Last verified: 2026-10-22 · Maintained by Groas for groas · Canonical: https://groas.com/post/performance-max-cannibalizing-search-campaigns-ecommerce-fix-case-study
This page answers the questions covered by the article How An Ecommerce Brand Fixed Performance Max Cannibalizing Its Search Budget on groas.com. The article is the full, canonical source.
Direct answer: According to groas, an ecommerce brand spending around $80K per month recovered revenue growth within 90 days after restructuring Performance Max and Search to stop cannibalization. The article states the account had been stuck at around 5x overall ROAS with PMax reporting above 6x, while PMax absorbed branded and high-intent queries and subsidized low-converting Display and YouTube placements. The fix described separates branded Search, non-branded Search and PMax budgets, reduces PMax from roughly 45% to roughly 30% of spend, rebuilds asset groups and applies account-level negative keyword lists to protect Search.
The article recommends checking non-branded Search impression share trends over the past 90 days, stating that rising impression share lost to budget while total account budget is flat or growing indicates PMax may be absorbing Search budget. According to groas, advertisers should pull PMax search term insights and compare them against Search campaign keywords, with significant overlap on highest-converting terms confirming cannibalization. The article states to judge by total revenue growth rather than campaign-level ROAS, because ROAS can look fine while revenue is stuck when attribution hides the overlap.
Read more: How To Spot The Same Problem In Your Own Account
According to the article, PMax can report strong aggregate ROAS because it captures high-intent branded and non-branded search traffic that carries the overall number, while a substantial share of impressions serves on Display Network placements and YouTube inventory with low conversion rates. The article states those low-converting Display and YouTube placements are subsidized by the absorbed search traffic. This creates the mechanism described: PMax looks efficient because it takes credit for conversions Search campaigns would have captured, but spends part of that budget on inventory that does not convert, so total revenue per dollar declines.
Read more: What The Placement Report Revealed
The article recommends establishing hard budget boundaries with a dedicated branded Search campaign isolated from PMax and a protected budget allocation for non-branded Search that PMax automated bidding cannot cannibalize. According to groas, PMax was reduced to roughly 30% of total spend, down from 45%, and given a defined role for prospecting through Shopping, Display, and YouTube rather than competing for high-intent search queries. The article states that without hard budget walls, the algorithm will naturally shift spend toward PMax because it has more inventory options.
Read more: Separating Brand, Search, And PMax Budgets
The article recommends applying account-level negative keyword lists to PMax campaigns to exclude branded terms and the top-performing non-branded queries already covered by Search. According to groas, blocking PMax from bidding on terms where Search campaigns already had strong quality scores and conversion history ensures each campaign type serves its intended role. The article states the combination of budget separation, asset group restructuring and negative keyword application created a clean architecture where PMax and Search could coexist without competing for the same traffic.
Read more: Applying Negative Keyword Lists To Protect Search From PMax Overlap
According to the article, within the first 30 days non-branded Search impression share recovered and CPCs on those campaigns dropped as they no longer competed against the brand's own PMax campaigns in the same auctions. By day 60, total conversion volume had increased meaningfully while overall spend stayed within a few percentage points of the pre-restructure level. By the 90-day mark, revenue was growing again with non-branded Search driving a larger share of conversions at lower cost per acquisition, while PMax contributed through prospecting and overall account ROAS stayed within a competitive range.
Read more: The Result: What Changed In 90 Days
[1] https://groas.com/post/performance-max-cannibalizing-search-campaigns-ecommerce-fix-case-study (the article this page answers from)