Mobile attribution has its own vocabulary. This glossary defines the 50 most important terms in mobile measurement and attribution. For deep linking terminology, see deep linking terminology: a glossary for developers. For a complete guide to mobile attribution, see mobile attribution: a developer's guide.
A
Ad Revenue Attribution
Attributing revenue from in-app ads (not purchases) back to the acquisition source. Important for ad-monetized apps (especially games) where a user's value includes the ad impressions they generate.
Adstock
A modeling technique that accounts for the lingering effect of advertising. An ad viewed today still influences behavior tomorrow. Adstock applies a decay function to past ad exposure to model this carryover effect.
App Tracking Transparency (ATT)
Apple's framework (introduced in iOS 14.5) that requires apps to request user permission before tracking them across apps and websites. Users who opt out cannot be tracked with IDFA.
Attribution Window
The time period after an ad click or impression during which a conversion can be attributed to that ad. Common windows: 7 days for clicks, 24 hours for views. Also called a "lookback window."
C
Click Flooding
A fraud technique where a network sends massive volumes of fake clicks for real device IDs. When any of those devices later installs the app organically, the fake click claims the attribution.
Click Injection
An Android-specific fraud technique where a malicious app on the device fires a fake click after detecting that an app installation has started, stealing attribution from the legitimate source.
Click-Through Attribution (CTA)
Attribution that credits a conversion to an ad that the user clicked. Stronger signal than view-through because the user took a deliberate action.
Click-Through Rate (CTR)
The percentage of ad impressions that result in a click. Calculated as clicks / impressions. Typical mobile ad CTR: 0.5-2%.
Cohort
A group of users who share a common characteristic, typically their install date. "The March 2026 cohort" refers to all users who installed in March 2026. Used for analyzing retention and LTV over time.
Conversion Value
In SKAdNetwork, a numeric value (0-63) that the app sets to communicate post-install behavior (engagement level, revenue tier) back to the ad network without revealing user identity.
Cost Per Action (CPA)
The cost of acquiring a user who completes a specific action (registration, purchase, subscription). More meaningful than CPI because it measures quality, not just installs.
Cost Per Install (CPI)
The cost of acquiring one app install. Calculated as ad spend / attributed installs. Industry average: $1-5 for casual apps, $5-20+ for fintech/subscription apps.
Cross-Device Attribution
Linking a user's activity across multiple devices (phone, tablet, desktop) to build a unified view of their journey. Uses deterministic (login-based) or probabilistic (IP/fingerprint) matching.
D
Data Clean Room
A secure environment where two parties can match their data without exposing raw data to each other. Used for privacy-safe attribution measurement.
Deferred Deep Link
A deep link that works even when the app is not installed. The user clicks the link, installs the app from the store, and then is routed to the intended content on first open.
Deterministic Matching
Matching users across touchpoints using definitive identifiers (device ID, email, login). Higher accuracy than probabilistic matching but requires a shared identifier.
Device ID
A unique identifier for a mobile device used for attribution matching. IDFA on iOS, GAID on Android.
F
Fingerprinting
A probabilistic matching technique that creates a "fingerprint" from device characteristics (IP address, user agent, screen resolution) to match users without a device ID. Increasingly restricted by platform policies.
First-Party Data
Data collected directly from your users through your own app or website. Includes user behavior, purchase history, and registration data. Not affected by third-party tracking restrictions.
Fraud
In mobile attribution, any activity designed to steal attribution credit or generate fake conversions for financial gain. Includes click flooding, click injection, device farms, and SDK spoofing.
G-I
GAID (Google Advertising ID)
The advertising identifier on Android devices. Users can reset or disable it in device settings. Equivalent to IDFA on iOS.
Holdout Group
A control group excluded from seeing ads in an incrementality test. The difference in conversion rates between the test group (sees ads) and holdout group (no ads) measures the incremental impact.
IDFA (Identifier for Advertisers)
Apple's advertising identifier for iOS devices. Since iOS 14.5, access requires user consent via ATT. Without consent, IDFA is zeroed out.
Impression
A single instance of an ad being served and viewed by a user. One impression does not mean the user noticed or engaged with the ad.
Incrementality
The additional conversions caused by a marketing campaign that would not have happened without it. Measured through controlled experiments (holdout tests).
Install Attribution
The process of determining which marketing touchpoint (ad click, deep link, referral) led to an app install.
Install Referrer
On Android, the Play Install Referrer API provides data about what referred the user to install the app, including campaign parameters and timestamps.
L-M
Last-Click Attribution
An attribution model that gives 100% credit to the last ad clicked before the conversion. The default and most common model, but ignores the contribution of earlier touchpoints.
Lifetime Value (LTV)
The total revenue a user generates over their entire relationship with the app. Used to evaluate whether the cost of acquiring a user (CPI/CPA) is justified.
Lookback Window
See Attribution Window.
Media Mix Modeling (MMM)
A statistical approach that uses aggregate data (total spend by channel, total outcomes) to measure the impact of marketing channels. Does not require user-level data, making it privacy-safe.
MMP (Mobile Measurement Partner)
A third-party attribution provider that tracks app installs and attributes them to marketing sources. Serves as a neutral arbiter between advertisers and ad networks.
Multi-Touch Attribution (MTA)
An attribution model that distributes credit across multiple touchpoints in the user journey, rather than giving 100% to the last click. Models include linear, time-decay, and data-driven.
P-R
Postback
A server-to-server HTTP callback sent from the attribution provider to the ad network when a conversion occurs. Contains campaign data, device identifiers, and event details.
Probabilistic Matching
Matching users across touchpoints using statistical signals (IP, user agent, timing) rather than definitive identifiers. Lower accuracy than deterministic matching.
Privacy Sandbox
Google's initiative to replace third-party cookies and advertising IDs with privacy-preserving APIs. Includes the Attribution Reporting API for Android.
Re-Engagement Attribution
Attributing a lapsed user's return to a specific re-engagement campaign (retargeting ad, push notification, email).
ROAS (Return on Ad Spend)
Revenue generated from a campaign divided by the cost of the campaign. ROAS of 150% means $1.50 in revenue for every $1.00 spent. Often calculated at multiple time windows (D7, D30, D90).
S
SDK Spoofing
A fraud technique where fake install or event signals are sent to the attribution provider without a real device or real app. The attacker reverse-engineers the SDK's communication protocol and generates fraudulent data.
Self-Attributing Network (SAN)
An ad platform that performs its own attribution instead of relying on a third-party MMP. Meta, Google, TikTok, Snapchat, and Apple Search Ads are all SANs.
SKAdNetwork (SKAN)
Apple's privacy-preserving attribution framework for iOS. Provides campaign-level (not user-level) attribution with delayed reporting and limited conversion values.
Smart Banner
A banner displayed on a mobile website that prompts users to open or install the native app. Can include a deep link to preserve context from the web page.
U-V
Universal Link
Apple's mechanism for deep linking on iOS. Universal Links use your website's domain and an apple-app-site-association file to open the app directly from web links.
View-Through Attribution (VTA)
Attribution that credits a conversion to an ad that the user saw but did not click. Uses shorter attribution windows (1-24 hours) and weaker signal strength than click-through.
View-Through Rate (VTR)
For video ads, the percentage of users who watched the video to completion. Higher VTR indicates better ad engagement.
Tolinku for Attribution
Tolinku provides deep link click tracking, conversion attribution, and analytics for mobile apps. The platform handles install attribution, deep link routing, and campaign measurement. Learn more in the Tolinku attribution docs.
For mobile attribution, see mobile attribution: a developer's guide. For deep linking terminology, see deep linking terminology: a glossary for developers.
Get deep linking tips in your inbox
One email per week. No spam.
