A/B Testing
Run controlled experiments to optimize your ad campaigns.
Test Types
Creative Testing
Test different ad creatives:
- Images vs videos
- Carousel vs single image
- Different visual styles
- Color variations
Audience Testing
Compare audience segments:
- Demographics (age, gender)
- Interest-based audiences
- Lookalike audiences
- Custom audiences
Placement Testing
Test ad placements:
- Feed vs Stories
- Reels vs Search
- Desktop vs Mobile
- Platform comparison
Bid Strategy Testing
Compare bidding approaches:
- Manual vs automated bidding
- Different bid amounts
- Cost cap vs bid cap
- ROAS targets
Copy Testing
Test ad messaging:
- Headlines variations
- Description alternatives
- Call-to-action options
- Tone variations
Landing Page Testing
Compare destinations:
- Different page designs
- Form variations
- Content layouts
Creating A/B Tests
Setup Process
- Go to Ad Campaigns → A/B Testing
- Click Create New Test
- Select test type
- Configure variations:
- Control group (original)
- Variation A, B, C...
- Set traffic allocation
- Define success metrics
- Set duration and budget
- Launch test
Configuration Options
| Setting | Description |
|---|---|
| Traffic Split | Percentage per variation |
| Duration | Test runtime |
| Budget | Total or per-variation |
| Confidence Level | Statistical threshold (default 95%) |
| Primary Metric | Main success indicator |
Managing Tests
Test Statuses
- Draft - Not yet started
- Running - Actively collecting data
- Paused - Temporarily stopped
- Completed - Reached statistical significance
- Stopped - Manually ended
Monitoring Progress
Track in real-time:
- Impressions per variation
- Clicks and CTR
- Conversions
- Cost metrics (CPC, CPA)
- ROAS per variation
Statistical Analysis
Metrics Tracked
| Metric | Description |
|---|---|
| Impressions | Views per variation |
| Clicks | Engagement count |
| CTR | Click-through rate |
| Conversions | Goal completions |
| CPC | Cost per click |
| CPA | Cost per acquisition |
| ROAS | Return on ad spend |
| P-value | Statistical significance |
Winner Detection
The system automatically:
- Calculates statistical significance
- Identifies winning variation
- Shows lift percentage
- Provides confidence level
AI Analysis
Get AI-powered insights:
- Performance explanations
- Recommended actions
- Predicted outcomes
- Optimization suggestions
Best Practices
Test Design
- Test one variable at a time
- Use sufficient sample size
- Run tests long enough
- Define clear success metrics
Traffic Allocation
- Start with even splits
- Minimum 1000 impressions per variation
- Consider budget constraints
Duration
- Run for at least 7 days
- Account for day-of-week patterns
- Don't end tests early
Next: Bid Optimization →