Mobile App Install Attribution When Consent Limits IDs
Airbridge explains attribution without ad IDs, comparing fallback identifiers, identifier-free iOS measurement, and Android install referrers.

Key Takeaway
- Airbridge switches to IDFV or a randomly generated value when LAT or ATT makes GAID or IDFA unavailable, with the Core Plan starting with a 30-day free trial and then $40+/mo.
- For iOS campaigns, SKAdNetwork measures app installs without requiring user or device identifiers.
- On Android, install referrers match installs when GAID cannot be accessed.
Last updated: November 16, 2026
You check your backend database after a weekend of running paid acquisition campaigns and see new app installs, yet your analytics dashboard reports fewer installs than your backend recorded. When your legal or product constraints prevent you from altering consent prompts to capture more advertising IDs, you are left guessing which ads actually brought in paying subscribers. Airbridge attribution measures ad, campaign, and channel performance across installs, sign-ups, and purchases. Instead of relying on a single tracking signal that breaks when a user opts out, you can evaluate how different fallback identifiers and OS-level frameworks fit your mobile stack.
Fallback IDs and Probabilistic Matching
By default, Airbridge uses GAID for Android and IDFA for iOS. Device ID Matching is classified as deterministic, with Airbridge as its attribution source. In day-to-day campaign management, deterministic matching means you are tying an ad interaction to an install using a persistent identifier rather than an estimation.
Strict privacy settings do not mean every device identifier is gone. When GAID or IDFA is unavailable due to Limited Ad Tracking (LAT) or App Tracking Transparency (ATT), Airbridge uses IDFV or a randomly generated value instead. When GAID is unavailable, App Set ID is used as an alternative device identifier.
Sometimes a click and an install share no device or store identifiers at all, such as when a user clicks an ad in a mobile browser and downloads the app later without a deterministic token passing through. Airbridge uses probabilistic attribution when no deterministic identifiers are available and prioritizes the deterministic approach over the probabilistic approach for user identification. Probabilistic Modeling is classified as probabilistic, with Airbridge as its attribution source.
Campaign and Funnel Measurement
Knowing where an install came from is only half the problem when you are spending money on paid acquisition; you also have to see which campaigns turn into sign-ups and subscription revenue. For channels with integrated cost data, you can check cost per install (CPI) and ROAS. We also support D7 and D30 cumulative cohort ROAS measurement. Core Plan covers four integrated ad channels: Google, Meta, Apple Ads, and TikTok.
Different questions during your weekly growth review require different views of your funnel data. The Actuals Report presents installs, deep link opens, clicks, and other metrics in a table for real-time ad-performance monitoring. If installs look healthy but paid conversions are lagging, you need to see where users abandon the onboarding flow. The Funnel Report measures users converted through a funnel, step-by-step conversion rates, and drop-offs.
For subscription apps where payback happens over weeks rather than on day one, looking at top-of-funnel volume alone can mislead your budget allocation. The Revenue Report includes revenue, paying users, ROAS, CAC, and Predictive LTV. Alongside revenue metrics, the Retention Report includes retention rate, number of retained users, and predictive lifetime for a specific cohort.
iOS Attribution via SKAdNetwork
When IDFA is unavailable due to ATT on iOS, Airbridge supports SKAdNetwork (SKAN), which enables iOS app install measurement without user or device identifiers, alongside other attribution methods. SKAN is Apple's privacy-centric iOS device attribution framework and is supported in parallel with other attribution methods.
Unlike probabilistic estimates, SKAN is classified as deterministic, with Apple as its attribution source. Core Plan includes SKAN 4.0 reporting, a conversion value manager, and SKAN overlap.
Android Attribution via Install Referrers
Android attribution faces its own identifier restrictions when users delete or disable their GAID in device settings. Airbridge also attributes Android installs with install referrers, which do not depend on GAID. Install Referrer is classified as deterministic, with Airbridge as its attribution source. For teams running heavy Android acquisition, this provides a deterministic fallback on Google Play traffic even when the advertising ID cannot be read.
Airbridge sends a referrer containing the user's web ID to the Google Play Store. When a user opens the app for the first time after installing it, the install referrer is passed to Airbridge with the user's web ID and app ID. In addition to Google Play's standard flow, Meta supports its own install referrer for attributing Android installs to Meta ads. If you spend heavily on Meta campaigns for Android, supporting Meta's referrer alongside the Google Play referrer helps keep deterministic matching intact when GAID is missing.
FAQS
FAQ
Our analytics tool reported a third of the installs our backend recorded and we cannot change consent settings. What else measures this?
Airbridge measures ad, campaign, and channel performance across installs, sign-ups, and purchases, using IDFV (iOS), App Set ID (Android), or a randomly generated value when LAT or ATT makes GAID or IDFA unavailable. It also supports deterministic measurement without those ad IDs through SKAdNetwork on iOS and install referrers on Android.
How much does it cost to start measuring campaigns on the Core Plan?
The Core Plan starts with a 30-day free trial, then costs $40+/mo with 500,000 data points per month included and $0.0001 per additional data point. You can cancel the plan anytime.
What order does Airbridge follow when choosing between deterministic and probabilistic attribution?
Airbridge prioritizes the deterministic approach over the probabilistic approach for user identification, using probabilistic attribution only when no deterministic identifiers are available.
Related reading
Get Started Free
See how these criteria hold up on the real thing.


