Deferred Deep Linking

Deferred deep linking allows you to attribute users and deliver them to specific content even when they didn't have your app installed when they clicked the link.

The Problem

Without deferred deep linking:

1. User sees ad for shoes on Instagram
2. Clicks link → App not installed
3. Redirected to App Store
4. Installs app
5. Opens app → Lands on home screen 😞
   (Lost: which shoes they wanted, that they came from Instagram)

The Solution

With Clippr deferred deep linking:

1. User sees ad for shoes on Instagram
2. Clicks link → Clippr records click + device fingerprint
3. Redirected to App Store
4. Installs app
5. Opens app → SDK matches device → Returns link data
6. App navigates to shoes page 🎉
   (Knows: shoes ID, came from Instagram, part of summer campaign)

How Matching Works

Clippr uses multiple strategies to match clicks to installs:

1. Install Referrer (Android Only)

Accuracy: 100%

When a user installs from Play Store after clicking a link, Google passes referrer data directly to the app. This is deterministic and cannot fail.

Click → Play Store installs with referrer → App reads referrer → Perfect match

2. Deterministic Matching

Accuracy: 100%

If the user has an advertising identifier (IDFA on iOS, GAID on Android) and consented to tracking:

Click captures IDFA/GAID → Install sends same ID → Exact match

3. Probabilistic Matching

Accuracy: ~85-95%

When deterministic methods aren't available, Clippr uses device fingerprinting:

Signals collected at click:

• IP address
• User agent
• Device model
• Screen resolution
• Timezone
• Language
• OS version

Matching algorithm:

• Finds recent clicks with similar fingerprints
• Calculates confidence score
• Returns match if confidence exceeds threshold

Match Types in Code

final link = await Clippr.getInitialLink();
if (link != null) {
  switch (link.matchType) {
    case MatchType.direct:
      // User clicked with app installed
      // 100% confidence
      break;

    case MatchType.deterministic:
      // Matched via Install Referrer or IDFA/GAID
      // 100% confidence
      break;

    case MatchType.probabilistic:
      // Matched via fingerprinting
      // Check link.confidence (0.0 - 1.0)
      if (link.confidence! > 0.8) {
        // High confidence match
      }
      break;

    case MatchType.none:
      // Organic install, no link clicked
      break;
  }
}

Attribution Window

The time between click and install matters:

Clippr's default window is 72 hours. Beyond this, probabilistic matching becomes less reliable.

Maximizing Match Rate

1. Prompt for Tracking Permission (iOS)

Request ATT permission for higher accuracy:

import AppTrackingTransparency

ATTrackingManager.requestTrackingAuthorization { status in
    // Initialize Clippr after permission is determined
    Clippr.initialize(apiKey: "YOUR_API_KEY")
}

2. Initialize Early

Initialize the SDK as early as possible:

void main() async {
  WidgetsFlutterBinding.ensureInitialized();
  await Clippr.initialize(apiKey: 'YOUR_API_KEY');
  runApp(MyApp());
}

3. Call getInitialLink() Once

The SDK performs the match on first call and caches the result:

// Do this once, early in app lifecycle
final link = await Clippr.getInitialLink();

4. Use Install Referrer (Android)

The SDK automatically uses Install Referrer when available. Ensure your Play Store listing is properly configured.

One-Time Matching

Deferred matching is one-time by design:

1. User clicks link
2. User installs app
3. First app open → SDK matches → Returns link
4. Subsequent opens → Returns null

This prevents:

• Counting the same install multiple times
• Showing the same deep link content repeatedly
• Attribution fraud

Handling Low Confidence Matches

For probabilistic matches with low confidence:

void handleDeepLink(ClipprLink link) {
  if (link.matchType == MatchType.probabilistic) {
    final confidence = link.confidence ?? 0;

    if (confidence < 0.5) {
      // Low confidence - might be wrong user
      // Options:
      // 1. Ignore the link
      // 2. Show confirmation: "Were you looking for [X]?"
      // 3. Log for analysis but don't navigate
      return;
    }
  }

  // High confidence - proceed normally
  navigateTo(link.path);
}

Debugging Deferred Links

Enable debug mode to see matching details:

await Clippr.initialize(apiKey: 'YOUR_API_KEY', debug: true);

Logs will show:

• Click lookup results
• Fingerprint comparison
• Confidence calculation
• Final match decision

Common Issues

Match Rate Lower Than Expected

1. Long install delay: Users installing days after clicking
2. VPN/proxy usage: Changes IP between click and install
3. Shared devices: Multiple users on same device
4. Privacy settings: Limited tracking enabled

No Match Found

1. User never clicked a link: Organic install
2. Click expired: Beyond attribution window
3. Different device: Clicked on one device, installed on another
4. SDK not initialized: Check initialization order

Next Steps

• Attribution Guide - Understanding attribution data
• SDK Quick Start - Implement deferred linking