Performance Max for Service Businesses: When to Use It and How to Control It

A roofing company came to us spending 70% of their budget on Performance Max. The campaign looked great in the dashboard — high impressions, low CPC. Then we checked the search terms report.

Forty percent of their PMax traffic was branded searches. People already looking for them by name, who would have clicked the organic result for free. Another 25% was Display network placements on mobile games — the kind of accidental "fat finger" clicks that never convert. The remaining 35% was a mix of legitimate Search traffic and YouTube pre-rolls with a conversion rate barely above 1%.

Their actual ROAS on PMax? About 2.4x. Their Search campaigns, running alongside PMax with half the budget? 5.8x. They were paying more to get less — and the dashboard made it look like everything was working.

That story isn't unusual. Performance Max delivers a median ROAS of 2.57 compared to Search campaigns at 5.17. Google pushes PMax because its algorithm weights 90% for engagement and only 10% for conversions. For service businesses, PMax should never be your first campaign. Use it only after Search is profitable, and only with brand exclusions, monitoring scripts, and realistic expectations about what you're getting.

---

The Numbers Google Doesn't Headline

These are the numbers that should frame every PMax decision:

ROAS by Campaign Type (2025 Benchmarks)

Search delivers 2x the ROAS of PMax. Not a marginal difference. A fundamental gap that PMax's cross-channel reach hasn't closed.

The Conversion Rate Problem

Performance Max campaigns for service businesses average a 1.83% conversion rate. Compare that to the overall Search Ads average of 7.52% across industries, or specific verticals:

The gap exists because PMax serves ads across channels where intent varies dramatically — Search (high intent) to Display and YouTube (low intent) — and you can't control the mix.

---

The 90/10 Problem: Why PMax Underperforms for Service Businesses

PMax isn't "bad" in an absolute sense. The problem is that the algorithm optimizes for the wrong thing.

How PMax Selects Which Assets to Show

Google's Performance Max asset selection algorithm weights factors as follows:

Read that again. 90% engagement. 10% conversions.

PMax systematically chooses assets that generate clicks and engagement — activities Google charges you for — over assets that generate conversions, which is what you actually need.

How Google's "Best" Asset Rating Misleads You

PMax rates your assets as "Best," "Good," or "Low." Advertisers naturally assume "Best" means best for their business. It doesn't.

Google's "Best" asset rating criteria:

Case study — Fashion retailer:

The worst-performing asset got rated "Best." The best-performing asset got rated "Learning." Google's algorithm promoted the asset that generated the most clicks (which Google profits from) while suppressing the one that generated the most conversions (which the advertiser profits from).

Not an isolated case. GROAS.ai identified seven systematic biases in PMax's algorithm, all favoring engagement over conversions:

1. Engagement Trap — Favors high-click assets regardless of conversion quality
2. Broad Appeal Priority — Generic creative that performs "adequately" everywhere beats specialized assets that perform exceptionally on specific placements
3. New Asset Penalty — New assets receive suppressed impressions for 4-8 weeks while inferior established assets dominate
4. Platform Revenue Optimization — High-CPC assets receive preferential selection during revenue shortfall periods despite worse ROAS
5. Cross-Account Learning Contamination — Your asset selection gets influenced by unrelated businesses
6. Creative Homogenization — Algorithm favors generic stock-photo aesthetics over distinctive brand positioning
7. Attribution Manipulation — Video and dynamic assets receive inflated conversion credit while static images are under-credited

A B2B software company documented the engagement trap directly: they saw 67% traffic cost increases with 34% fewer qualified leads after switching to PMax.

---

Transparency Without Control: The Design Decision

PMax gives you channel-level reporting. You can see exactly how much budget went to Search, Display, YouTube, and Discovery. What you can't do is change the allocation.

The TechStart Case Study

Display received 45% of the budget but generated only 18% of conversions. Search received just 28% of the budget but drove 67% of conversions. The obvious fix — shift budget from Display to Search — was impossible within PMax.

TechStart abandoned PMax entirely and shifted budgets to traditional Search campaigns where they could control allocation.

The Former Google Engineer's Admission

An anonymous former Google engineer stated:

"The decision to provide transparency without control was deliberate. Google's data shows advertisers make suboptimal decisions with control, so reporting satisfies demands without enabling budget misallocation."

Translation: Google shows you the data so you feel informed, but withholds the controls so the algorithm keeps spending on channels that benefit Google's publisher network — even when those channels underperform for you.

Industry response confirms the frustration:

• 84% of PPC professionals appreciate the channel visibility
• 91% wish they could adjust channel budgets
• 83% continue using traditional campaigns for clients requiring control

---

Brand Cannibalization: PMax's Hidden Tax

This is the most expensive problem PMax creates for service businesses. And the one Google talks about least.

How Brand Cannibalization Works

As Sarah Vlietstra from MAKROZ Marketing puts it, Performance Max "prioritizes easy wins, bidding heavily on branded terms and often inflating campaign-level ROAS, making results appear stronger than they actually are."

The mechanics:

1. Someone searches your business name (e.g., "Smith Plumbing")
2. They would have found you through organic results or your dedicated brand campaign
3. PMax bids on your brand name and serves an ad
4. The person clicks the PMax ad instead of the organic result
5. PMax claims the conversion
6. Your PMax ROAS looks great — but you just paid for a click you would have gotten for free

The math is straightforward. If a plumber pays $5-$15 CPC on branded terms through PMax, and 20% of PMax conversions are actually cannibalized brand traffic, that's direct waste. An account spending $5,000/month on PMax could be losing $1,000/month paying for clicks that would have happened organically.

The Elite Retail Case Study

Channel reporting revealed:

• 73% product overlap between PMax and their Shopping campaigns
• PMax showed 23% higher CPCs than Shopping for the same products
• No controls existed to reduce this cannibalization

How to Set Up Brand Exclusions

Google now allows brand exclusions in Performance Max — a feature that was notably absent at launch. Here's the setup:

1. Navigate to your PMax campaign
2. Click Settings
3. Under Brand Exclusions, click Edit
4. Add your brand name and common variations:
   • Exact business name
   • Common misspellings
   • Abbreviations
   • Domain name without TLD
5. Save and monitor

Example for "Smith's Premium Plumbing":

Important limitation: Brand exclusions only apply to Search and Shopping placements within PMax. They do not affect Display, YouTube, or Discovery placements — which means PMax can still target your brand audience on those channels.

Run a dedicated brand Search campaign alongside PMax with exact match keywords and manual CPC bidding. This ensures you capture brand traffic at the lowest possible CPC rather than letting PMax overbid for it. As covered in our campaign structure guide, brand campaigns should receive their own budget allocation.

---

When PMax Actually Works for Service Businesses

PMax can add value in specific, well-defined scenarios. The key word is "supplement." Never "replacement."

Prerequisites Before Launching PMax

The Right Account Structure

In the accounts I manage, PMax fits in as Campaign 6, not Campaign 1:

Account Structure for Service Business
|
+-- Campaign 1: Local Service Ads (if eligible) — Pay-per-lead
+-- Campaign 2: Search - Emergency/Urgent Services
+-- Campaign 3: Search - Specific Services
+-- Campaign 4: Search - Branded Terms (Exact match, Manual CPC)
+-- Campaign 5: Remarketing (when audience > 1,000)
+-- Campaign 6: Performance Max (AFTER Search is profitable)
    Budget: 15-20% of total ad spend, NOT more

Budget allocation when using PMax:

Cap PMax at 15-20% of total spend. If it can't prove itself with 15% of budget, giving it more won't help — it'll just scale the inefficiency.

Scenarios Where PMax Adds Value

1. Reaching new audiences beyond Search

If your Search campaigns are maxed out (90%+ impression share on core keywords) and you want to expand into Display and YouTube without managing those campaigns separately, PMax can automate that expansion.

Caveat: Standalone Display campaigns have a ROAS of 0.12 and standalone Video campaigns have a ROAS of 0.52. PMax performs better than running those channels independently — but not better than just spending more on Search.

2. Remarketing at scale

PMax's cross-channel reach can be useful for remarketing to previous visitors across YouTube, Gmail, and Display. This works because remarketing audiences are already familiar with your brand — reducing the "low intent" problem.

3. Data-rich accounts with complex conversion paths

If your customers interact with your brand multiple times before converting (research on YouTube, compare on Search, convert on your website), PMax can optimize across that journey. This is more common in high-ticket services (remodeling, legal cases) than emergency services (plumbing, locksmith).

---

When to Avoid PMax Entirely

1. New Accounts (Under 6 Months)

PMax needs conversion data to learn. Zero historical data gives it nothing to optimize toward. The algorithm will spread your budget across all channels — mostly Display and YouTube where impressions are cheapest — generating engagement metrics that look active but don't convert.

What to do instead: Start with Search campaigns using exact and phrase match keywords. Build conversion data for 3-6 months. Graduate to PMax only when you have 50+ conversions/month consistently.

2. Small Budgets (Under $3,000/month total)

With limited budget, PMax can't test across channels meaningfully. Spread $3,000 across Search, Display, YouTube, Gmail, and Discovery, and each channel gets $500-$600/month. That isn't enough for any single channel to generate statistically significant results.

What to do instead: Put 100% of budget into Search campaigns targeting your highest-intent keywords. A focused $3,000/month Search campaign will dramatically outperform a scattered $3,000/month PMax campaign.

3. Businesses Without Phone Tracking

For service businesses, phone calls convert at approximately 10x the rate of website clicks. If you're not tracking phone calls as conversions, PMax has no signal to optimize toward your most valuable conversion action. It'll optimize for form fills or page views instead — metrics that don't correlate with actual revenue for most service businesses.

What to do instead: Implement call tracking first (Google forwarding numbers + third-party like CallRail), then reconsider PMax after 60-90 days of call conversion data.

4. Tight Geographic Service Areas

PMax's geographic targeting is less precise than Search. Serve a 15-mile radius, and PMax's algorithmic reach can bleed budget into adjacent areas you don't service.

For law firms paying $8.58 average CPC or dentists at $7.85, even a few mislocated clicks per day add up to hundreds of dollars monthly. Search campaigns with strict geo-targeting and "Presence" only (not "Presence or Interest") provide the control PMax cannot.

5. Industries Requiring Message Control

Legal, healthcare, and financial services face strict advertising regulations. PMax's automated creative combinations can produce ad variations that violate those regulations. Google admitted the problem by introducing brand guidelines — because automated solutions previously "created creatives that aren't on brand."

A regulatory fine for a non-compliant ad that PMax auto-generated outweighs any efficiency gain.

---

Real-World Cost of PMax Misuse: Case Studies

The numbers become concrete when you see what happens to real businesses running PMax without safeguards.

Case Study: Professional Services Firm

A professional services firm running PMax discovered through script analysis that Google was showing generic stock photos across all placements instead of their specialized infographics. The infographics converted at 340% better on Search placements — but PMax's algorithm preferred the stock photos because they generated higher click-through rates across Display and YouTube.

The cost: Higher traffic volume at lower conversion rates. More clicks billed, fewer clients acquired.

Case Study: B2B Software Company

After switching to PMax, a B2B software company documented:

• 67% increase in traffic costs
• 34% fewer qualified leads

The algorithm optimized for engagement — in B2B, that means people clicking out of curiosity, not purchase intent. Every casual Display click cost the same as a high-intent Search click, but converted at a fraction of the rate.

Case Study: Luxury Watch Retailer

PMax's cross-account learning contamination created an unusual problem: the algorithm optimized the retailer's ads based on mass-market electronics retailers' learning data. The result was ad creative emphasizing price messaging — the exact wrong approach for luxury positioning.

Case Study: End-of-Quarter Revenue Pressure

GROAS.ai documented a pattern where Google's algorithm adjusted PMax behavior during end-of-quarter periods. High-CPC assets received 280% more impressions despite 45% worse ROAS — a pattern consistent with Google needing to hit internal revenue targets.

This aligns with a broader structural concern: PMax's algorithm doesn't just optimize for engagement over conversions. During revenue-critical periods, it may actively favor the most expensive assets in your account.

Quantifying the Gap for Service Businesses

Using industry benchmarks, here's what the PMax-versus-Search gap looks like in real dollars for a service business:

Median figures — individual results vary. But the direction is clear: shifting budget from Search to PMax cuts return roughly in half for most service businesses.

The counterargument: PMax reaches audiences Search can't. True. But those audiences convert at dramatically lower rates. The question isn't whether PMax finds new people. It's whether those people are worth finding at the cost PMax charges.

---

PMax's Evolution: What Changed and What Didn't

PMax's history helps contextualize Google's claims about improvements.

What Google Has Added (Improvements)

What Google Has NOT Added (Still Missing)

The improvements are real but insufficient. More transparency (you can see more data), still no control (you can't act on it). The core algorithm weighting — 90% engagement, 10% conversion — hasn't changed.

Google's "Commerce Media" Expansion

Google expanded PMax's reach in 2025-2026 to include Waze, AI Overviews, and additional placements. Each expansion spreads your PMax budget across more channels — channels where your service business may have zero relevance.

Claire Jarrett, a Google Ads consultant, recommends: "Placement expansions only if they align with how your customers buy." For most service businesses, customers buy through Search and phone calls — not through Waze ads or AI Overview placements.

---

Audience Signals: The One PMax Lever You Should Use

Audience signals are the most valuable of PMax's limited controls. They don't restrict who sees your ads — PMax will still expand beyond your signals — but they guide the algorithm toward better starting audiences.

How to Configure Audience Signals

Custom Segments (Highest Value)

Create custom segments based on:

• Search terms people use: Add your highest-converting keywords from Search campaigns
• URLs people visit: Add competitor websites, industry directories, review sites
• Apps people use: Add apps relevant to your service (e.g., Angi, Thumbtack, HomeAdvisor for home services)

Example for a plumbing company:

In-Market Audiences

Select Google's pre-built audiences of people actively researching your service category:

• Home Services > Plumbing Services
• Home Services > HVAC Services
• Professional Services > Legal Services
• etc.

Remarketing Lists

Add your existing remarketing audiences:

• Website visitors (last 30/60/90 days)
• Customers who called but didn't convert
• Past customers (for upselling or review follow-up)

Audience Signals vs. Audience Targeting

Critical distinction: signals are suggestions, not restrictions. PMax uses your signals as a starting point but expands far beyond them:

• You cannot rely on audience signals to keep PMax focused
• PMax will spend budget on audiences outside your signals
• The algorithm weights its own learning over your signals after the initial period

Fundamentally different from Search campaign audience targeting, where you can restrict ads to specific segments. Another instance of the transparency-without-control design philosophy.

---

Monitoring PMax with Scripts

Scripts are your primary tool for understanding what PMax is actually doing. The full script setup process is covered in our scripts automation guide — here are the PMax-specific ones you need:

PMax Insights Script (Mike Rhodes)

The most comprehensive PMax visibility tool available. It extracts:

• Asset group performance data Google buries in the UI
• Channel-level spend and conversion breakdowns
• Search term data for PMax campaigns
• Audience signal performance

Setup: Weekly schedule. Outputs to Google Sheets for trending analysis.

PMax Non-Converting Search Terms Alert

Logs search terms triggered by PMax that generate clicks but zero conversions. PMax's keyword matching is notoriously broad — it will match queries that no manual Search campaign would target.

Setup: Daily schedule. Email alert when non-converting spend exceeds threshold.

What to do with the data: Use non-converting search terms to build your account-level negative keyword list. Note that PMax now supports campaign-level negative keywords — apply them aggressively.

PMax Brand Traffic Analyzer

Visualizes what percentage of your PMax traffic is brand versus non-brand. If brand traffic exceeds 30% of PMax conversions despite brand exclusions, your exclusion list is incomplete.

Setup: Weekly schedule. Track the brand/non-brand ratio over time.

Target: Brand traffic should be under 10% of PMax conversions. Anything higher indicates cannibalization.

PMax Placement Exclusion Suggestions

Surfaces low-quality placements — websites, apps, YouTube channels — where PMax is serving your ads. Helps you build an account-level placement exclusion list.

Common findings for service businesses:

• Mobile game apps (accidental "fat finger" clicks)
• Children's content websites (zero conversion potential)
• Foreign-language sites (if you serve English-speaking markets)
• Low-quality Display Network sites with near-zero engagement

Setup: Weekly schedule. Review and exclude flagged placements manually.

---

PMax vs. Search: A Direct Comparison for Service Businesses

For demand capture — reaching people actively searching for your service right now — Search outperforms PMax by a wide margin. PMax's advantage is audience expansion across channels, which only matters after you've captured all available Search demand.

---

The PMax Decision Framework

Use this framework to decide whether PMax belongs in your account:

Step 1: Is Search Already Profitable?

If your Search campaigns are not yet profitable (ROAS above your break-even point), do not add PMax. Fix Search first.

PMax cannot compensate for a broken Search strategy. It will spread the same problems across more channels at lower returns.

Step 2: Are You Capturing Available Search Demand?

Check your impression share on core non-brand keywords:

• Below 60%: Increase Search budget first — there's unfilled demand you're missing
• 60-80%: Room to grow Search; PMax is premature
• 80-90%+: Search is near capacity; PMax expansion may be justified

Step 3: Do You Have Enough Conversion Data?

PMax needs conversion volume to learn. Without it, the algorithm defaults to optimizing for engagement (clicks, views) — which is exactly what you don't want.

Step 4: Can You Monitor It?

If you cannot set up the monitoring scripts described above (or hire someone who can), do not run PMax. The black box nature of PMax means that without active monitoring, waste accumulates silently.

Decision Matrix

---

If You Deploy PMax: The Setup Checklist

Before Launch

• Search campaigns profitable for 3+ months
• 50+ monthly conversions in Search
• All conversion actions tracked (calls, forms, offline)
• Brand exclusion list configured (all variations)
• Account-level placement exclusion list built
• PMax monitoring scripts deployed and tested
• Dedicated brand Search campaign running (exact match, manual CPC)
• Budget capped at 15-20% of total ad spend

Asset Group Configuration

• Separate asset groups by service type (not one group for everything)
• 15+ images, 5+ videos, 5+ headlines, 5+ descriptions per group
• Audience signals configured (custom segments, in-market, remarketing)
• Final URLs point to service-specific landing pages (not homepage)

First 30 Days

• Review PMax channel reporting weekly
• Check brand traffic percentage (target: under 10%)
• Review non-converting search terms weekly — add negatives
• Compare PMax ROAS against Search ROAS (expect PMax to be lower)
• Do NOT increase budget during the first 30 days (learning period)

Ongoing Monthly Review

• PMax ROAS within acceptable range (define your floor before launch)
• Brand cannibalization under control (brand traffic <10%)
• No significant Search campaign performance degradation
• Placement exclusions updated based on script data
• Negative keywords expanded based on non-converting terms

Kill Criteria: When to Shut PMax Down

Define these before you launch, not after:

---

What Google Won't Tell You: The Honest Summary

Google says: "Performance Max uses AI to find customers across all Google channels."

Reality: PMax's algorithm weights 90% for engagement and 10% for conversions. It finds clicks across all Google channels — but clicks are what you pay for, and conversions are what you need.

Google says: "Performance Max simplifies campaign management."

Reality: PMax removes controls that expert advertisers need. 91% of PPC professionals wish they could adjust channel budgets within PMax. Simplification that removes control isn't simplification — it's abdication.

Google says: "Asset ratings help you understand which creative works best."

Reality: "Best" rated assets often achieve the worst ROAS. The fashion retailer case study showed a "Best" asset at 1.8x ROAS while a "Learning" asset delivered 5.1x ROAS. Google's rating system optimizes for Google's metrics, not yours.

Google says: "Channel reporting gives you visibility into performance."

Reality: A former Google engineer confirmed: "The decision to provide transparency without control was deliberate." You can see that Display gets 45% of budget for 18% of conversions — you just can't do anything about it.

Google says: "Performance Max is the future of Google Ads."

Reality: PMax median ROAS is 2.57 versus Search at 5.17. For service businesses capturing high-intent demand, Search remains the superior channel by a factor of 2x. PMax is a supplemental channel, not a replacement.

---

The Verdict

Performance Max is a tool. Not inherently good or bad. But its structural design favors engagement over conversions, benefits Google's publisher network, and reduces advertiser control. For service businesses, these trade-offs are rarely worth making as a primary strategy.

The Numbers

The Decision

• New account or small budget? Do not use PMax. Invest 100% in Search.
• Search profitable but not maxed out? Do not use PMax yet. Scale Search first.
• Search maxed, strong data, scripts ready? Test PMax at 15% budget with full monitoring.
• PMax underperforming after 60 days? Kill it and reallocate to Search. No regrets.

Where to Go From Here

1. If your Search campaigns aren't optimized yet, start with our campaign setup guide
2. Set up the monitoring scripts from our scripts automation guide — you need them whether or not you run PMax
3. Build your negative keyword strategy — it protects both Search and PMax campaigns
4. Only after steps 1-3 are solid, evaluate PMax against the decision framework above

The advertisers who make PMax work treat it as a measured experiment, not a default strategy. They deploy it with monitoring, measure it against Search, and kill it without hesitation when it underperforms.

The advertisers who waste money on PMax accept Google's framing: that PMax is the future, that the algorithm knows best, and that giving up control is the price of progress.

Your budget. Your data. Your call.