Performance Max Asset-Group Rebuild Delivered 5.8x ROAS for an Illustrative DTC Brand

Summary

An illustrative DTC consumer-goods brand was running a single monolithic Performance Max campaign with thin creative and no audience signals, burning budget for a 2.1x ROAS. Asset-group segmentation, first-party data seeding, and a structured creative pipeline lifted ROAS to roughly 5.8x within five months without increasing total media spend.

The challenge

The brand had migrated from Smart Shopping to Performance Max when Google forced the change and let the default structure stand: one campaign, one asset group, a few stock images, and no customer match data. Google's algorithm was optimizing toward the easiest conversions it could find — mostly existing customers and branded searches — rather than driving incremental revenue. Reported ROAS looked passable but actual new-customer acquisition had stalled.

Approach

1. 1

   Week 1 — Asset-group segmentation plan

   Audited the product catalog and identified four natural asset-group segments by customer use case. Built a segmentation plan that matched the creative library to specific buyer intents rather than running one campaign for everyone.

2. 2

   Weeks 2-3 — First-party data seeding

   Uploaded 18 months of customer data as a Customer Match list. Built lookalikes at 1%, 3%, and 5% thresholds. Attached segmented audience signals to each of the four asset groups so Google's algorithm had targeting inputs beyond "find our customers for us."

3. 3

   Weeks 3-4 — Creative production

   Produced 10+ image assets and 3+ video assets per asset group. Mix of studio shots, UGC-style content, and lifestyle photography. Each asset group got creative tuned to its specific use case rather than one generic library.

4. 4

   Weeks 4-6 — Launch and learning phase

   Launched all four segmented asset groups simultaneously. Protected the learning phase by leaving asset groups untouched for 3 weeks. Account-level brand exclusion added to prevent PMax from claiming credit on branded-search conversions.

5. 5

   Months 2-5 — Creative rotation and scaling

   Bi-weekly creative rotation based on asset-level performance reports. New-customer-acquisition toggle turned on in month 3 after sufficient conversion data accumulated. Budget scaled cautiously (20-30% per week) to avoid re-entering learning phase.

Results

ROAS climbed steadily from month 2 as Smart Bidding learned against the segmented asset groups. The biggest single driver was the customer match list — PMax suddenly had real first-party data to shape its targeting instead of relying on its own black-box interpretation of site traffic. New-customer revenue (excluding remarketing and branded-search noise) grew disproportionately, which was what the business actually needed. By month five, the account was producing meaningful new-customer acquisition at a ROAS that justified scaling.

Services used