Performance Max · Complete Guide 2026

Performance Max is an AI-first campaign type launched in 2021 and expanded to replace Smart Shopping and Local campaigns in 2022. A single PMax campaign serves ads on every Google channel simultaneously — Google's algorithm determines which channel, audience, and creative combination produces the best result for each individual auction.

PMax uses Google's full suite of AI capabilities: automatically generated ad copy from your assets, automatic audience expansion beyond your specified signals, real-time cross-channel attribution, and continuous creative testing. It represents Google's vision of fully automated advertising: you provide business goals and creative inputs; Google handles all targeting and optimisation decisions.

Asset groups are PMax's creative unit — each asset group contains a set of creative assets (text, images, videos) that Google combines and serves across channels. A PMax campaign can have up to 100 asset groups, each with its own creative assets and audience signal. Typically, asset groups are structured by product category, audience type, or promotional theme.

Asset group contents

Audience signals in PMax are hints — not restrictions. You provide audience segments (remarketing lists, Customer Match, In-Market audiences, Custom Segments) to indicate who your most valuable customers are. Google uses this as a starting point for targeting but will expand beyond these signals if it finds additional users likely to convert.

Audience signals do not limit targeting — PMax will show ads outside your specified signals whenever its models predict a conversion is likely. This is fundamentally different from audience targeting in Display campaigns, where audience selection genuinely restricts who sees your ads.

Recommended audience signals to provide

• Your remarketing list (all website visitors)
• Customer Match list (existing customers)
• Relevant In-Market audiences
• Custom Segments based on competitor URLs or relevant search terms

Providing high-quality, relevant audience signals speeds up the learning phase by giving Google a better starting point for conversion probability modelling.

PMax supports two bidding objectives:

• Maximise Conversion Value with Target ROAS. Recommended for e-commerce — optimises for total revenue value at a specified return on ad spend target. Requires accurate conversion values (transaction revenue) passed through conversion tracking or GA4 integration.
• Maximise Conversions with Target CPA. Used when conversion count is the primary goal and individual conversion values are not tracked (lead generation, app installs). Optimises for maximum conversions at a specified cost per acquisition.

Start without a ROAS/CPA target during the learning period — allowing the algorithm to collect data without the constraint of meeting a specific target. Add the target after the learning period (6 weeks) using your historical performance as a benchmark.

PMax provides significantly less transparency than Standard campaigns. Key reporting limitations:

• No full search term visibility. PMax shows "search categories" rather than individual search queries. You cannot see the specific queries triggering your ads at the level available in Search and Standard Shopping campaigns.
• Limited asset performance data. Asset performance shows "Low", "Good", or "Best" labels — not individual click, impression, or conversion data per asset.
• No channel breakdown by default. Performance data shows totals — attributing performance to Shopping vs Display vs YouTube vs Search is not directly available in the standard interface.
• Negative keyword limitations. Adding negative keywords to PMax is possible but limited — you can add them through account-level negative keyword lists or through Google support for specific brand/competitor terms.

The Insights tab in PMax provides some search themes, audience insights, and top-performing assets — this is the primary source of performance intelligence within the limited reporting available.

Google positions Performance Max as the recommended shopping campaign type. However, many practitioners find Standard Shopping campaigns remain valuable alongside PMax:

• Standard Shopping provides better transparency for product-level optimisation
• Standard Shopping allows more granular bid control by product group
• Standard Shopping has better negative keyword management
• For accounts with limited conversion data, Standard Shopping with Manual CPC or Target ROAS outperforms underfuelled PMax

A common structure: Standard Shopping for the majority of campaigns (full control and transparency), Performance Max for top-performing product groups where the AI can efficiently scale volume, with brand keyword exclusions in PMax to prevent cannibalisation of brand search campaigns.

• Separate asset groups by product category or audience intent. Do not mix all products into a single asset group — the creative and messaging for running shoes differs from hiking boots. Separate asset groups allow tailored creative per category.
• Provide high-quality assets in all formats. The more quality assets provided, the better the AI has to work with. Provide landscape, square, and portrait images; professional product photography; a real video; complete text assets.
• Exclude brand from PMax. Use account-level brand exclusions to prevent PMax from serving on branded searches — run a separate brand Search campaign for maximum control and efficiency on brand terms.
• Allow the learning period. Do not evaluate PMax performance during the 6-week learning phase. Do not change budgets, targets, or asset groups during learning — changes restart the learning clock.

Authentic Sources