Cross-Platform Creative Testing

A/B tests ad creatives across all connected advertising platforms. Creates test variants, monitors performance, determines statistical winners, and applies winning creatives platform-wide.

Safety Guardrails

Read-only by default. This skill:

• Analyzes creative performance and outputs recommendations only
• Never auto-pauses or promotes variants without explicit flags
• Requires --apply flag to make any changes
• Generates detailed reports before modifications

Capabilities

1. Create A/B Tests - Set up multi-variant creative tests across platforms
2. Monitor Performance - Track CTR, CPC, CVR, ROAS per creative variant
3. Statistical Analysis - Calculate confidence intervals and significance
4. Winner Detection - Automatically identify winning variants at threshold
5. Apply Winners - Promote winning creatives, pause losers
6. Cross-Platform Sync - Replicate winning concepts to other platforms

Workflow

Step 1: Create Test Configuration

python3 scripts/create_test.py \
  --name "Q1 2026 Hero Image Test" \
  --platforms google,meta,linkedin \
  --variants 3 \
  --metric ctr \
  --confidence 95 \
  --min-impressions 1000 \
  --output test_config.json

Step 2: Upload Creative Variants

python3 scripts/upload_variants.py \
  --test-id test_config.json \
  --variant-a /path/to/hero_v1.png --headline-a "Try Free Today" \
  --variant-b /path/to/hero_v2.png --headline-b "Start Your Trial" \
  --variant-c /path/to/hero_v3.png --headline-c "Get Started Free"

Step 3: Monitor Test Performance

python3 scripts/monitor_test.py \
  --test-id test_config.json \
  --output performance.json

Step 4: Analyze Results

python3 scripts/analyze_results.py \
  --test-id test_config.json \
  --performance performance.json \
  --output analysis.json

Step 5: Apply Winner (Optional)

python3 scripts/apply_winner.py \
  --test-id test_config.json \
  --analysis analysis.json \
  --dry-run

# Remove --dry-run to execute
python3 scripts/apply_winner.py \
  --test-id test_config.json \
  --analysis analysis.json \
  --apply

Common Operations

Quick performance check on active tests

python3 scripts/monitor_test.py \
  --all-active \
  --summary-only

Create headline-only test

python3 scripts/create_test.py \
  --name "Headline Test" \
  --platforms meta \
  --type headline \
  --variants 4 \
  --metric ctr

Create image-only test

python3 scripts/create_test.py \
  --name "Image Test" \
  --platforms google,meta \
  --type image \
  --variants 3 \
  --metric cvr

Generate creative test report

python3 scripts/generate_report.py \
  --test-id test_config.json \
  --format markdown \
  --output creative_report.md

Sync winning creative to other platforms

python3 scripts/sync_winner.py \
  --test-id test_config.json \
  --source-platform meta \
  --target-platforms google,linkedin,tiktok \
  --dry-run

Statistical Methodology

Significance Testing

Uses Bayesian A/B testing with Beta-Binomial model for CTR/CVR:

1. Prior: Beta(1, 1) - uninformative prior
2. Posterior: Beta(1 + conversions, 1 + non-conversions)
3. Winner Probability: Monte Carlo simulation (10,000 draws)
4. Threshold: Variant wins if P(best) > confidence level

Minimum Sample Size

n = 2 × (Z_α/2 + Z_β)² × p(1-p) / δ²

Where:
- Z_α/2 = 1.96 (95% confidence)
- Z_β = 0.84 (80% power)
- p = expected baseline rate
- δ = minimum detectable effect

Metrics Tracked

Parameters

Scripts Reference

Output Format

Test Configuration JSON

{
  "test_id": "ct_2026011801",
  "name": "Q1 2026 Hero Image Test",
  "created_at": "2026-01-18T10:00:00Z",
  "status": "running",
  "platforms": ["google", "meta", "linkedin"],
  "type": "full",
  "primary_metric": "ctr",
  "confidence_level": 95,
  "min_impressions": 1000,
  "max_duration_days": 14,
  "variants": [
    {
      "id": "A",
      "name": "Control",
      "headline": "Try Free Today",
      "image_url": "https://your-cdn.com/creatives/hero_v1.png"
    },
    {
      "id": "B",
      "name": "Variant B",
      "headline": "Start Your Trial",
      "image_url": "https://your-cdn.com/creatives/hero_v2.png"
    }
  ],
  "budget_split": {
    "A": 50,
    "B": 50
  }
}

Analysis Results JSON

{
  "test_id": "ct_2026011801",
  "analyzed_at": "2026-01-25T10:00:00Z",
  "status": "winner_found",
  "winner": "B",
  "confidence": 97.3,
  "platforms": {
    "google": {
      "variants": {
        "A": {
          "impressions": 12500,
          "clicks": 375,
          "conversions": 45,
          "spend": 187.50,
          "ctr": 3.0,
          "cvr": 12.0,
          "cpc": 0.50
        },
        "B": {
          "impressions": 12500,
          "clicks": 500,
          "conversions": 65,
          "spend": 187.50,
          "ctr": 4.0,
          "cvr": 13.0,
          "cpc": 0.375
        }
      },
      "winner": "B",
      "confidence": 98.5,
      "lift": {
        "ctr": 33.3,
        "cvr": 8.3,
        "cpc": -25.0
      }
    },
    "meta": {
      "variants": {
        "A": {
          "impressions": 18000,
          "clicks": 540,
          "conversions": 54,
          "spend": 270.00,
          "ctr": 3.0,
          "cvr": 10.0,
          "cpc": 0.50
        },
        "B": {
          "impressions": 18000,
          "clicks": 720,
          "conversions": 79,
          "spend": 270.00,
          "ctr": 4.0,
          "cvr": 11.0,
          "cpc": 0.375
        }
      },
      "winner": "B",
      "confidence": 99.1,
      "lift": {
        "ctr": 33.3,
        "cvr": 10.0,
        "cpc": -25.0
      }
    }
  },
  "aggregate": {
    "winner": "B",
    "confidence": 97.3,
    "total_impressions": 61000,
    "lift": {
      "ctr": 33.3,
      "cvr": 9.5,
      "cpc": -25.0
    },
    "recommendation": "Promote Variant B across all platforms. Expected +33% CTR improvement."
  }
}

Platform-Specific Notes

Google Ads

• Creates Responsive Search Ads for headline tests
• Responsive Display Ads for image tests
• Asset-level reporting available
• Ad rotation: Optimize or Rotate Evenly

Meta Ads

• Uses Dynamic Creative for testing
• A/B test framework built-in
• Automatic placements recommended
• Breakdown by creative element

LinkedIn Ads

• Sponsored Content A/B testing
• Single image, carousel, video formats
• Member traits for audience splits
• Conversion tracking via Insight Tag

X Ads

• Promoted Tweets with variants
• Image, video, carousel formats
• Engagement objectives for CTR
• Website conversion tracking

Reddit Ads

• Link ads and text ads
• Subreddit targeting splits
• Conversion pixel for CVR
• API v3 creative management

TikTok Ads

• In-Feed ads, TopView, Spark
• Video-first creative testing
• Custom audiences for splits
• TikTok Pixel for conversions

Amazon Ads

• Sponsored Products/Brands/Display
• Custom headline tests
• ASIN-level performance
• Attribution for conversions

Best Practices

1. Test one element at a time - Isolate headlines, images, or CTAs
2. Ensure statistical significance - Wait for min impressions before deciding
3. Use realistic budgets - Don't starve variants with low spend
4. Match audiences - Ensure fair comparison with same targeting
5. Document learnings - Track what works for future campaigns
6. Consider seasonality - Account for weekly/monthly patterns
7. Iterate on winners - Use insights for next round of tests
8. Cross-platform sync - Apply learnings across all platforms

Environment Variables

Required environment variables:

# Platform Connection IDs

# Platform Account IDs
GOOGLE_ADS_CUSTOMER_ID
META_AD_ACCOUNT_ID
LINKEDIN_AD_ACCOUNT_ID
TWITTER_AD_ACCOUNT_ID
REDDIT_AD_ACCOUNT_ID
TIKTOK_ADVERTISER_ID
AMAZON_ADS_PROFILE_ID

# Google Ads specific
GOOGLE_ADS_DEVELOPER_TOKEN
GOOGLE_ADS_LOGIN_CUSTOMER_ID

# Database for test tracking
DATABASE_URL

Error Handling

Integration with Other Skills

• ad-copy-generation - Generate headline variants for testing
• cross-platform-launcher - Deploy winning creatives to new campaigns
• multi-channel-reporting - Include test results in reports
• budget-optimizer - Allocate more budget to winning creatives

About this skill

Part of the Synter free skills collection — open-source agent skills for advertising, PPC, and marketing automation.

Want this skill (and 40+ more) running on autopilot against your live ad accounts? Try Synter — AI Agent Media Buyers that connect to Google, Meta, LinkedIn, TikTok, Reddit, Amazon, and 7+ more platforms.

• 🌐 syntermedia.ai
• 📚 All free skills
• 💬 Built by @syntermedia — questions? Open an issue.