What Performance Max Actually Is
Performance Max (PMax) is Google's AI-driven campaign type that runs ads across every Google channel from a single campaign: Search, Display, YouTube, Gmail, Discover, and Maps.
You provide the creative assets (images, videos, headlines, descriptions), set a goal, give Google an audience signal, and the AI figures out where to show your ads and to whom.
On paper, it sounds like a dream. In practice, it's a powerful tool that requires careful setup — or it becomes an expensive way to burn budget on irrelevant placements.
The uncomfortable truth: Most advertisers running PMax are letting Google spend their money however it wants because they haven't configured the campaign properly. Google's AI is good, but it optimises for its definition of success, which may not match yours.
How PMax Differs From Traditional Campaigns
| Feature | Traditional Campaigns | Performance Max | |---------|----------------------|----------------| | Channel control | You choose (Search, Display, etc.) | Google chooses all | | Keyword targeting | You select keywords | Google matches intent | | Bidding | Manual or automated per campaign | Fully automated | | Placement visibility | Full transparency | Limited reporting | | Creative | You control per ad | AI mixes and matches assets | | Audience targeting | You define audiences | You provide "signals" |
The trade-off is clear: You get Google's full AI power but give up granular control and transparency.
When to Use PMax (and When Not To)
Good Fit
- E-commerce with a product feed — PMax replaced Smart Shopping and works well for product-based businesses
- Lead generation with clear conversion tracking — the AI needs reliable conversion data to optimise
- Businesses with 30+ conversions per month — the AI needs data volume to learn effectively
- When you've maxed out standard campaigns — PMax can find incremental reach
- Multi-channel advertising goals — you want presence across Search, YouTube, Display, etc.
Poor Fit
- Fewer than 15 conversions per month — not enough data for the AI to learn
- Unclear conversion tracking — garbage in, garbage out
- Brand-new accounts with no history — the AI has nothing to learn from
- When you need tight control over placements — PMax won't give you that
- Very small budgets — the AI needs room to test across channels
Anatomy of a PMax Campaign
Asset Groups
Asset groups are the building blocks of PMax. Each asset group contains:
Text assets:
- Up to 5 headlines (30 characters each)
- Up to 5 long headlines (90 characters each)
- Up to 5 descriptions (90 characters each)
- Business name
- Display URL path
- Call to action
Image assets:
- Landscape images (1200x628)
- Square images (1200x1200)
- Portrait images (960x1200)
- Logo (1200x1200 and 1200x300)
Video assets:
- At least one video (YouTube)
- Landscape, square, and vertical formats
- If you don't provide video, Google will auto-generate one (usually terrible — always provide your own)
The golden rule: Provide as many high-quality assets as possible. The AI can only mix and match what you give it. Weak assets = weak ads.
Audience Signals
Audience signals tell Google's AI who your ideal customer looks like. They're not hard targeting — they're suggestions.
Custom segments:
- People who searched for specific terms
- People who visited specific websites
- People who use specific apps
Your data:
- Customer lists (email uploads)
- Website visitors (remarketing)
- App users
Interests and demographics:
- In-market audiences
- Affinity audiences
- Life events
- Demographics
Strong audience signals are the single most important PMax optimization lever. Without them, Google's AI starts from scratch. With them, it has a head start on finding the right people.
Conversion Goals
PMax optimises toward whatever conversion goal you set. Get this wrong and nothing else matters.
Best practices:
- Use primary conversions only for your most valuable actions (purchases, qualified leads)
- Set secondary conversions for softer goals (page views, add to cart) — these inform the AI but don't drive bidding
- Assign conversion values if possible (even estimated ones)
- Use "Maximise conversion value" instead of "Maximise conversions" when values are available
Structuring Your PMax Campaigns
The Hybrid Approach (Recommended)
Don't rely solely on PMax. Run it alongside standard campaigns.
Structure:
- Standard Search campaigns for your highest-value, most controlled keywords
- PMax for broader reach and discovery
- Standard Shopping (if e-commerce) alongside PMax for comparison
Why hybrid works:
- You maintain control over your core keywords
- PMax fills in the gaps and finds new audiences
- You can compare PMax performance against standard campaigns
- If PMax underperforms on a channel, your standard campaigns still cover it
Asset Group Strategy
Option 1: Segment by product/service category
- One asset group per product line or service
- Tailored messaging for each
- Different audience signals per group
Option 2: Segment by audience
- One asset group for prospects, one for remarketing
- Different messaging for cold vs. warm audiences
- Different conversion goals
Option 3: Segment by funnel stage
- Awareness asset groups (broad messaging)
- Consideration asset groups (value-focused)
- Decision asset groups (offer-driven)
Optimising PMax Campaigns
Week 1-2: Learning Phase
Don't touch anything. The AI needs time to learn.
- Budget will fluctuate
- Performance will be inconsistent
- Resist the urge to make changes
- Monitor but don't optimise yet
Week 3-4: Initial Optimisation
Check asset performance:
- Review the asset report (which headlines, images, videos perform best)
- Replace "Low" performing assets with new variations
- Keep "Best" and "Good" performing assets
Review audience insights:
- Which audience segments are converting?
- Are there surprises (audiences you didn't expect)?
- Refine audience signals based on what's working
Assess channel distribution:
- Where is budget being spent? (Insights tab)
- Is Display eating too much budget with low-value traffic?
- Are Search impressions cannibalising your standard Search campaigns?
Monthly Ongoing
- Refresh creative assets (new images, headlines, videos)
- Update audience signals based on latest customer data
- Review search term insights (limited but available)
- Compare PMax against standard campaign performance
- Adjust budget allocation between PMax and standard campaigns
- Test new asset groups for new products/services
The Reporting Problem (and How to Work Around It)
PMax's biggest criticism is limited transparency. You can't see:
- Specific search terms (only "search term insights" categories)
- Exact placement URLs
- Individual keyword performance
- Per-channel ROAS in detail
What You Can See
Asset performance: Which individual assets are rated Best, Good, Low, or "Learning"
Audience insights: Which segments are engaging and converting
Conversion data: Total conversions, conversion value, CPA, ROAS
Search term insights: Themed categories of search terms (not individual terms)
Placement reports: Some visibility into where ads appeared
Workarounds
Use UTM parameters and GA4:
- Track PMax traffic in Google Analytics separately
- Analyse landing page performance from PMax vs. other campaigns
- Monitor user behaviour (bounce rate, time on site, pages per session)
Run brand exclusions:
- Exclude your brand terms from PMax so it doesn't cannibalise branded search
- Forces PMax to find genuinely new customers
- Available as an account-level or campaign-level setting
Use scripts:
- Third-party scripts can extract more granular PMax data
- Mike Rhodes' PMax script is widely used in the industry
- Provides channel-level spend breakdowns
PMax for E-commerce vs. Lead Gen
E-commerce
PMax shines for e-commerce because of the product feed integration.
Setup:
- Connect Google Merchant Center
- Use product-level ROAS targets
- Segment asset groups by product category
- Provide high-quality product imagery
Key metric: ROAS (target 4:1+)
Lead Generation
PMax for lead gen requires more careful setup because lead quality is harder to measure than purchases.
Challenges:
- The AI may optimise for form fills, not qualified leads
- Lead quality can vary wildly
- No immediate revenue data to feed back to the AI
Solutions:
- Import offline conversions (tell Google which leads actually closed)
- Use enhanced conversions for leads
- Set up value-based bidding with estimated lead values
- Monitor lead quality weekly, not just volume
Common PMax Mistakes
- Not providing video assets — Google auto-generates awful videos. Always upload your own
- Weak audience signals — the AI needs direction. Upload customer lists and define custom segments
- Not excluding brand terms — PMax will happily spend on branded search that would convert anyway
- Only one asset group — limits the AI's ability to match messaging to audiences
- Setting it and forgetting it — PMax needs ongoing creative refreshes and signal updates
- Using "Maximise clicks" — always use conversion-based bidding
- Launching with too few conversions — need 30+ per month for reliable optimisation
- Not running standard campaigns alongside — PMax alone leaves you with no control and no comparison point
Getting Started
- Set up proper conversion tracking (this is non-negotiable)
- Upload your customer list for audience signals
- Create 2-3 asset groups with strong creative
- Upload at least one custom video per asset group
- Exclude brand terms
- Set a realistic budget (enough for the AI to learn)
- Use "Maximise conversion value" with a target ROAS if you have value data
- Run alongside existing standard campaigns
- Don't touch it for 2 weeks
- Optimise monthly based on data
PMax is not a magic button. It's a sophisticated tool that rewards thoughtful setup and ongoing attention. The advertisers seeing 5-10x ROAS aren't lucky — they're disciplined about structure, signals, and creative quality.