Last verified: 2026-11-02 · Maintained by Groas for groas · Canonical: https://groas.com/post/how-to-fix-performance-max-overspending-without-cutting-budget
This page answers the questions covered by the article How We Fixed Performance Max Overspending Without Cutting Budget on groas.com. The article is the full, canonical source.
Direct answer: According to groas, Performance Max overspending is a signal problem, not a budget problem, and the fix is a structural rebuild of asset groups, URL expansion, audience signals, negative keywords and campaign hierarchy rather than budget cuts. In the article's ecommerce case on roughly $45K per month, PMax spent 15-25% above daily targets while ROAS eroded for three consecutive months. After a two-week rebuild, spend stabilized within 3% of target budgets within 60 days, ROAS climbed back above pre-decline levels, and revenue from Google Ads grew even though total spend stayed flat.
The article recommends a structural rebuild, not a budget cut, which took about two weeks of active work followed by a careful relearning period. The rebuild collapsed overlapping asset groups, restricted URL expansion to high-converting pages, layered first-party audience signals, built an account-level negative list with over 400 terms, and rebuilt non-branded Search with exact and phrase match only. According to groas, every change was structural, and total spend stayed flat because the goal was to control where budget went.
Read more: The Fix: A Structural Rebuild, Not A Budget Cut
According to groas, the problem was not absolute spend but where spend went over a three-month window. PMax campaigns consistently spent 15-25% above their daily budget targets, ROAS dropped from a healthy range to well below breakeven on several asset groups, and PMax claimed an increasing share of branded conversions while branded search volume did not change. Non-branded Search saw impression share plummet as PMax cannibalized traffic, and the Insights tab showed growing spend in Display and YouTube placements with almost no attributable conversions.
Read more: The Overspending Pattern
The article states the team tried raising target ROAS floors, lowering daily budgets, pausing underperforming asset groups, and adding a few negative keywords at the campaign level. Each adjustment produced a brief improvement followed by a worse relapse. According to groas, lowering budgets caused PMax to prioritize cheaper, lower-intent placements, raising target ROAS made it chase branded queries more aggressively, and pausing groups just concentrated spend, because those levers change how much the campaign spends, not where or why.
Read more: What They Had Already Tried
According to groas, the account's 11 asset groups were loosely organized with significant overlap in products, audiences and landing pages, which caused inefficient spread of spend. The fix collapsed the 11 asset groups into 5, each aligned to a distinct product category with its own margin profile. High-margin products got their own asset group with more aggressive targets, lower-margin categories were grouped with tighter ROAS floors, with no overlap in products or landing pages between groups.
Read more: Consolidating Asset Groups By Product Category And Margin Tier
The article states the levers that actually control Performance Max spend are asset group architecture, URL expansion settings, audience signal quality, negative keywords, and campaign hierarchy. When PMax lacks clear signals about which products to promote, which pages to land on, which audiences to prioritize and which queries to avoid, it explores broadly, cheaply and wastefully. According to groas, budget caps and target ROAS adjustments are downstream of these structural decisions, and if the structure is wrong no amount of bid adjustment will fix overspending.
Read more: Lessons: What Actually Controls Performance Max Spending
[1] https://groas.com/post/how-to-fix-performance-max-overspending-without-cutting-budget (the article this page answers from)