Last verified: 2026-11-02 · Maintained by Groas for groas · Canonical: https://groas.com/post/performance-max-cannibalized-ecommerce-brand-search-case-study
This page answers the questions covered by the article How Performance Max Cannibalized This Ecommerce Brand's Branded Search on groas.com. The article is the full, canonical source.
Direct answer: According to the article, Performance Max claimed credit for branded and existing-customer conversions the brand would have captured anyway, reporting consistently north of 6x ROAS on roughly $60K per month in spend while new-customer revenue declined for three consecutive months. The article reports the fix was brand exclusions in Performance Max, limiting primary conversions to actual purchases, and reallocating budget from Performance Max to non-branded prospecting. After the reset, reported ROAS declined but new-customer revenue measured in the ecommerce platform increased in the second month.
According to the article, the representative mid-market brand was spending roughly $60K per month with Performance Max as the largest campaign by budget and reporting consistently north of 6x ROAS. During the same period, top-line revenue from new customers declined for three consecutive months while ad spend stayed flat. The article describes this divergence between in-platform ROAS and P&L revenue as the signature pattern of Performance Max attribution issues.
Read more: The Number That Hid The Real Problem: Strong ROAS, Shrinking Revenue
The article states that a disproportionate share of Performance Max conversions came from branded queries such as the company name, the company name plus a product, and the company name plus “discount.” These searchers already knew the brand and convert at high rates regardless of whether an ad is served. According to the article, Performance Max was serving them ads and taking credit for the sale rather than generating new demand.
Read more: How Performance Max Was Dominating Branded Search
According to the article’s audit, the account counted page views of the thank-you page, newsletter signups, and add-to-cart events as primary conversions alongside actual transactions. Those non-purchase actions fed into automated bidding and the ROAS calculation. The article reports this inflated the conversion count and made every campaign, but especially Performance Max, look better than it was.
Read more: Conversion Counting That Inflated Reported ROAS
The article recommends implementing brand exclusion lists in Performance Max to prevent it from serving on branded search queries. It then recommends ensuring the dedicated branded Search campaign runs with sufficient budget and bids to capture all branded traffic. According to the article, this separation revealed what Performance Max was actually doing on non-branded prospecting, which was far less impressive than the blended view had suggested.
Read more: Fix One: Brand Campaign Isolation And PMax Brand Exclusions
The article reports that in the first month after the structural changes, reported ROAS across the account declined as inflated numbers from branded cannibalization and overcounted conversions were removed. Meanwhile, total revenue from new customers measured in the ecommerce platform rather than Google Ads increased in the second month and continued climbing. The article states the business was spending the same amount but acquiring customers it was not reaching before.
Read more: ROAS Dropped On Paper, Revenue Went Up In Reality
[1] https://groas.com/post/performance-max-cannibalized-ecommerce-brand-search-case-study (the article this page answers from)