AI Creative

How to Use AI for Google Display and PMax Ad Creative at Scale

Mukund Srivathsan, CTO8 min read
Google Display and PMax advertising

The Google Display Challenge

Google Display Network reaches 90% of internet users across 2 million sites. It's the largest advertising platform in the world. But Google Display creative has a reputation problem: the ads are often generic, poorly designed, and easy to ignore.

The reason is technical constraint. Google Display ads exist in dozens of formats: 300x250 squares, 728x90 leaderboards, 300x600 half-pages, plus infinite custom sizes. Building separate creative for each format manually is economically infeasible. Most brands create one ad and let it stretch and resize to fit each format, resulting in distorted, low-quality creative.

Google's answer was responsive display ads, which auto-resize across formats. But the challenge remained: humans need to provide all the raw assets, and getting responsive creative to look good across 10+ formats requires understanding how text, images, and branding elements scale and reflow.

Performance Max (PMax) compounds this challenge. PMax is Google's end-to-end campaign format that runs across Search, Display, Gmail, Maps, and YouTube. It requires dozens of assets in different ratios, dimensions, and aspect ratios. Providing all required assets manually is a nightmare. Most brands under-deliver assets, and PMax performs below its potential as a result.

PMax is Google's most powerful format, but most brands starve it of assets. This is where AI changes the game.

PMax's Evolution and AI Creative Requirements

Performance Max has evolved from a niche product to Google's primary focus. Google is directing more budget toward PMax and increasingly de-prioritizing traditional Search and Display. Brands that master PMax will have a competitive advantage; brands that don't will lose scale.

PMax requires:

  • At least 20 assets (images, headlines, descriptions) to perform optimally
  • Ideally 30-50 assets to unlock PMax's full machine learning potential
  • Assets in multiple aspect ratios: 1:1, 4:3, 16:9 (horizontal and vertical variants)
  • Multiple image styles, backgrounds, and creative angles
  • Different headline and description variations (200+ text combinations)
  • Regular asset refreshing as Google's algorithm adapts to creative fatigue

Traditional creative production can't support this volume. Agencies typically create 5-10 assets for PMax and hope it's enough. The result: PMax campaigns that underperform because they lack asset diversity.

Asset Types Google Requires

Image Assets for PMax

  • Landscape: 1200x628 (16:9), 1200x627 (LinkedIn ratio sometimes used)
  • Square: 1080x1080, 1200x1200
  • Vertical: 1080x1920, 1080x1350
  • Gmail: 600x400, 300x400

Video Assets (Optional but Recommended)

  • Short-form: 9:16 (TikTok/Reels style)
  • Standard: 16:9 (YouTube style)
  • Duration: 15 seconds recommended (6-60 seconds supported)

Headline and Description Assets

  • 15+ unique headlines (30 characters max)
  • 15+ unique descriptions (90 characters max)
  • Each headline can pair with any description (200+ combinations tested)

AI-Powered Asset Generation

Instead of hiring designers to create responsive ads in 10 formats, AI generates all assets in seconds.

Image Asset Generation

Provide a brief: "E-commerce electronics brand. Product: new wireless headphones. Create 30 images across all PMax aspect ratios. Variations: product on white, product in lifestyle (person wearing them), product on colored backgrounds (tech blue, black, minimal gray), detailed hero shots, lifestyle context (gym, commute, office). Mix professional and lifestyle aesthetics."

AI generates all 30 images at the correct aspect ratios, automatically optimized for responsive display (focal point centered, text safe zones respected, composition readable at all sizes).

Responsive Design Intelligence

AI doesn't just generate images—it understands responsive design. It ensures product positioning remains focal across aspect ratio changes. It keeps critical brand elements in safe zones. It designs compositions that don't lose meaning when cropped from 16:9 to 1:1 to 9:16.

Headline and Description Generation

AI generates copy variations that fit Google's character limits and follow ad copy best practices:

  • Value-focused headlines: "Save 30% on Wireless Headphones"
  • Social proof headlines: "Trusted by 500K+ customers"
  • Feature-focused headlines: "40-Hour Battery Life"
  • Urgency headlines: "Limited Time Offer"

Descriptions expand on headlines: "Premium sound, all-day comfort. Ships free. 30-day returns."

The system generates 15+ variation pairs, Google's algorithm tests all combinations, and the best-performing combinations get boosted automatically.

Dive deeper: Explore our complete guide to AI Creative for Google, LinkedIn, Pinterest & Beyond for advanced multi-platform strategy.

Responsive Design Principles

The secret to PMax success is understanding how creative scales. Responsive ads that look good in one format might look terrible in another.

Safe Zone Hierarchy

  • Critical elements: Must be visible in all formats. Typically logo, key product detail.
  • Primary elements: Should be visible in most formats. Product hero shot, main value proposition.
  • Secondary elements: Can crop in extreme formats. Supporting imagery, secondary copy.

Aspect Ratio Transitions

When content shifts from 16:9 (wide) to 1:1 (square) to 9:16 (tall), composition must reflow intelligently. AI handles this by designing primary focal points for the center of the image, ensuring visibility across crops.

Text Overlay Safety

Text overlaid on images is common in Google Display ads. AI places text to remain legible at all aspect ratios and ensures it doesn't obscure important product details.

Strategic Implementation for PMax

Step 1: Asset Planning

Decide on asset quantity and variety. Starting strategy: 25-30 images across all ratios, 15+ headline variations, 15+ description variations.

Step 2: Creative Brief

Provide a comprehensive brief including: product details, key benefits, brand color palette, target audience, competing offers, and emotional tone (premium, budget-friendly, innovative, etc.).

Step 3: AI Generation

AI generates all assets in one batch. Review for brand alignment, product accuracy, and responsive design effectiveness.

Step 4: Asset Upload to Google Ads

Upload all assets to a PMax campaign. Google's algorithm automatically tests combinations and optimizes for performance.

Step 5: Refresh Cadence

Generate new asset batches every 4-6 weeks. Google's algorithm gets smart about current creative, so rotation prevents fatigue.

The result: PMax campaigns that have 3-5x more assets than competitors, continuously refreshed, and deeply tested by Google's machine learning. Performance typically improves 20-35% when asset volume increases from 10 to 40+.

That's the AI advantage in PMax: not just time saved, but performance unlocked through asset scale that's otherwise economically infeasible.