AI UGC for Google Demand Gen Campaigns: The Complete Guide

Google Demand Gen campaigns replaced Discovery ads in late 2023 and have since become the primary upper-funnel campaign type for e-commerce brands on Google. They combine the visual richness of social ads with Google's intent signals and reach—serving image and video ads across 3 billion+ daily users on YouTube, Gmail, and Discover. The bottleneck for most brands is creative: Demand Gen's algorithm needs 10–15+ image variants per asset group to fully optimize, and the images must look native to each surface. AI UGC solves both problems.

Step 1: Understand Demand Gen Placements and Asset Requirements

Demand Gen ads appear across four distinct surfaces, each with different visual expectations:

Key insight: Each surface has a different “native look.” YouTube rewards bold, aspirational imagery. Discover rewards editorial, magazine-quality visuals. Gmail rewards clean, product-forward creative. AI UGC lets you generate surface-appropriate lifestyle imagery for each placement without three separate photo shoots.

Image Spec Requirements

• Landscape (1.91:1): Minimum 600×314 px. Used across all placements. This is your most important format.
• Square (1:1): Minimum 300×300 px. Used in YouTube, Discover, and responsive formats.
• Portrait (4:5): Minimum 480×600 px. Primarily for Discover and YouTube Shorts interstitials.
• Max file size: 5 MB per image. JPEG or PNG.
• Recommended images per asset group: 10–15 unique images across all aspect ratios.

Google's machine learning tests every combination of your images with headlines and descriptions. More high-quality image variants means more combinations to optimize across. Brands uploading 15+ images consistently outperform those uploading the minimum 3.

Step 2: Define Audience Segments and Intent Signals

Demand Gen campaigns offer audience targeting that combines Google's search intent data with social-style audience building:

• Lookalike segments: Based on your customer lists or website visitors. These are your highest-intent prospects and should receive your most product-specific lifestyle imagery.
• Custom segments: Built from search terms, URLs, and app usage. Match creative to the interests that define each segment—fitness-focused imagery for fitness keyword segments, home-focused imagery for home decor URL segments.
• In-market audiences: Users actively researching products in your category. Conversion-focused lifestyle imagery showing products in use performs best here.
• Optimized targeting: Google expands beyond your selected audiences to find converters. Provide diverse lifestyle imagery so the algorithm can match the right creative to the right user.

Planning tip: Create a matrix of audience segments × products × creative angles. Each cell in the matrix represents a specific image you need to generate. For a brand with 10 products and 3 audience segments, that is 30 unique lifestyle image needs—before accounting for aspect ratio variants.

Step 3: Create AI Experts Matched to Audience Segments

The AI experts you create should mirror the demographics and lifestyles of your target audience segments. Here is a recommended approach:

• Map expert profiles to segments. If you are targeting fitness enthusiasts, create AI experts in athletic wear and gym settings. If you are targeting parents, create experts in home and family settings.
• Build 4–8 experts. This gives you enough diversity to cover your key segments while keeping production manageable. See our guide to creating AI experts.
• Vary demographics intentionally. Age, ethnicity, body type, and personal style should reflect the diversity of your customer base. Google's algorithm will learn which expert resonates with which audience.
• Consider the placement context. Experts in aspirational, polished settings work for YouTube. Experts in natural, everyday settings work for Discover. Plan your expert profiles with placement in mind.

Step 4: Upload Products and Plan Scene Compositions

Upload your product photos to the Props Library and plan scene compositions tailored to each Demand Gen surface:

• YouTube scenes: Bold, aspirational, lifestyle-driven. Show products being used in dynamic, visually striking settings. Think: morning routine with a skincare product, outdoor workout with activewear, cooking scene with a kitchen gadget.
• Discover scenes: Editorial, magazine-quality, content-style. These should look like they belong in a lifestyle article, not an ad. Think: styled flat lay, a person reading with a product on the table, a candid coffee-shop moment.
• Gmail scenes: Clean, product-forward, professional. The product should be the hero of the image with a clean background or minimal setting. Think: product on a marble counter, held against a neutral background, placed on a desk.

For comprehensive product photography planning, see our product photo shot list guide.

Step 5: Batch-Generate Images in Required Aspect Ratios

Generate images systematically to cover every asset group need:

Total per product: 14–22 images. For a catalog of 15 products, that is 210–330 total images. With AI UGC, this batch is achievable in 1–2 focused sessions. For batch workflows, see our batch production guide.

Step 6: Export and Upload to Google Ads

When exporting AI UGC for Demand Gen campaigns, follow these specifications:

• Resolution: Export at the highest available resolution. Google accepts up to 5 MB per image but will downscale as needed for each placement.
• Format: JPEG for lifestyle photos (smaller file sizes), PNG only if you need transparency.
• Text on images: Avoid text overlays. Google combines your images with headlines and descriptions automatically. Text on images competes with Google's own ad assembly and can reduce quality scores.
• Logo placement: Google adds your logo separately. Do not burn logos into your lifestyle images.
• Naming convention: Use descriptive file names (e.g., skincare-vanity-scene-landscape.jpg) to keep your asset library organized.

Upload organization: In Google Ads, create separate asset groups within your Demand Gen campaign for each product category or audience segment. Upload the relevant AI UGC images to each asset group so Google tests the right creative with the right audience.

Step 7: Configure Asset Groups and Launch A/B Experiments

Demand Gen campaigns are organized around asset groups—each one containing a set of images, headlines, descriptions, and audience targeting. Here is how to structure them for maximum AI UGC impact:

Asset Group Structure

• One asset group per product-audience combination. This lets Google optimize creative combinations within a focused context rather than mixing signals across unrelated products and audiences.
• Upload all aspect ratios per asset group. Include landscape, square, and portrait images in every asset group so Google can serve your ads across all placements.
• Use 10–15 images per asset group. This is the sweet spot for giving the algorithm enough testing surface without diluting signal across too many variants.

A/B Experiments

Use Google's campaign experiments feature to test AI UGC against your current creative:

• Test 1: AI UGC lifestyle images vs. white-background product shots. Expect 20–40% higher CTR from lifestyle images.
• Test 2: AI UGC lifestyle images vs. stock photography. Expect 10–25% higher CTR from AI UGC due to authenticity and specificity.
• Test 3: High volume (15+ images per asset group) vs. low volume (5 images). More variants typically produce better algorithmic optimization.

For comprehensive creative testing methodology, see how to A/B test AI UGC ad creative.

Step 8: Optimize Based on Asset Performance Signals

Google Demand Gen provides asset-level performance reporting that tells you how each image is performing:

• Best: Top-performing assets. Keep these and generate similar variants.
• Good: Solid performers. Keep for now but test replacements.
• Low: Underperforming. Replace immediately with new AI UGC variants.
• Learning: Not enough data yet. Wait for 1,000+ impressions before judging.

Refresh Cadence

Demand Gen creative decays faster than search ads because users see the same visuals across YouTube, Gmail, and Discover. Follow this refresh schedule:

• Every 2 weeks: Replace all “Low” rated images with new AI UGC variants.
• Every 4 weeks: Refresh 30–50% of all images, even “Good” performers, to prevent ad fatigue.
• Quarterly: Full creative overhaul with new AI experts, scenes, and seasonal context.

Without AI UGC, this cadence is unsustainable for most brands. With it, generating a full refresh batch takes hours, not weeks. For the complete creative refresh playbook, see how to beat ad fatigue.

Performance Benchmarks and ROI

Here are the performance benchmarks e-commerce brands can expect when running AI UGC in Demand Gen campaigns:

The ROI math is straightforward: a brand spending $10K/month on Demand Gen that reduces CPA by 40% through better creative saves $4,000/month while generating the same number of conversions. The AI UGC production cost for that creative library is under $500. For more on tracking returns, see how to track AI UGC ROI.