Attribution Methods

Attribution Overview

Attribution is the process of matching a subscription purchase to the influencer who referred that user. InfluTo supports multiple attribution methods.

Method 1: Link-Based Attribution (Primary)

How It Works

1. User clicks influencer link
2. InfluTo stores click with device fingerprint
3. User installs app (can be hours/days later)
4. SDK sends device fingerprint on first launch
5. InfluTo matches fingerprint to recent click
6. Attribution success!

Device Fingerprinting

We match based on:

• IP Address: User's IP when clicking vs installing
• Device Info: Brand, model, OS version
• Screen Resolution: Unique to device
• Timezone: Geographic indicator
• Language: Device language setting

Matching Algorithm

Score = 0

if (ip_match):
    Score += 50  # Strongest signal

if (device_model_match):
    Score += 20

if (os_version_match):
    Score += 10

if (screen_resolution_match):
    Score += 10

if (timezone_match):
    Score += 5

if (language_match):
    Score += 5

if Score >= 70:
    → Attribution SUCCESS (95%+ confidence)
else:
    → No attribution (organic install)

Accuracy

• Same WiFi network: 98%+ accuracy
• Same cellular network: 95%+ accuracy
• Different network: 70-85% accuracy (device fingerprint only)

Method 2: Manual Code Entry (New in v1.2.0)

How It Works

1. User hears code on podcast/YouTube (no link click)
2. User installs app
3. User enters code manually in app
4. SDK validates code via API
5. SDK sets code in RevenueCat
6. Attribution recorded

Validation

POST /api/sdk/validate-code
{ "code": "FITGURU30" }

Response:
{
  "valid": true,
  "campaign": { ... },
  "influencer": { ... }
}

→ Code exists and campaign is active
→ Safe to apply

Use Cases

• Podcast mentions (verbal codes)
• YouTube videos (codes in description)
• Social media posts (codes in captions)
• In-person events (verbal sharing)

Attribution Window

Default: 72 Hours

User must install app within 72 hours of clicking link:

Monday 10:00 AM: User clicks link
Monday 10:05 AM: User installs app
→ Attribution SUCCESS (within window)

Monday 10:00 AM: User clicks link
Friday 9:00 AM: User installs app (95 hours later)
→ No attribution (outside 72-hour window)

Why 72 Hours?

• Balances accuracy vs conversion opportunity
• Industry standard (similar to Facebook Ads)
• Accounts for "install later" behavior

Customizable

Change attribution window in app settings:

• 24 hours: High confidence, lower conversions
• 72 hours: Balanced (recommended)
• 168 hours (7 days): Maximum conversions, lower confidence

Lifetime Attribution

First Touch Wins

Once a user is attributed, they stay attributed forever:

Month 1: User subscribes via FITGURU30
→ Influencer earns commission

Month 2: User renews (no referral code in webhook)
→ Influencer STILL earns commission (lifetime attribution)

Month 6: User upgrades plan
→ Influencer earns commission on new price

Month 12: User downgrades
→ Influencer earns commission on lower price

Why Lifetime Attribution?

• Fair to influencers (they brought the user)
• Incentivizes quality over quantity
• Influencers focus on retention, not just acquisition

Attribution Methods Comparison

Attribution Conflicts

Multiple Codes

What if user clicks multiple influencer links?

Day 1: User clicks @fitguru link (CODE1)
Day 2: User clicks @wellness link (CODE2)
Day 3: User installs app

→ Attribution goes to MOST RECENT click (CODE2)
→ Last touch wins

Link + Manual Code

What if user clicks link AND enters code?

User clicks @fitguru link → CODE1 stored
User installs app → CODE1 detected
User manually enters @wellness code → CODE2

→ Manual entry OVERRIDES link attribution
→ CODE2 used for final attribution
→ User choice wins

Privacy & Compliance

GDPR Compliance

• Device fingerprints are anonymized
• No personal data stored in clicks
• IP addresses hashed after 30 days
• Users can request data deletion

iOS ATT (App Tracking Transparency)

• Attribution works WITHOUT IDFA
• No tracking permission needed
• Uses first-party attribution only
• Privacy-friendly fingerprinting

Fraud Prevention

Anti-Fraud Measures

• IP abuse detection (same IP clicking many codes)
• VPN detection (suspicious patterns)
• Bot user-agent filtering
• Velocity checks (too many installs too fast)

Suspicious Patterns

Flagged if:
  - 100+ clicks from same IP in 1 hour
  - User-agent indicates bot/scraper
  - Device fingerprint impossible (screen: 0x0)
  - Install rate > 95% (organic is 10-20%)

Action: Attribution blocked, flagged for review

Next Steps

1. Learn about commission structure
2. Troubleshoot attribution issues