PMax Not Performing? The Complete Diagnostic Framework for Performance Max Campaigns

14 min read · Google Ads · Last updated July 2026

Quick answer: PMax underperformance has six root causes: insufficient budget (under $50/day), weak asset quality, missing or wrong audience signals, conversion tracking errors, brand cannibalisation, and insufficient learning data. Diagnose in that order before making any campaign changes.

Introduction

Performance Max is simultaneously the most powerful and the most frustrating campaign type Google has ever released. When it works, it’s extraordinary — finding customers across Search, Display, YouTube, Gmail, Discover, and Maps with a single campaign. When it doesn’t, it behaves like a black box that burns budget on irrelevant inventory with zero transparency.

The problem: most advertisers diagnose PMax incorrectly. They change bid strategies after three days. They pause the campaign after two weeks. They add more keywords to a campaign type that doesn’t use keywords the way you think it does. They add audience signals based on job titles and then wonder why a B2C campaign isn’t converting.

Here is the real diagnostic process — the same one the Ignited Nepal team uses when inheriting a broken PMax campaign.

In this guide you will learn:

• The six-stage diagnostic framework for PMax campaigns
• How to identify and fix asset quality problems
• What audience signals actually do (and why most are configured wrong)
• The budget minimums that must be met for PMax to learn
• How to detect and stop brand cannibalisation

Table of Contents

1. Understanding How PMax Actually Works
2. Stage 1: Budget Minimum Diagnosis
3. Stage 2: Conversion Tracking Audit
4. Stage 3: Asset Quality Diagnosis
5. Stage 4: Audience Signal Diagnosis
6. Stage 5: Brand Cannibalisation Check
7. Stage 6: Learning Period and Data Volume Check
8. Advanced Fixes: Exclusions, Negative Keywords, and Asset Groups
9. PMax Performance Benchmarks by Industry
10. FAQ
11. Conclusion

1. Understanding How PMax Actually Works

Before diagnosing what’s wrong, you need a mental model of what PMax is actually doing. Most problems stem from misunderstanding the mechanism.

PMax is a Goal-Based Campaign. It optimises toward the conversion actions you define, using Google’s full inventory and audience graph to find people who are likely to complete those actions. It does not use traditional keyword targeting — instead, it uses audience signals as a starting point and then expands beyond them.

PMax has two discovery modes:
1. Audience signal mode — targets the audience signals you provide as a starting point
2. Open targeting mode — discovers new audiences beyond your signals, using your creative assets and conversion data

Most PMax problems happen in mode 1 (wrong signals) or in the transition to mode 2 (insufficient conversion data to expand intelligently).

What PMax needs to work:
– Minimum $50/day budget (absolute floor; $100+/day for meaningful learning)
– At least 30 conversions per month in the campaign (ideally 50+)
– High-quality creative assets across all formats (images, headlines, logos, videos)
– Audience signals that represent your best customers (not all potential customers)
– Correct conversion tracking with accurate values or goal-based CPA targets

Fail any one of these, and PMax will underperform — not because the campaign type is broken, but because you haven’t given it what it needs.

2. Stage 1: Budget Minimum Diagnosis

The single most common reason PMax underperforms is insufficient budget. This is not a metaphor for “spend more money.” It is a technical requirement of how Google’s Smart Bidding algorithm learns.

The learning phase math:

Smart Bidding needs 30–50 conversions to exit the learning phase for a single asset group. PMax distributes budget across multiple channels simultaneously. If your daily budget is $20 and your CPA target is $80, you’re generating 7–8 conversions per month — far below the minimum threshold.

The algorithm never exits learning. It keeps rotating creatives, testing audiences, and testing placements — never stabilising on what works.

Budget diagnostic questions:

1. Is your daily budget at least $50? ($100+ strongly recommended)
2. Does your budget * 30 days generate at least 30 expected conversions? (Budget ÷ target CPA × 30 ≥ 30)
3. Have you been running at this budget for at least 6 weeks?

Fix: If budget is the constraint, you have two options:
– Increase budget to the minimum viable level
– Raise your target CPA to reduce the required conversions per dollar spent (lower the bar for the algorithm)

Do not reduce budget and tighten CPA targets simultaneously. That’s the fastest way to kill a PMax campaign permanently.

3. Stage 2: Conversion Tracking Audit

The second most common PMax failure is bad conversion tracking — specifically, tracking the wrong thing or tracking inflated conversion numbers.

The most dangerous PMax conversion tracking mistakes:

Mistake 1: Counting page views as conversions. If you have “Thank you page visit” as a conversion action and users are landing on that page from sources other than ad-submitted forms, PMax is optimising toward traffic — not revenue.

Mistake 2: Tracking micro-conversions alongside macro-conversions with equal weight. If “Newsletter sign-up” has the same conversion value as “Demo booked,” PMax will optimise toward newsletter sign-ups (much easier to obtain) and ignore demo bookings.

Mistake 3: Duplicate conversion tracking. GA4 import + Google Ads tag both firing for the same event. Your conversion volume looks great. Your actual revenue does not match.

Conversion tracking audit steps:

1. Go to Tools → Conversions → Click each conversion action
2. Check “Count” setting — set macro-conversions (purchases, leads) to “One” not “Every”
3. Check “Primary vs Secondary” — only macro-conversions should be set as Primary
4. Cross-reference conversion volume in Google Ads against actual leads in your CRM for the same period
5. If there’s a >20% discrepancy, you have a tracking issue

Fix for B2B lead gen: Use offline conversion imports from your CRM to feed qualified leads back into Google Ads with values. PMax will then optimise toward the leads that actually become customers, not just form fills.

4. Stage 3: Asset Quality Diagnosis

PMax uses an Asset Strength score (Poor / Good / Excellent) for each asset group. This score directly affects delivery — low-quality asset groups receive less traffic.

The Asset Quality minimum requirements:

The most common asset quality failure: relying on auto-generated video.

When you don’t provide a video, Google auto-generates one from your images and headlines. These are visually poor and perform badly on YouTube inventory. If PMax is allocating significant budget to YouTube (check Insights → Asset Group) and you have auto-generated video, that’s a significant waste.

Diagnosing asset quality:

1. Go to your PMax campaign → Asset Groups → Click “View Asset Details”
2. Review the Asset Strength indicator
3. Look at “Asset Performance” ratings: “Best,” “Good,” “Low,” “Learning”
4. Any asset rated “Low” is actively hurting delivery — replace or remove it

Common weak assets:
– Generic stock photos (low distinctiveness signal)
– Headlines that are your company name only
– Descriptions that describe features, not outcomes
– Logos with white backgrounds on images (looks bad on Display placements)

Fix: Audit every headline for specificity. “Award-Winning Marketing Agency” is weak. “93% of Clients Renew After Year One” is strong. Replace every “Low” performing asset within 48 hours of review.

5. Stage 4: Audience Signal Diagnosis

Audience signals are the most misunderstood element of PMax. Here is the key clarification that changes everything:

Audience signals are NOT targeting — they are starting points.

When you add an audience signal, you are telling Google: “Start your learning by focusing on people like this. Then expand from there.” PMax will expand beyond your signals based on conversion data. The signals just accelerate the learning phase.

The most common audience signal mistakes:

Mistake 1: Adding too-broad signals. If you sell accounting software to small businesses and your audience signal is “Business Professionals” (200 million+ people), you’ve given the algorithm almost no direction. It will spend weeks testing random placements before finding signal.

Mistake 2: Using demographic signals instead of intent signals. Age, gender, and income are weak signals for most B2B and service businesses. Behavioural signals (website visitors, similar audiences, customer match lists) are 4–6x more effective as starting points.

Mistake 3: No customer match list. The single most powerful audience signal is a customer match list — your actual customer email addresses. Google matches these to real Google accounts and builds a “lookalike” profile of your best customers. If you have 500+ customer emails, you should be using them.

The recommended audience signal stack (in order of priority):

1. Customer match list — Upload your customer list (CSV with email, phone, first name, last name)
2. Website visitors (all pages, 90 days) — People who already know you
3. Converters — Past leads or buyers (from GA4 import)
4. Your top-performing remarketing segment — High-intent behavioural segment
5. Custom intent audience — Built from keywords your customers search (not campaign keywords — audience keywords)

What NOT to use as primary signals:
– Interest categories from Google’s predefined lists (too broad)
– Demographic-only signals (gender + age)
– Topics (too broad)
– YouTube viewers (unless video is your primary acquisition channel)

6. Stage 5: Brand Cannibalisation Check

This is the most overlooked PMax problem, and it can silently inflate your ROAS while your actual new customer acquisition stalls.

What brand cannibalisation looks like:

PMax will take credit for converting users who were going to find you anyway through branded search. If someone searches “[YourBrand] login” and clicks a PMax ad, that’s a conversion that your brand campaign would have captured at $0.50 CPC — and PMax just charged you $8 CPC.

When brand cannibalisation is severe, your PMax ROAS looks excellent (3–5x+) but your non-brand ROAS is terrible. New customer acquisition has plateaued. Budget is being consumed by branded traffic.

How to detect cannibalisation:

1. Go to Insights → Search Terms → Click “Campaign” view
2. Look for branded queries in your PMax search term report
3. Check percentage of PMax conversions coming from brand terms
4. Compare your brand campaign impression share — if it has dropped since launching PMax, cannibalisation is occurring

Fix — Brand Exclusions:

PMax now supports brand exclusion lists (as of 2023). Go to Campaign Settings → Brand Exclusions → Add your brand name and all common misspellings. This prevents PMax from serving on branded queries.

This is not optional for any account that also runs a brand campaign. Without brand exclusions, you are paying a premium for traffic you could have won cheaply.

7. Stage 6: Learning Period and Data Volume Check

PMax campaigns need time and data volume to exit the learning phase. Most advertisers intervene too early.

The PMax learning timeline:

• Days 1–7: Initial exploration across all inventory types
• Days 7–14: Signal processing — beginning to identify high-converting audience patterns
• Days 14–30: Optimisation begins — budget allocation shifts toward better-performing inventory
• Days 30–60: Meaningful performance emerges — ROAS/CPA stabilises
• Days 60+: Full optimisation mode — campaign finds its equilibrium

The cardinal rule: Do not make structural changes (bid strategy change, budget cut >20%, asset group additions) in the first 14 days. Every structural change resets the learning period.

Common learning interruptions:
– Changing bid strategy from Maximise Conversions to Target ROAS before 30 conversions
– Adding new asset groups before existing ones have enough data
– Pausing and resuming the campaign
– Major budget fluctuations (>20% change per week)

Patience test: If you’re on day 25 with 15 conversions, you have not failed — you’re below the data threshold. Either increase budget to accelerate learning or extend your evaluation window to 60 days.

8. Advanced Fixes: Exclusions, Negative Keywords, and Asset Groups

Placement Exclusions

PMax allows you to exclude specific URLs, mobile apps, and YouTube channels. Use this to stop showing on:
– Mobile app categories (especially gaming apps, which have terrible conversion rates)
– Parked domain sites
– YouTube channels with irrelevant audiences
– Specific competitor domains (controversial — some argue keeping them improves your data)

To add placement exclusions: Go to Campaign → Placements → Exclusions → Add URLs or app categories.

Negative Keywords (Limited but Available)

PMax does not have a keyword targeting interface, but you can apply negative keyword lists at the campaign level. To do this:
1. Create a Negative Keyword List in Shared Library
2. Apply it to your PMax campaign

This is particularly important for excluding irrelevant search intent that consistently doesn’t convert. Common negative keywords for PMax:
– Your own brand terms (if using brand exclusions tool instead)
– Competitor names (if you’re not running conquest campaigns)
– Free / DIY / tutorial / how to (for paid service businesses)
– Jobs / careers / salary (to avoid job seekers triggering service ads)

Multiple Asset Groups

Use separate asset groups for:
– Different product or service lines (different messaging, different images)
– Different funnel stages (prospecting vs remarketing)
– Different geographies (if messaging differs significantly)

Do not use separate asset groups just to test headlines — that’s what the headline variation feature is for. Asset groups are for fundamentally different creative and audience approaches.

9. PMax Performance Benchmarks by Industry

Use these benchmarks to calibrate expectations. If you’re significantly below these ranges after 60 days, you have a structural problem.

Key takeaway: PMax is not broken — it is undertrained. Every PMax problem has a root cause that predates the campaign launch: wrong budget, wrong signals, wrong assets, or wrong conversion setup. Fix the foundation, give the algorithm time, and measure at 60 days.

FAQ

Should I run PMax and Search campaigns at the same time?
Yes, and you should. PMax and Search campaigns complement each other. Search campaigns give you explicit keyword control and brand term ownership. PMax discovers new audiences and inventory. Run them together, use brand exclusions in PMax, and let each do its job.

Why does PMax show a great ROAS but pipeline hasn’t grown?
Brand cannibalisation. Check your search term insights in the PMax campaign — if you see your brand name appearing frequently, that’s the cause. Add brand exclusions immediately.

How often should I make changes to a PMax campaign?
Maximum one meaningful change per week after the initial 14-day no-touch period. Each change triggers a mini-learning phase. Constant changes keep the campaign in perpetual learning and prevent it from ever optimising.

Can I see which placements PMax is running on?
Partially. Go to Insights → Placement Performance → you can see category-level placement data. For detailed URL-level data, check your Placement report. Google does not show full granular placement data for PMax.

What is the minimum number of asset groups I need?
One. More asset groups are useful when you have fundamentally different product lines or messaging, not for minor variations. Start with one well-built asset group before adding more.

Should I use Target ROAS or Target CPA for PMax?
Start with Maximise Conversions (no target) for the first 2–4 weeks to gather data. Switch to Target CPA when you have 30+ conversions. Switch to Target ROAS only when you have 50+ conversions and are tracking conversion values.

Conclusion

PMax is not magic, and it’s not a black box. It is a sophisticated algorithm that requires specific inputs to produce useful outputs. Budget, conversion data, asset quality, audience signals, and brand exclusions are the five control levers. When any one of them is wrong, the campaign underperforms — and the fix is always in the foundation, not in the bidding settings.

Use the diagnostic tool above to identify your specific issue. Fix the highest-priority issue first. Wait 2 weeks. Then re-evaluate.

Struggling with a PMax campaign that’s burning budget without producing qualified pipeline? The Ignited Nepal growth engineering team audits and rebuilds PMax campaigns across e-commerce, SaaS, and professional services — with a clear diagnostic report before any changes are made.

Book a free PMax audit at ignitednepal.com

Written by the Ignited Nepal team. ignitednepal.com