App Tracking Transparency (ATT)

IDFA, short for Identifier for Advertisers, is a distinct and unpredictable alphanumeric code that is assigned to each iOS device by Apple. Advertisers use this identifier to deliver targeted ads and measure the effectiveness of their advertising campaigns. With the changes to Apple's privacy policy, app developers are now required to explicitly ask for user permission to track their IDFA. Users can choose to opt out of IDFA tracking by turning on the "Limit Ad Tracking" (LAT) option in their device's settings.

SKAdNetwork (SKAN) is Apple's privacy-preserving attribution framework for measuring the effectiveness of advertising campaigns that drive app installs on iOS. Introduced as an alternative to IDFA-based tracking after ATT severely limited user-level tracking, SKAN provides aggregated, anonymized attribution data directly from Apple's servers — without revealing any information about individual users. Under SKAN, when a user clicks an ad and installs an app, Apple validates the install and sends an attribution postback to the ad network after a delay, containing limited information: the ad network ID, campaign ID, and a conversion value set by the developer. The conversion value (initially 6 bits allowing 64 possible values, expanded in SKAN 4.0 with coarse and fine values) is the developer's only mechanism for encoding post-install user behavior into the attribution signal. This means developers must carefully decide which events to encode — such as whether the user started a trial, made a purchase, or reached an engagement threshold — within a constrained time window. SKAN's design prioritizes user privacy through delayed, aggregated reporting with built-in noise, which makes campaign-level optimization significantly more challenging than traditional attribution methods.

Mobile attribution is the process of connecting app installs and in-app actions to specific marketing campaigns, ads, or channels that drove them. It enables marketers to understand which advertising efforts deliver results, optimize ad spend across channels, and make data-driven decisions about user acquisition strategies.

First-party data is information collected directly by an app developer from their own users through interactions on owned channels — including the app itself, websites, web funnels, email communications, and customer support touchpoints. This data includes behavioral signals (screens viewed, features used, session duration), purchase history (subscriptions, transactions, plan changes), declared preferences (onboarding survey responses, settings), and device-level signals (OS, device model, app version). In the post-ATT privacy landscape, first-party data has become the most valuable data asset for mobile app businesses. Unlike third-party data, which is collected by external entities and is increasingly restricted by platform privacy policies, first-party data is consented, reliable, and privacy-compliant by nature. Subscription apps that build robust first-party data strategies can use this information to personalize onboarding experiences, optimize paywall presentations, build predictive LTV models, power targeted re-engagement campaigns, and create lookalike audiences for acquisition — all without depending on third-party cookies or mobile advertising identifiers that are being deprecated across the ecosystem.

Fingerprinting, also called probabilistic attribution, is a method of identifying and matching users across touchpoints using non-deterministic data signals such as IP address, device type, operating system version, screen resolution, language settings, and timestamp proximity. Unlike deterministic attribution methods that rely on unique identifiers like IDFA or GAID, fingerprinting creates a statistical probability that a particular ad click and subsequent app install belong to the same user. Before Apple's App Tracking Transparency framework, fingerprinting served as a fallback when device-level identifiers were unavailable. However, Apple has taken an increasingly strict stance against fingerprinting, explicitly prohibiting it in their developer guidelines and enforcing against it in App Store reviews. Despite these restrictions, probabilistic methods remain relevant in certain contexts — particularly in web-to-app flows where a user's web session attributes can be matched against app-side signals to maintain attribution continuity without relying on restricted mobile identifiers. The accuracy of fingerprinting varies widely depending on the signals available and the matching window, typically ranging from 60–80% compared to near-100% accuracy for deterministic methods.