A probabilistic mobile attribution method that infers which affiliate link click originated a mobile app installation by comparing device characteristics (operating system version, device model, screen resolution, language settings, timezone, IP address) observed at click time with device characteristics observed at install time — used when deterministic device-level identification (advertising IDs, login-based matching) is unavailable. Device fingerprinting is the primary probabilistic attribution method used by mobile measurement partners (MMPs) to attribute app installs to affiliate clicks after Apple's App Tracking Transparency (ATT) framework reduced deterministic IDFA-based attribution. How fingerprinting works: when a consumer clicks an affiliate link, the click generates a server-side log of available device signals (IP address, user agent string, device type, OS version, timestamp); when the consumer installs the app within a time window (typically 24-72 hours), the MMP compares the new device's signals to recent click logs; if the signals match with sufficient confidence, the install is probabilistically attributed to that click. Accuracy limitations: device fingerprinting is less accurate than deterministic attribution because multiple consumers on the same network (shared IP address), multiple users on the same device, or consumers using VPNs can create false matches; typical fingerprinting accuracy is 60-80% vs. 95%+ for deterministic ID-based attribution; accuracy decreases in high-volume markets where many similar devices generate many simultaneous clicks. Privacy considerations: Apple has restricted device fingerprinting in iOS by limiting the precision of device signals available to third parties; technically, ATT prohibits using device characteristics for cross-app tracking purposes even when advertising IDs are not available; MMPs walk a fine line between probabilistic matching and Apple's stated fingerprinting restrictions.
Related Resources