Dallas Auto Dealer: Performance Max Restructure + Inventory Feed Repair

Client: Lone Star Pre-Owned · Market: Dallas–Fort Worth Metro · Timeframe: 45 Days

The Challenge: PMax Running Blind

Lone Star Pre-Owned is an independent dealership in the Dallas metro with 85–110 vehicles in inventory at any given time. When Jordan, the General Manager, contacted us, they were running a single Performance Max campaign with a $22,000 monthly budget and getting a 1.9x ROAS — barely above break-even after overhead.

Their Google rep had assured them PMax was "working as designed." Our audit told a different story. We identified three compounding problems that were gutting the campaign's efficiency simultaneously.

Problem 1 — Broken Inventory Feed: The vehicle feed had 214 active errors including 47 vehicles listed at $0 price, 38 vehicles with missing condition attributes, and 11 vehicles that had sold months prior still receiving ad impressions and clicks. Google was serving ads for a 2021 F-150 that had been sold in January.

Problem 2 — No Asset Group Segmentation: The entire inventory — sedans, trucks, SUVs, luxury, sub-$15K economy vehicles — was running through a single asset group with generic creative. A first-time buyer searching for an affordable commuter car was seeing the same ad as someone shopping for a $45,000 pickup. The algorithm had no signal to serve the right message to the right person.

Problem 3 — Brand and Non-Intent Traffic Mixed: Branded search terms (people searching the dealership's name directly) were being credited as PMax conversions, inflating reported ROAS. When we stripped brand traffic from the conversion attribution, the true ROAS for conquest traffic was under 1.4x — worse than break-even.

The Fix: Three Parallel Tracks

Track 1: Feed Repair and Automation

• Audited all 214 errors and categorized by type: missing attributes, sold inventory, price errors, image quality failures
• Worked with the dealership's DMS (dealer management system) to set up an automated nightly feed refresh via Google Merchant Center
• Built a feed monitoring script that flags any new errors within 24 hours
• Implemented supplemental feed rules to auto-populate missing attributes from VIN data
• Reduced feed errors from 214 to 3 (minor image size warnings) within 10 days

Track 2: Asset Group Segmentation

• Rebuilt the single PMax campaign into 4 asset groups aligned with buyer intent: Economy (<$20K), Family SUV ($20K–$35K), Truck + Work Vehicle, and Luxury/Premium ($40K+)
• Wrote dedicated headlines, descriptions, and calls-to-action for each segment
• Sourced segment-appropriate imagery: real inventory photos, lifestyle shots matched to buyer profile
• Set audience signals for each group based on in-market segments, customer list lookalikes, and prior site visitors by page category

Track 3: Brand Campaign Separation

• Built a dedicated branded search campaign at low CPC to capture high-intent direct searches efficiently
• Excluded branded terms from PMax using negative keyword lists (requires a workaround via campaign-level negatives)
• This alone revealed that real conquest ROAS was being subsidized — and gave us a clean baseline to improve against

Results: Before vs. After

Metric	Before BZM	After 45 Days
Overall ROAS	1.9x	5.3x
Conquest ROAS (ex-brand)	~1.4x	4.1x
Feed errors	214	3 (minor)
Asset groups	1 (all inventory)	4 (segmented by buyer type)
Sold vehicles still advertising	11	0 (automated nightly refresh)
Monthly ad spend	$22,000	$22,000 (same budget, better results)
Form inquiries/month	~190	441

What the GM Said

"Our Google rep told us Performance Max was working great. BZM showed us it was sending traffic to vehicles we didn't even have in stock anymore — the inventory feed had over 200 errors and we had no idea. Within six weeks we were getting real buyer traffic and our ROAS nearly tripled. My sales team is busier and we're not spending a dollar more on ads. I wish we'd done this two years ago." — Jordan K., General Manager, Lone Star Pre-Owned · Dallas, TX

The Lesson for Auto Dealers on Performance Max

Performance Max is a powerful tool in the right hands and a money furnace in the wrong ones. The algorithm is only as good as the data you feed it. A broken inventory feed, generic asset groups, and brand traffic contamination will make PMax look like it's working while your actual conquest performance rots underneath the surface. The fix requires expertise in feed management, audience strategy, and campaign architecture — not just "turning it on" and letting Google optimize.