Performance Max is Google's most ambitious campaign type — an AI-driven campaign that runs across all of Google's channels: Search, Shopping, Display, YouTube, Discover, Gmail, and Maps. Since its launch, PMax has become the default choice for e-commerce and an increasingly used tool for lead generation. But PMax is not magic. Without proper setup, strong assets, and strategic audience signals, it burns through budget with poor results. This guide gives you everything you need to set up, optimize, and scale PMax correctly.
What Performance Max is and how it differs from other campaign types
Performance Max is a goal-based campaign type that uses Google's AI to optimize bidding, targeting, creative, and placements automatically across all of Google's channels. Instead of creating separate campaigns for Search, Shopping, Display, and YouTube, PMax combines everything into one campaign.
The fundamental difference from traditional campaigns is that you give up granular control in exchange for AI-driven optimization. In a Search campaign, you choose keywords, write ads, and set bids manually. In PMax, you give the AI assets (headlines, descriptions, images, videos), audience signals, and a conversion goal — and let the system find the best combinations.
Google reports that advertisers using PMax see an average of 18 percent more conversions at similar CPA compared to previous campaign types (Google Ads Performance Max Guide, 2025). But this average hides significant variation. Accounts with strong assets and good conversion data perform well above average, while accounts with weak setups perform well below. Read our complete Google Ads guide to paid advertising for a complete overview of all campaign types. For an understanding of Quality Score and how it affects ad costs, see our guide to Google Ads Quality Score.
How PMax works under the hood
PMax uses three AI systems in combination: Smart Bidding for bid optimization, audience expansion for targeting, and asset combination for creative. Together, these create a system that tests thousands of combinations and optimizes toward your conversion goal.
Smart Bidding in PMax uses machine learning based on hundreds of real-time signals: device, location, time of day, search history, browsing behavior, and more. The system sets bids per auction to maximize conversions within the budget. Audience expansion starts with the audience signals you provide but gradually expands to new segments when the AI finds patterns indicating conversion probability.
Asset combination takes the headlines, descriptions, images, and videos you upload and automatically creates ads tailored to each channel and placement. A headline that works on Search can be combined with an image for Display and a different description for YouTube. The system tests combinations and allocates more impressions to those that perform best.
Important to understand: PMax is not transparent. You get limited insight into which search terms trigger, which placements are used, and which audiences convert. This is the biggest downside and requires alternative reporting methods that we cover below.
When to use PMax vs. Standard Shopping vs. Search campaigns
The choice between PMax and traditional campaign types should be based on a framework with four factors: data maturity, control needs, product breadth, and conversion volume.
Use PMax when: you have at least 30 conversions per month (ideally 50+), you have a broad product assortment, you want to run across channels without managing separate campaigns, and you have strong creative assets (images, videos, text). PMax is especially strong for e-commerce with product feeds because it replaces Smart Shopping with better AI and broader reach.
Use Standard Shopping when: you need full control over which products appear for which search terms, you have tight margins that require keyword-level bid control, or you want to test new products in isolation before scaling. Use standard Search campaigns when: you have specific keywords with high conversion value that you want full control over, or you run brand campaigns where you want to ensure position 1.
The most common approach in 2026 is hybrid: PMax for main volume, supplemented with standard Search campaigns for brand search and high-value keywords. At LeapJet, we typically start with this hybrid model and adjust the distribution based on data after 4-6 weeks.
Setting up PMax correctly: Campaign structure and naming
A good PMax setup starts with clear campaign structure. The main rule is that each PMax campaign should have a clear conversion goal and a theme that hangs together. Do not mix different product categories with entirely different margins and conversion rates in the same campaign.
For e-commerce, we recommend one campaign per major product group, with asset groups as subcategories. Example: a campaign for "Winter Jackets" with asset groups for "Down Jackets", "Rain Jackets", and "Wool Jackets". Each asset group has its own assets and product groups.
Naming must be consistent to make reporting possible. Use a format like: [Type]_[Goal]_[Theme]_[Detail]. Example: PMax_Sales_WinterJackets_Q1-2026. This makes it easy to filter and compare in reports.
Set the conversion goal explicitly. Choose between "Maximize conversions" (most conversions within budget) and "Maximize conversion value" (highest possible conversion value). For e-commerce with varying order values, "Maximize conversion value" with target ROAS is usually best. For lead generation, "Maximize conversions" with target CPA is standard.
Asset group optimization: Headlines, descriptions, images, and video
Asset quality is the single factor you have the most control over in PMax, and it has direct impact on results. Google lets you upload up to 15 headlines, 5 descriptions, 20 images, 5 videos, and 5 logos per asset group.
For headlines: fill all 15 slots. Include variation between product-focused, benefit-focused, and action-focused headlines. Use keywords naturally. Google reports that asset groups with all 15 headlines get 10 percent better reach than those with fewer (Google Ads Best Practices, 2025).
For images: upload minimum 10, ideally 20. Use both product images, lifestyle images, and graphic content. Include images in all recommended formats: square (1:1), landscape (1.91:1), and portrait (4:5). Avoid images with lots of text — they perform poorly on Display and Discovery.
For video: upload at least 1 video, ideally 3-5. If you do not upload video, Google automatically generates videos from your images — and these are almost always worse than self-produced content. Use short videos (15-30 seconds) with clear hook and CTA. Vertical format for Shorts, horizontal for YouTube in-stream.
- Fill all asset slots: 15 headlines, 5 descriptions, 20 images, 5 videos
- Upload images in all three formats: 1:1, 1.91:1, and 4:5
- Never let Google auto-generate video — produce your own short videos
- Use varied angles: product, benefit, social proof, urgency
- Check asset performance weekly and replace "Low"-rated assets
Audience signals: Custom Segments, your data, and demographics
Audience signals are directional, not binding. They tell Google's AI where to start searching for converting users, but the AI will expand beyond these signals when it finds good results. Nevertheless, strong signals are critical for accelerating the learning phase.
Custom Segments are the strongest signal. You can create segments based on keywords people have searched for on Google, apps they have used, or websites they have visited. For an online store selling running gear: create Custom Segments with keywords like "best running shoes 2026", competitor websites, and apps like Strava.
Your Data includes customer lists, website visitors, and app users. Upload customer lists with at least 1,000 rows for best matching. Combine with Lookalike expansion to reach similar users. Demographics and interests can be added as secondary signals but should not be too narrow.
A critical mistake is not including audience signals at all. Without signals, PMax takes much longer to learn, and it spends most of the budget on Display and YouTube with low conversion rates during the learning period. Strong signals from day one deliver better results faster.
Search themes and keyword-like control in PMax
Search themes were introduced to give advertisers more control over which search terms PMax appears for. You can add up to 25 search themes per asset group, and these function as signals to the AI about which search intent you want to capture.
Search themes are not keywords in the traditional sense — you cannot set match types or negative keywords. But they provide important steering, especially for new accounts where the AI lacks historical data. Use search themes that mirror your most important keywords from standard Search campaigns.
For e-commerce: add product names, categories, and purchase-intent phrases as search themes. For lead generation: add service descriptions, problem formulations, and comparative search phrases. Avoid overly generic themes like "shoes" — be as specific as possible.
Search themes and account-level negative keywords together constitute the search control in PMax. Add negative keywords at the account level to exclude irrelevant traffic. Check the search terms insights report weekly to identify terms that should be excluded.
PMax for e-commerce: Product feed optimization and supplemental feeds
For e-commerce, the product feed is the most important element in PMax. A well-optimized feed can make the difference between profitable and unprofitable campaigns. The feed delivers product data to the Shopping component of PMax, which typically accounts for 60-80 percent of conversions.
Feed optimization starts with titles. Include brand, product type, key attribute, and variant (color, size) in the title. "Nike Air Max 90 Running Shoes Men Black" performs far better than "Air Max 90." Read our guide to Google Shopping Ads to understand how product titles and descriptions affect visibility in both organic and paid search.
Supplemental feeds let you enrich product data without changing the main feed. Use them to add custom labels that segment products by margin, bestseller status, or season. Custom Label 0: "high-margin" / "low-margin." Custom Label 1: "bestseller" / "standard." Custom Label 2: "spring" / "summer" / "fall" / "winter."
Use custom labels to create separate asset groups in PMax for high-margin categories, bestsellers, and seasonal products. This gives the AI better context and lets you allocate budget strategically to the most profitable products.
PMax for lead generation: Lead form assets and offline conversions
PMax for lead generation requires a different approach than for e-commerce. Without a product feed, you must compensate with stronger assets, clearer audience signals, and — critically — offline conversion imports to avoid poor lead quality.
Lead Form Assets let users submit contact information directly in the ad without visiting the website. They yield higher volume but often lower quality than website forms. Use qualifying questions in the form to filter out unqualified leads. Test both approaches (lead form vs. website) and compare lead quality, not just volume.
Offline Conversion Import is the most important step for B2B and service businesses. By importing data about which leads actually became customers back into Google Ads, the AI learns to optimize toward quality leads instead of just form submissions. Set up automatic import from CRM via Google Ads API or weekly manual CSV upload.
Without offline conversion data, PMax optimizes toward the cheapest lead type, which is almost always the least valuable. We typically see 2-3 times better lead quality after offline conversion import has been in place for 4-6 weeks. For conversion tracking setup, see our guide to conversion tracking. Also ensure landing pages convert effectively — read our guide to PPC landing page optimization. For negative keyword strategy, see our guide to negative keywords. Use our ad budget calculator to estimate expected ROAS.
Reporting challenges and solutions
PMax is notoriously poor at reporting. You get no search terms report at the campaign level, the placement report shows only channel-level data, and audience insight is limited. This is the biggest frustration for experienced PPC managers — and it requires creative solutions.
The Insights tab in PMax campaigns shows search themes, audience segments, and asset performance. Use it weekly to identify patterns. The placement report (via "Where ads showed") shows the distribution between Search, Shopping, Display, YouTube, and other channels. If Display and YouTube dominate, it may indicate the AI is not finding enough converting traffic in Search.
Script-based reporting is the best solution for deeper insight. Google Ads scripts can retrieve search term data, placement details, and other metrics not available in the standard report. Community-developed scripts exist that automate this.
Compare PMax results with your standard Search and Shopping campaigns to identify incremental value. If PMax only "steals" conversions from existing campaigns without delivering new ones, it is not providing real value. Use holdout tests (on/off in different regions) to measure incrementality.
PMax optimization cycle: Weekly, monthly, and quarterly tasks
PMax requires a different optimization rhythm than traditional campaigns because the AI needs stability to learn. Too frequent changes send the campaign back into learning mode and destroy performance.
Weekly: Check asset performance and replace "Low"-rated assets. Review search term insights and add negative keywords at account level. Verify that budget distribution between channels is acceptable. Check conversion volume — is it above the minimum 30 per month?
Monthly: Evaluate ROAS/CPA against target. Adjust target ROAS/CPA by maximum 15-20 percent at a time. Update audience signals with fresh data. Add new creative assets based on learning data. Review placement report for unwanted placements.
Quarterly: Evaluate campaign structure — should asset groups be merged or split? Assess whether PMax is delivering incremental value compared to standard campaigns. Update product feed strategy and custom labels. Make a holistic assessment of whether the PMax share of budget should increase or decrease.
- Weekly: asset performance, negative keywords, budget distribution
- Monthly: ROAS/CPA evaluation, audience signal updates, new assets
- Quarterly: structure evaluation, incrementality testing, feed optimization
- Never change target ROAS/CPA by more than 15-20% at a time
- Avoid making multiple major changes simultaneously — isolate variables
Common PMax mistakes and how to avoid them
PMax mistakes are costly because they take a long time to detect in a campaign type with limited transparency. Here are the most common mistakes we see at LeapJet when we take over PMax accounts.
Weak assets are mistake number one. Many advertisers fill the minimum requirements and expect the AI to do the rest. But Google's own data shows that asset groups with "Excellent" asset strength get 6 times more conversions than groups with "Poor" strength (Google Ads Performance Max Best Practices, 2025). Fill all slots, vary angles, and replace poor assets continuously.
No audience signals is mistake number two. Without signals, PMax starts by spending budget on Display and YouTube because those are the cheapest channels. You get many impressions but few conversions during the learning period. Strong signals steer the AI toward converting users faster.
A single asset group for everything is mistake number three. When all products or services are mixed into one asset group, the AI cannot show relevant creative for specific search intents. Segment asset groups by theme, product category, or conversion type.
- Fill all asset slots and aim for "Excellent" asset strength
- Include audience signals from day one
- Segment asset groups by theme — do not mix everything into one
- Upload your own videos to avoid auto-generated ones
- Use search themes to steer search intent
- Add negative keywords at account level
- Wait at least 2 weeks before evaluating changes
Sources
This article draws on the following sources:
- Google — Ads Performance Max Guide: Campaign Setup and Best Practices (2025)
- Google — Ads Best Practices: Asset Group Optimization Data (2025)
- Google — Performance Max vs. Standard Shopping Conversion Lift Study (2025)
- Google — Ads Performance Max Best Practices: Asset Strength Impact (2025)
- Optmyzr — Performance Max Benchmark Report: Cross-Industry Performance Data (2025)
- Search Engine Land — PMax Reporting Scripts and Transparency Tools (2025)
