Lecture Notes on Mobile Advertising and Attribution

Overview of Mobile Ads

• Platforms discussed: TikTok, Google Ads, Facebook (Meta)
• Key result: Determine how many customers installed the app from each campaign and their subsequent behaviors.

Key Metrics to Track

• Installations: Number of devices that installed the app via each campaign.
• User Behavior After Installation:
  • Sign-ups
  • Building attendance in the app
  • Interacting with sellers (chat)
  • Making offers and closing deals

Attribution Basics

• Attribution: Process of attributing user actions (e.g., installs, sign-ups) to specific campaigns across different platforms.
• Importance of data: Helps optimize ads and creative strategies.

Traditional Attribution Method

• User actions are tracked at the device level using device IDs:
  • iOS: IDFA (Identifier for Advertisers)
  • Android: AD ID
• The old method involved tracking each ad platform's performance based on these device IDs.
• Limitations: User privacy regulations limit the tracking of individual device identifiers.

SKAdNetwork Overview

• SKAdNetwork: New privacy-focused attribution framework by Apple.
• Users must authorize tracking for IDFA; otherwise, it cannot be accessed.
• SKAdNetwork abstracts user identifiers while tracking campaign performance.

Attribution Process with Branch

1. User Interaction: User clicks on an ad (e.g., on TikTok).
2. Ad Click Event: TikTok sends an ad click event to Branch, including the device identifier.
3. App Installation: User installs the app; Branch tracks this as an install event.
4. Matching Events: Branch matches install events with ad click events based on device ID.

Last Click Attribution Logic

• Last click attribution: Credits the last ad clicked before installation.
• Other models:
  • First Click: Crediting the first ad a user interacted with.
  • Data Driven: Distributing credit among multiple interactions based on influence.

Challenges in Attribution

• Data Discrepancy: Differences in reported installs between Branch and individual ad platforms.
• Self-Reporting Networks (SRN): Major ad platforms often report their own credit for installs they attribute.
• Aggressiveness of platforms (e.g., Facebook claims credit even on ad impressions).

Transition to SKAdNetwork

• SKAdNetwork diminishes the granularity of device-level data:
  • Campaign-level data only (no device IDs).
  • Data delays for reporting conversions.
  • Conversion values sent to Apple after a set timer.

Key Differences in SKAdNetwork Versions

• Scan 3 vs. Scan 4:
  • Scan 3: Single postback from Apple; delay in data reporting.
  • Scan 4: Three postbacks allowing for more timely and detailed reports (not yet widely adopted).

Setting Up Conversion Values

• Conversion Values: Define key behaviors to track (e.g., install, view item, make an offer, purchase).
• Requirement: At least 100 installs per day per campaign for Apple to send data.

Conclusion

• Need to collaborate on the events to track and their corresponding conversion values.
• Consideration for development documentation to implement tracking.
• Importance of understanding attribution in optimizing ad campaigns.