A/B Testing

Split traffic between different destinations to optimize conversions.

Overview

A/B testing allows you to distribute visitors across multiple target URLs based on configured weights. This helps you compare different landing pages, offers, or experiences to find what works best.

How It Works

Traffic Distribution

Creating Variants

Add a Variant

POST /api/urls/{id}/variants

{
  "name": "Variant A - Original",
  "targetUrl": "https://example.com/landing-v1",
  "weight": 50,
  "isActive": true
}

Parameters:

Parameter	Description	Required	Default
name	Variant name	✅	-
targetUrl	Destination URL	✅	-
weight	Traffic percentage (0-100)	❌	50
isActive	Enable variant	❌	true

Example: 50/50 Split Test

// Variant A
{
  "name": "Control - Original Page",
  "targetUrl": "https://example.com/page-original",
  "weight": 50
}

// Variant B
{
  "name": "Test - New Design",
  "targetUrl": "https://example.com/page-new",
  "weight": 50
}

Example: Multi-variant Test

// Variant A (40%)
{
  "name": "Original",
  "targetUrl": "https://example.com/v1",
  "weight": 40
}

// Variant B (40%)
{
  "name": "New Header",
  "targetUrl": "https://example.com/v2",
  "weight": 40
}

// Variant C (20%)
{
  "name": "Bold CTA",
  "targetUrl": "https://example.com/v3",
  "weight": 20
}

Weight Configuration

How Weights Work

• Weights are percentages (0-100)
• Total should equal 100 for predictable distribution
• If totals don't equal 100, distribution is proportional

Example proportional distribution:

Variant A: weight 60
Variant B: weight 40
Total: 100

Distribution:
- A receives 60% of traffic
- B receives 40% of traffic

Adjusting Weights

Update weights in real-time without downtime:

PUT /api/urls/{id}/variants/{variantId}

{
  "weight": 70
}

Traffic distribution changes immediately.

Managing Variants

List All Variants

GET /api/urls/{id}/variants

Response:

{
  "variants": [
    {
      "id": "var_123",
      "name": "Control",
      "targetUrl": "https://example.com/v1",
      "weight": 50,
      "isActive": true,
      "clickCount": 1500,
      "createdAt": "2025-01-01T00:00:00Z"
    },
    {
      "id": "var_456",
      "name": "Test",
      "targetUrl": "https://example.com/v2",
      "weight": 50,
      "isActive": true,
      "clickCount": 1480,
      "createdAt": "2025-01-01T00:00:00Z"
    }
  ]
}

Update a Variant

PUT /api/urls/{id}/variants/{variantId}

{
  "name": "Updated Name",
  "targetUrl": "https://example.com/new-url",
  "weight": 60,
  "isActive": true
}

Delete a Variant

DELETE /api/urls/{id}/variants/{variantId}

WARNING

Deleting the last variant automatically disables A/B testing for that URL.

Enable/Disable Variants

Temporarily disable a variant without deleting it:

PUT /api/urls/{id}/variants/{variantId}

{
  "isActive": false
}

Disabled variants receive no traffic but retain their statistics.

A/B Testing Analytics

View Test Performance

GET /api/analytics/ab-tests

Response includes:

• All URLs with A/B testing enabled
• Click distribution per variant
• Performance comparison

Per-Variant Metrics

Each variant tracks:

Metric	Description
Click Count	Total clicks received
Percentage	Share of total traffic
Geographic Distribution	Where clicks came from
Device Distribution	Device type breakdown

Comparing Results

{
  "urlId": "url_123",
  "slug": "landing",
  "totalClicks": 2980,
  "variants": [
    {
      "id": "var_123",
      "name": "Control",
      "clicks": 1500,
      "percentage": 50.3
    },
    {
      "id": "var_456",
      "name": "Test",
      "clicks": 1480,
      "percentage": 49.7
    }
  ]
}

Traffic Distribution

Random Assignment

Each visitor is randomly assigned to a variant based on weights:

1. Random number generated (0-100)
2. Mapped to variant based on cumulative weights
3. Visitor redirected to selected variant's URL

Consistent Experience

Once assigned, clicks are recorded with the variant ID for accurate tracking. However, each click is independently assigned - returning visitors may see different variants.

TIP

For consistent user experience across sessions, consider implementing client-side storage or user-based routing with Smart Routing.

Automatic A/B Testing

Auto-Enable

A/B testing is automatically enabled when you create the first variant for a URL.

Auto-Disable

A/B testing is automatically disabled when:

• All variants are deleted
• All variants are deactivated

When disabled, the URL redirects to its original target.

Best Practices

1. Start Simple

Begin with a 50/50 split between two variants:

• Control (original)
• Test (one change)

2. Test One Variable

Change only one element per test:

• Headline
• Call-to-action
• Image
• Layout

3. Gather Sufficient Data

Run tests long enough to get statistically significant results:

• Minimum 100 clicks per variant
• Run for at least 1-2 weeks
• Account for daily/weekly patterns

4. Document Your Tests

Use clear, descriptive variant names:

• ✅ "Blue CTA Button - Jan 2025"
• ❌ "Test 1"

5. Monitor Regularly

Check results frequently but avoid making changes too quickly:

• Review daily for anomalies
• Make decisions after sufficient data

6. Winner Takes All

After determining a winner:

1. Set winner weight to 100%
2. Or update original URL to winner
3. Delete losing variants

Use Cases

Landing Page Optimization

Test: Which headline converts better?

Variant A: "Save 20% Today"
Variant B: "Limited Time Offer"

Pricing Page Testing

Test: Annual vs Monthly first

Variant A: Annual pricing highlighted
Variant B: Monthly pricing highlighted

Call-to-Action Testing

Test: Button text effectiveness

Variant A: "Sign Up Free"
Variant B: "Get Started"
Variant C: "Try It Now"

Combining with Smart Routing

A/B testing can be combined with Smart Routing:

1. Smart routing rules evaluated first
2. If no rule matches, A/B testing applies
3. Enables targeted experiments for specific segments

Example:

• Mobile users → A/B test mobile landing pages
• Desktop users → A/B test desktop landing pages

Next Steps

• Analytics - Track test performance
• Smart Routing - Conditional redirects
• Webhooks - Click notifications