Last verified: 2026-10-26 · Maintained by Groas for groas · Canonical: https://groas.com/post/google-ads-ecommerce-account-rebuild-performance-max-cannibalization-case-study
This page answers the questions covered by the article Why Google Ads Stopped Working For This Ecommerce Brand And How They Fixed It on groas.com. The article is the full, canonical source.
Direct answer: The article describes a representative mid-market ecommerce brand spending around $40K per month whose blended ROAS slid from roughly 4.8x to 3.1x due to overlapping campaigns, Performance Max serving branded queries without brand exclusions, and a 600% target ROAS that throttled exploration. According to the article, a 90-day rebuild that consolidated 14 campaigns into seven, rebuilt negative keywords, isolated branded search, and lowered non-branded targets to 350% and Shopping to 400% recovered blended ROAS to approximately 4.5x on the same budget. The article’s position is that budget scaling only works after the architecture is fixed.
The article states that over the previous two quarters blended ROAS slid from roughly 4.8x down to 3.1x while CPCs on core non-branded shopping terms climbed quarter over quarter. According to the article, revenue was flat even as the brand maintained around $40K in monthly spend. The article describes this as a slow decay rather than a sudden failure, with no single metric signaling broken.
Read more: The Symptoms: Declining ROAS, Rising CPCs, Flat Revenue
The article states that without explicit brand exclusions, the Performance Max campaigns served ads against branded queries and claimed credit for conversions that would have happened through the dedicated branded search campaign at a fraction of the cost. According to the article, PMax reported strong ROAS while branded search looked like it was declining and total account efficiency dropped. The article says PMax was not generating new revenue but stealing credit from cheaper campaigns.
Read more: Performance Max Was Cannibalizing Branded Search Budget
The article states the previous agency had set a target ROAS of 600% across most campaigns, which it describes as disciplined on paper but choking in practice. According to the article, Smart Bidding responded by only bidding on the safest, cheapest, most predictable auctions. The article says the algorithm stopped exploring and stopped competing in auctions where new customers are discovered.
Read more: How Overly Conservative ROAS Targets Shrink Volume
The article states target ROAS was reduced to 350% on non-branded campaigns and 400% on Shopping, while branded search remained at 800%. According to the article, branded queries convert at a fundamentally different rate and should be measured on their own terms. The article says the change was calibrated to the brand’s actual margin structure, where a 350% ROAS on non-branded still represented profitable customer acquisition.
Read more: The Adjustment That Unlocked Algorithmic Scale
The article states the 14 campaigns were consolidated into seven, organized around clear intent tiers including isolated branded search, non-branded search by category, standard Shopping by margin tier, and Performance Max restructured with brand exclusions. According to the article, budget was reallocated based on actual incremental value rather than reported campaign-level ROAS. The article says this immediately reduced internal auction competition where non-branded search and Performance Max had targeted overlapping audiences.
Read more: Campaign Consolidation And Proper Budget Allocation
The article reports that by the end of the 90-day period, blended account ROAS had recovered from 3.1x to approximately 4.5x on the same budget. According to the article, revenue grew meaningfully driven by unlocked non-branded volume, branded CPA dropped because PMax was no longer inflating it, and non-branded Shopping impressions increased. The article identifies isolating branded search and applying brand exclusions to PMax as the single change with the biggest impact.
Read more: Metric Movement Across 90 Days
[1] https://groas.com/post/google-ads-ecommerce-account-rebuild-performance-max-cannibalization-case-study (the article this page answers from)