advanced
How to Optimize Performance Max Campaigns
Optimize Performance Max campaigns by structuring asset groups around distinct product themes, providing strong audience signals with first-party data, using search themes to guide intent matching, controlling URL expansion carefully, and continuously replacing low-performing assets based on quality ratings.
Performance Max campaigns use Google’s AI to serve ads across all Google properties — Search, Display, YouTube, Gmail, Discover, and Maps — from a single campaign. Optimization requires a different approach than standard Search campaigns because you are managing inputs (assets, signals, goals) rather than individual placements. This guide covers the seven areas where your inputs have the most impact.
Key Takeaways
- Asset group structure is the primary lever for PMax performance — organize by theme, not by channel
- Audience signals are suggestions, not targeting restrictions — they guide Google’s AI but do not limit reach
- Search themes give you some control over which queries trigger your ads in Search inventory
- URL expansion is powerful but dangerous without exclusions — control where traffic lands
- Asset quality ratings are your diagnostic tool — replace “Low” assets regularly and test new variations
Introduction
Performance Max campaigns fundamentally shift the advertiser’s role. Instead of choosing keywords, placements, and bids, you provide creative assets, conversion goals, and audience signals. Google’s AI handles the rest, distributing your ads across channels based on predicted conversion likelihood.
This abstraction is powerful but comes with a trade-off: less control, less visibility. Optimization is less about granular bid adjustments and more about feeding the algorithm better inputs. The advertisers who get the best results from PMax are those who master the input layer — asset groups, audience signals, and conversion data.
This guide assumes you have an active PMax campaign with at least 2 weeks of data. If you are starting from scratch, focus on Steps 1-3 first, let the campaign learn for 2-4 weeks, then apply Steps 4-7.
Step 1: Structure Asset Groups by Theme
Asset groups are the organizational unit within PMax. Each group contains a set of creative assets (text, images, videos) and a listing group (product feed filter for e-commerce). Google combines these assets dynamically to create ads tailored to each placement and audience.
Asset group strategy:
| Approach | When to Use | Example |
|---|---|---|
| By product line | E-commerce with distinct categories | ”Running Shoes,” “Trail Shoes,” “Casual Shoes” |
| By audience segment | Service businesses | ”Enterprise Clients,” “SMB Clients,” “Startups” |
| By value proposition | Single-product businesses | ”Speed,” “Reliability,” “Price” |
| By funnel stage | Lead generation | ”Awareness,” “Consideration,” “Decision” |
Each asset group needs:
- 5-15 headlines (30 character max, mix of benefit-driven and keyword-rich)
- 5 long headlines (90 character max)
- 4-5 descriptions (90 character max)
- 5-20 images (landscape 1200x628, square 1200x1200, portrait 960x1200)
- 1-5 videos (if you have them; if not, Google auto-generates basic videos from your images)
- 1 final URL and optional display paths
Do not mix themes within an asset group. If your running shoes and hiking boots share an asset group, Google cannot optimize messaging for each product line. The AI needs thematic coherence within each group to test effectively.
Step 2: Provide Strong Audience Signals
Audience signals tell Google who your ideal customers are. Unlike audience targeting in standard campaigns, PMax signals are suggestions — Google uses them as a starting point and then expands to similar users based on performance data.
Signal types (in order of strength):
- Customer match lists — Upload your existing customer email lists. This is the strongest signal because it tells Google exactly who converts.
- Website visitors — Create remarketing audiences from your analytics data. People who visited product pages, added to cart, or started but did not complete a purchase.
- Custom segments — Define audiences based on search behavior (“people who searched for running shoes”) or app/URL interests.
- In-market audiences — Google’s pre-built audiences of people actively researching products in your category.
- Demographics — Age, gender, household income, parental status. Use only when your product has a clear demographic skew.
Best practices for signals:
- Provide signals in all asset groups, even if they are the same signals — asset groups without signals get less initial traffic.
- Use your highest-quality customer list as the primary signal. A list of repeat purchasers is better than a list of all customers.
- Layer multiple signal types. Customer match plus in-market audiences gives Google both your known converters and a broader pool of likely buyers.
- Update customer match lists at least monthly to keep the signal current.
Step 3: Use Search Themes to Guide Intent
Search themes, introduced in 2024 and expanded in 2025, give advertisers more control over which search queries trigger PMax ads. They function like keyword hints — telling Google what topics and intents your campaign should focus on.
How to set search themes:
- Open your PMax campaign
- Navigate to the asset group settings
- Under “Search themes,” add up to 25 themes per asset group
- Use specific, intent-rich phrases (not single words)
Effective search themes:
| Good | Why | Bad | Why |
|---|---|---|---|
| ”buy running shoes online” | Specific intent, commercial | ”shoes” | Too broad |
| ”best trail running shoes 2026” | Research intent, specific | ”running” | No product context |
| ”running shoes for flat feet” | Specific need | ”running shoes” | Generic, no differentiation |
Align search themes with asset groups:
Each asset group’s search themes should match its creative assets. If your “Trail Running” asset group has trail-specific images and copy, its search themes should focus on trail running queries, not general running shoe terms.
Negative keywords in PMax:
Since 2025, Google allows account-level negative keywords that apply to PMax campaigns. Use them to block:
- Brand terms (if you have a separate brand Search campaign)
- Completely irrelevant queries
- Competitor terms you do not want to appear for
Analyzing PMax search terms to identify which themes are driving conversions versus waste is more complex than standard Search campaigns. Lyra’s PMax Asset Diagnostics breaks down search term performance by asset group and theme, identifying which search themes drive conversions and which attract irrelevant traffic.
Step 4: Control URL Expansion
URL expansion allows Google to send traffic to any page on your site that it deems relevant, not just the final URL specified in your asset group. This can be powerful or problematic.
When URL expansion helps:
- E-commerce sites with many product pages — Google can match specific product pages to specific searches
- Content-rich sites where blog posts or guides might convert informational queries
When URL expansion hurts:
- Sites with pages that are not conversion-ready (about pages, blog posts without CTAs)
- Sites with pages for products or services you do not want to advertise
- Sites with login pages, support portals, or other internal pages
How to control it:
- Open your PMax campaign settings
- Navigate to “Final URL expansion”
- Choose to keep it on but add URL exclusions, or turn it off entirely
Recommended approach: Keep URL expansion on but aggressively exclude:
/blog/paths (unless blog content converts)/support/or/help/paths/login/or/account/paths- Any discontinued product pages
- Pages for services you are not currently promoting
Review your PMax placement report monthly to see which URLs Google is sending traffic to. If you find traffic going to irrelevant pages, add exclusions.
Step 5: Optimize Asset Quality Based on Ratings
Google rates each asset as “Best,” “Good,” “Low,” or “Learning.” These ratings reflect how well each asset performs relative to others in the group and across similar advertisers.
Asset rating actions:
| Rating | Meaning | Action |
|---|---|---|
| Best | Top performer, driving conversions | Keep, create variations on this theme |
| Good | Performing adequately | Keep, test improvements |
| Low | Underperforming | Replace with new creative |
| Learning | Not enough data yet | Wait 2 weeks before evaluating |
Asset optimization cycle:
- Wait until assets exit “Learning” status (typically 2 weeks with sufficient budget)
- Identify all “Low” rated assets
- Create replacement assets — do not just edit, create genuinely new variations
- Swap low performers for new assets
- Wait another 2 weeks for the new assets to learn
- Repeat
Creative guidelines by asset type:
Headlines:
- Mix benefit-driven (“Save 40% on Running Shoes”) and keyword-rich (“Premium Running Shoes”)
- Include at least one headline with a clear call to action
- Avoid generic headlines (“Welcome to Our Store”) that lack specificity
Images:
- Use high-quality product photography, not stock photos
- Include lifestyle images showing the product in use
- Test both clean/minimal backgrounds and contextual settings
- Ensure text overlays are minimal — Google often adds its own text
Descriptions:
- Lead with the value proposition
- Include social proof when possible (“Rated #1 by…”)
- End with a call to action
Manually tracking asset performance across multiple PMax campaigns and asset groups is tedious and easy to neglect. Lyra’s PMax Asset Diagnostics monitors asset quality ratings across all asset groups, alerts you when assets are rated “Low,” and provides AI-generated replacement suggestions based on patterns from your best-performing assets.
Step 6: Refine Conversion Goals and Values
PMax optimizes toward the conversion goals you set. If your goals are wrong or poorly valued, the algorithm optimizes for the wrong outcomes.
Conversion goal hygiene:
- Use only primary conversion actions that represent real business value
- Remove or set to secondary any micro-conversions (page views, scroll depth) that should not influence bidding
- If using value-based bidding (Target ROAS), ensure conversion values accurately reflect business outcomes
Value assignment for lead generation:
If you generate leads with different values, implement value rules or use offline conversion imports to give PMax accurate value signals:
| Lead Type | Conversion Action | Assigned Value | Rationale |
|---|---|---|---|
| Demo request | Primary | $500 | 10% close rate at $5,000 ACV |
| Trial signup | Primary | $200 | 4% close rate at $5,000 ACV |
| Content download | Secondary | $0 | Observation only, does not influence bidding |
New customer acquisition goals:
PMax offers a “New customer acquisition” goal that biases spend toward first-time buyers. Enable this if your business prioritizes customer acquisition over repeat purchases. Set a higher target CPA for new customers to reflect their lifetime value.
Step 7: Monitor and Iterate with Performance Insights
PMax provides less granular reporting than standard campaigns, but Google has expanded reporting significantly in 2025-2026. Use every signal available.
Reports to review weekly:
| Report | What to Check |
|---|---|
| Asset group performance | Conversions, conversion value, CPA by asset group |
| Asset details | Quality ratings, performance trends |
| Search terms | Queries triggering your PMax ads (limited visibility) |
| Placement report | Which channels (Search, YouTube, Display, etc.) drive performance |
| Audience insights | Which audience segments are converting |
Optimization cadence:
| Timeframe | Actions |
|---|---|
| Week 1-2 | Monitor only. Campaign is learning. |
| Week 3-4 | Review asset ratings, replace any “Low” assets |
| Month 2 | Adjust audience signals based on audience insights. Add search themes based on search terms. |
| Month 3+ | Test new asset groups, refine conversion goals, adjust budget allocation between PMax and Search. |
Lyra’s PMax Asset Diagnostics provides a unified view of PMax campaign health across all asset groups, with AI-powered recommendations for asset improvements, search theme adjustments, and audience signal refinements.
Practical Example
A D2C skincare brand runs one PMax campaign with three asset groups:
| Asset Group | Theme | Search Themes | Audience Signal |
|---|---|---|---|
| Anti-Aging | Anti-aging products | ”best anti-aging serum,” “retinol cream for wrinkles” | Customer match (past buyers), In-market: Skin Care |
| Acne Treatment | Acne solutions | ”acne treatment that works,” “clear skin products” | Custom segment: searched for acne solutions |
| Gift Sets | Holiday/gift bundles | ”skincare gift set,” “beauty gift for her” | In-market: Gifts, Demographics: Women 25-54 |
After 4 weeks, performance reveals:
- Anti-Aging asset group has the best CPA ($18) and highest conversion volume
- Acne Treatment has a high CPA ($45) and “Low” rated images
- Gift Sets performs well seasonally but poorly outside holiday periods
Actions taken:
- Replaced Acne Treatment images with before/after style creative and product-in-use photos
- Added stronger search themes to Acne Treatment based on search terms report
- Paused Gift Sets during off-season, reallocated budget to Anti-Aging
- Added customer match list of repeat purchasers as a signal to all asset groups
Result: Overall campaign CPA dropped 22% over the following 6 weeks.
Common Mistakes
- Not providing enough creative assets — PMax thrives on variety. Under-providing assets limits the AI’s ability to test and optimize. Max out image slots and provide at least 10 headlines.
- Setting and forgetting audience signals — Update customer match lists regularly and adjust custom segments based on performance data. Stale signals degrade over time.
- Ignoring URL expansion — Leaving URL expansion on without exclusions sends traffic to irrelevant pages, wasting budget and hurting conversion rates.
- Making too many changes at once — PMax enters a new learning period after significant changes. Batch small changes together and space major changes 2-4 weeks apart.
- Judging PMax on a short timeframe — The learning period is real. Evaluating PMax performance in the first 2 weeks leads to premature optimization or campaign cancellation.
Frequently Asked Questions
How many asset groups should a Performance Max campaign have? +
Can I see which search terms trigger my Performance Max ads? +
How long should I wait before optimizing a Performance Max campaign? +
Should I run Performance Max alongside standard Search campaigns? +
Try Lyra Free
19 Google Ads optimization tools. 14-day free trial.
Start Free TrialNo credit card charged until trial ends
Start Optimizing Your Google Ads Today
14-day free trial. All 19 tools included. No credit card charged until trial ends.
Start Free Trial