a fallback you're accepting by default, not a feature, and replace it with something you'd choose deliberately.
4. Give a new or heavily-changed asset group real time before judging it — check ad strength and asset performance again after a couple of weeks, not a couple of days, and don't restructure on day-three volatility.
5. Watch for brand cannibalization specifically: if branded queries are converting through Performance Max at a worse cost than a dedicated brand search campaign already covering them, that's not incremental reach — it's the same demand at a worse price. Exclude your own brand once you can see that pattern.
6. Read search term insights the same way you'd read a search terms report anywhere else — clicks with cost and no conversions are negative-keyword candidates, even though applying them here usually takes a different path than a standard search campaign.

What good looks like

• The best operators treat "Low" performance labels on individual assets as a standing queue to clear, not a one-time cleanup — replacing the weakest asset in a group is cheaper than restructuring the whole group later.
• The common mistake is judging a fresh asset group on its first few days of volatility and restructuring it before it's had a real chance to settle — that resets the very learning period the impatience was trying to fix.
• A good structure is legible from the outside: you can point to any asset group and say which product or segment it exists for, and every low performer in it has a named replacement candidate, not just a general sense that "creative could be better."

Rules

• MUST split asset groups by product line or audience segment rather than serving everything from one group.
• MUST provide at least one deliberately-chosen video per asset group rather than relying on auto-generated video by default.
• NEVER judge or restructure a new asset group before giving it a real settling period — treat early-days volatility as expected, not diagnostic.
• NEVER leave brand cannibalization unexamined if a dedicated brand search campaign already covers the same queries at a better cost.