Four AI Tools for Ad Creatives, Compared

These are the main AI tools used for Meta ad creatives. Each has clear strengths and weaknesses.

Recommendations by Use Case

Product shots (need accurate product imagery):

• First choice: DALL·E 3 or Adobe Firefly
• Second choice: Manual photography (AI-assisted editing)
• Why: Midjourney struggles with product accuracy

Lifestyle scenes (users and environments):

• First choice: Midjourney
• Second choice: Advantage+ Creative variations
• Why: Midjourney's emotional quality works well in ads

Images with embedded text:

• First choice: DALL·E 3 (strong text rendering)
• Second choice: Canva AI (templates + AI combo)

Video creatives:

• First choice: Sora (OpenAI, rolling out)
• Second choice: Runway Gen-3
• Third choice: Real footage + AI post-processing

Real-World Workflow

Step 1: Planning (human-led)

• Lock down brand tone and message
• Draft 3-5 creative concepts
• Define target audience

Step 2: AI drafts

• 10-20 lifestyle images in Midjourney
• 5-10 product shot variations in DALL·E
• Automated variations via Advantage+ Creative inside Meta

Step 3: Selection and editing (human-led)

• Pick the ones that match brand tone
• Fine-tune in Photoshop or Canva
• Add copy

Step 4: Launch and measure

• Upload 3-5 to Meta Ads Manager
• Pick winners via A/B testing

Cost Efficiency Comparison

Small brands (monthly ad spend $500-2,000):

• Midjourney $10/month + Advantage+ (free) = $10/month
• Saves $500-1,000 per photographer session

Mid-tier ($2,000-10,000):

• Midjourney $30/month + ChatGPT Plus $20/month + Advantage+ = $50/month
• Reduces creative production headcount

Large scale ($10,000+):

• Pro AI tool stack + local Stable Diffusion = $100-500/month
• Roughly equivalent to one in-house AI artist

Quality Gotchas

Midjourney mistakes:

• Six fingers and asymmetric faces are common
• Always review at full size
• Be careful with ads featuring people

DALL·E mistakes:

• Accuracy drops with multiple objects
• Product color and size can be distorted
• Compare against real product

Shared risks:

• AI artifacts (unnatural patterns)
• Potential copyright or trademark issues (famous logos sneaking in)

Policy Compliance

Patterns that make Meta's ad review suspect AI imagery:

• Unrealistically "perfect" skin
• Impossible body proportions
• Repeating AI-style background patterns

How to handle it:

• Add human editing to AI drafts (post-processing)
• Flag AI generation in metadata (Meta requires it)
• Never use celebrity faces that could be mistaken for deepfakes

So What Do We Do?

Advertisers just starting out:

• Midjourney free trial, then $10/month subscription
• Turn on Advantage+ Creative options (already included)
• Add Canva Pro ($12/month) alongside

After 1-2 months, once stable:

• Add DALL·E 3 (via ChatGPT Plus)
• Build your own brand-style prompt library

After 6 months:

• External AI + Meta automation pipeline
• Can produce 20+ creatives per week

Long-Term Outlook

Meta keeps strengthening Advantage+ Creative. Dependency on external tools will likely decrease over time. In 2-3 years:

• Meta's built-in AI may handle most needs
• External AI stays for specialized use cases
• Advertisers focus on planning and validation

Creative planning and high-volume AI creative production are covered in Meta Ads Book 3.