Case Study: How an Auto Dealer Recovered $47K/Month in Wasted PMax Spend

The Client

A regional auto dealer group operating 8 dealerships across 3 states, selling both new and used vehicles. Combined monthly Google Ads spend: $215,000, with approximately 60% allocated to Performance Max campaigns.

The dealer had been running PMax for 18 months and reported satisfaction with results. Their agency reported a 5.2x ROAS—well above their target of 4x.

The Problem They Didn't Know They Had

The dealer's CFO requested an independent audit after reading about PMax transparency issues. The results shocked everyone:

Finding #1: Brand Cannibalization ($28,400/month)

The forensic audit revealed that 22% of PMax spend was going to searches containing the dealer's own brand names. When someone searched "Smith Honda Miami"—clearly already intending to visit Smith Honda—PMax was intercepting that search and charging for the click.

This traffic would have arrived organically or through the dealer's dedicated Brand campaign (at much lower CPCs). Instead, PMax was claiming credit—and charging premium prices.

The Fix

• Implemented brand exclusion lists for all 8 dealership names
• Added make + dealer name combinations to exclusions
• Set up a dedicated Brand campaign with controlled budgets

Finding #2: Zero-Conversion Placements ($11,200/month)

PMax was placing ads across Google's Display Network and mobile apps. Analysis of the Placements report showed significant spend on:

• Mobile gaming apps (unlikely car buyers)
• Children's content sites
• International sites with mismatched geo-targeting

Over 90 days, these placements had accumulated $33,600 in spend with exactly zero conversions.

The Fix

• Created placement exclusion list of 847 low-quality sites/apps
• Excluded mobile app inventory entirely
• Tightened content category exclusions

Finding #3: Geographic Expansion ($7,400/month)

Despite setting location targeting to a 75-mile radius around each dealership, PMax was showing ads to users 200+ miles away. This is a known behavior: PMax interprets location targeting as "interests" rather than strict boundaries.

A user in another state who had previously searched for car information in the dealer's area could still see ads—even though they'd never realistically visit for a purchase.

The Fix

• Added explicit geographic exclusions for states outside service area
• Layered in ZIP code exclusions for known non-serviceable areas
• Set up monitoring for geographic drift

The Real ROAS

After removing brand traffic from the calculation, the dealer's true non-brand ROAS was 2.8x—below their 4x target. The "great performance" they'd been celebrating was largely an illusion created by brand cannibalization inflating the numbers.

Metric	Reported	After Audit
PMax ROAS	5.2x	2.8x (non-brand)
Monthly waste	Unknown	$47,000
Brand cannibalization	Not measured	22% of spend

Results After Implementation

Three weeks after implementing fixes:

• Monthly spend reduced from $215K to $168K (reallocated to higher-performing Search campaigns)
• Non-brand ROAS improved to 3.9x (approaching target)
• Lead volume held steady despite 22% budget reduction
• $564,000 annualized savings identified

"We thought we were sophisticated advertisers. Turns out we were paying premium prices for customers who were already looking for us by name. The audit paid for itself in the first week."

---

Note: Client details anonymized. Financial figures represent actual audit findings with identifying information removed.