Deduplication System

Convultra uses a multi-tier deduplication system to prevent double-counting conversions.

Why Deduplication Matters

Without deduplication, conversions can be counted multiple times:

• Page refresh after purchase
• User clicking "back" to confirmation page
• Network retries
• Multiple tracking sources (client + server)

This leads to:

• Inflated conversion numbers
• Inaccurate ROAS calculations
• Wasted ad spend on over-attributed campaigns

Deduplication Tiers

Convultra applies four levels of deduplication:

Tier 1: Conversion ID (Order ID)

Window: 24 hours (configurable)

The primary deduplication method using your provided order/transaction ID:

Convultra.trackPurchase({
  orderId: 'ORD-12345', // Used for deduplication
  value: 99.99
})

Tier 2: Event ID

Window: 5 minutes (configurable)

Unique identifier generated per event:

// Automatically generated
{
  event_id: 'evt_1704067200000_abc123',
  event_type: 'purchase',
  ...
}

Catches:

• Exact duplicate requests (network retries)
• Rapid-fire button clicks

Tier 3: Click ID + Value

Window: 1 hour (configurable)

Same click ID + same value = likely duplicate:

// Scenario: User refreshes thank you page
{
  gclid: 'Cj0KCQ...',
  value: 99.99
}
// Second request with same gclid + value = duplicate

Tier 4: Browser Fingerprint

Window: 5 minutes (configurable)

Same fingerprint + same value within short window:

{
  fingerprint: 'fp_a1b2c3...',
  value: 99.99
}

Catches:

• Users without click IDs
• Direct traffic conversions

Deduplication Flow

Event Received
      ↓
[1] Check Order ID (24h window)
      ↓ not duplicate
[2] Check Event ID (5min window)
      ↓ not duplicate
[3] Check Click ID + Value (1h window)
      ↓ not duplicate
[4] Check Fingerprint + Value (5min window)
      ↓ not duplicate
✅ Process as unique conversion

Configuration

Dashboard Settings

Dashboard → Settings → Deduplication

SDK Configuration

Convultra.init({
  apiKey: 'your-key',
  endpoint: 'https://your-endpoint.com',
  deduplication: {
    orderIdWindow: 86400,     // 24 hours in seconds
    eventIdWindow: 300,       // 5 minutes
    clickIdWindow: 3600,      // 1 hour
    fingerprintWindow: 300    // 5 minutes
  }
})

Best Practices

1. Always Use Order IDs

// ✅ Good
Convultra.trackPurchase({
  orderId: 'ORD-12345',
  value: 99.99
})
 
// ❌ Bad - no orderId
Convultra.trackPurchase({
  value: 99.99
})

2. Use Unique Order IDs

// ✅ Good - unique per transaction
orderId: `ORD-${timestamp}-${random}`
 
// ❌ Bad - reused IDs
orderId: 'purchase'

3. Include Lead IDs

For leads, use a unique identifier:

Convultra.trackLead({
  leadId: `LEAD-${formId}-${timestamp}`,
  value: 100
})

Viewing Duplicates

Dashboard

Dashboard → Conversions → Duplicates

Shows:

• Original event
• Duplicate attempts
• Deduplication tier triggered

API

// Query duplicates
GET /v1/events/duplicates?date_from=2024-01-01

Platform Deduplication

In addition to Convultra's deduplication, each ad platform has its own:

Convultra sends these fields to enable platform-side deduplication too.

Troubleshooting

Duplicates Still Appearing

1. Check orderId uniqueness - Are IDs truly unique?
2. Check time windows - Is the gap between events too long?
3. Check multi-source - Are you tracking same event client + server?

Legitimate Events Marked Duplicate

1. Extend time windows - If legitimate events are within window
2. Check value differences - Small value changes should be unique
3. Verify orderId format - Ensure IDs are properly generated

Next Steps

• Attribution Models →
• Conversion Recovery →