SKAdNetwork mobile tracking is used to measure iOS acquisition campaigns without accessing the IDFA, Apple’s advertising identifier. Since the introduction of App Tracking Transparency in 2021, it has become the foundation of app attribution on the Apple side, supplemented by AdAttributionKit since 2024. The direct result: less individual data and delayed reports, but actionable insights if the tagging plan, events, and business expectations are properly defined from the start.
SKAdNetwork Mobile Tracking: What Has Really Changed
Before iOS 14.5, many advertisers tracked their installs using the IDFA, a device-based advertising identifier. Since April 26, 2021, Apple has required App Tracking Transparency (ATT) to track a user across apps or websites and access this identifier. Without explicit consent, the IDFA is not available.
Apple also prohibits fingerprinting—that is, attempts to identify a device by combining technical signals such as the IP address, model, or configuration. For a business leader, this issue is therefore not just about marketing. It involves balancing metrics, compliance, media budget, and the risk of app rejection.
SKAdNetwork apporte a response: Apple attributes an install to a campaign without revealing the user’s identity. The ad network receives a postback—in other words, an attribution message—containing limited informations. You know that a campaign has generated installs and certain quality signals, but not the complete individual user journey.
In the projects we work on, we often see the same mistake: treating SKAdNetwork as just an SDK (software development kit) setting. It’s too late. The quality of the measurement depends primarily on the events selected before launch: sign-up, free trial, purchase, subscription, level achieved, or adding to cart. There’s no magic solution on the reporting side if these milestones are poorly designed.
What does SKAdNetwork mean in business terms?
SKAdNetwork is Apple’s ad attribution framework for apps distributed through the App Store. A framework is a set of rules and technical tools provided to developers. In this case, it allows an installation to be attributed to an ad while limiting the available data, in order to protect user privacy.
Starting with SKAdNetwork 4.0, rolled out in 2022–2023, Apple allows for three conversion windows to measure what happens after installation. The first covers days 0 through 2, the second covers days 3 through 7, and the third covers days 8 through 35. This is useful for apps whose value isn’t immediately apparent, such as a subscription purchased after a few days of a free trial.
The distinction is clear. The more detailed conversion values are only available in the first postback. Subsequent postbacks use coarse values—low, medium, or high—and remain subject to Apple’s privacy thresholds, known as “crowd anonymity.” If the volume is too low, certain information is not reported.
Another point that’s often underestimated: postbacks 2 and 3 arrive after their window has closed, with a random delay ranging from 24 to 144 hours. A campaign launched on Monday therefore does not reveal its full picture by Friday. To manage a budget on a day-to-day basis, you must accept a degree of uncertainty and supplement your analysis with other metrics, such as cost per click, aggregated App Store installs, or anonymized internal metrics.
How can you track your installs without IDFA?
The right approach is to distinguish between three levels. First, media attribution: which campaign appears to have driven the installation. Next, user engagement: what do users do after installation, without identifying them across services. Finally, financial performance: what acquisition cost remains acceptable based on margin, retention, and average order value.
In practical terms, the product team must define a conversion matrix. A conversion value is a signal sent to Apple to indicate a level of value after installation. For example: opening an account, performing the first business action, making a purchase, or starting a trial. The pitfall is trying to measure everything. With SKAdNetwork, it’s better to have three signals that are closely tied to revenue than an exhaustive mapping that becomes impossible to interpret.
AdAttributionKit, introduced by Apple in 2024, extends this approach. It also limits the data in postbacks and interoperates with SKAdNetwork depending on the ad network’s integration. An important difference: AdAttributionKit covers the App Store and alternative marketplaces, while SKAdNetwork remains specific to the App Store. Since iOS 17.4, alternative marketplaces have been available in the European Union, making this distinction less theoretical for certain apps.
Your MMP—short for Mobile Measurement Partner—often serves as the centralization tool. AppsFlyer, Adjust, Branch, and Singular help consolidate data from networks, but they do not circumvent Apple’s rules. They translate, normalize, and model the data. They do not reveal the IDFA when the user has not authorized it.
- Define business events Before SDK integration: qualified sign-up, first order, subscription, repeat purchase.
- Building a SKAN Grid Suitable for windows of 0–2, 3–7, and 8–35 days, without striving for impossible precision.
- Testing Postbacks with ad networks and the MMP before allocating media budgets.
- Aligning Marketing and Finance based on realistic benchmarks: cost per installation, cost per activated user, and time to return on investment.
- Documenting ATT Consent and data processing practices to ensure compliance with the 2018 GDPR.
If your app relies on sensitive APIs, marketing efforts must not compromise security. SDKs added for attribution, advertising, or analytics must be audited just like the rest of the app; the risks are similar to those described in our guide on Mobile API Security.
iOS, Android: A Comparison of Two Privacy-First Approaches
Android doesn’t work like iOS, but the direction is similar: fewer persistent identifiers, more privacy-preserving attribution APIs. Google made the Privacy Sandbox APIs available in beta on public Android devices in 2023. The Attribution API supports app install attributions and post-install conversions without cross-app advertising identifiers.
The Google 2025 documentation distinguishes, in particular, between event-level reports—which are closer to individual events but are limited—and aggregatable reports, which produce aggregated reports. During testing, debug keys can provide unaltered rapports alongside protected rapports, which helps technical teams validate the integration.
For a decision-maker, the implication is simple: you need to set up a different system for iOS and Android, then reconcile the results at the business level. A single bord table is still possible. A perfect solution? No. Honestly, any promise of comprehensive user measurement without consent should be viewed with caution.
| Environment | Award Framework | Available Data | Effect on steering |
|---|---|---|---|
| iOS App Store | SKAdNetwork 4.0 | 3 postbacks: fine-grained values at the start, followed by low/medium/high values based on Apple's thresholds | Delayed analysis; sufficient volumes are needed |
| The Store iOS app and alternative app stores | AdAttributionKit, introduced in 2024 | Limited postbacks; interoperability with SKAdNetwork varies by network | What to Expect for hors App Store Apps in the Relevant Markets |
| Android | Privacy Sandbox Attribution API, in public beta since 2023 | Event-level rapports and aggregated rapports, with testing mechanisms | Scalable, privacy-first measure (encore); integration to be monitored over time |
| Mobile web | Cookies, consent, server-side solutions | Depends on the browser, the CMP, and the analytics settings | Comparing mobile apps and the web is challenging without a common attribution model |
This difference between platforms also matters when choosing an application technology. Native iOS/Android development, Flutter, React Native, or Kotlin Multiplatform do not prevent SKAdNetwork, but they do affect the integration and testing process. For architecture trade-offs, our analysis on Kotlin Multiplatform and Its Actual Costs provides a good reference point.
Budget, Timelines, and Risks: Key Figures to Keep in Mind
In the French market, the in-house integration of an app attribution solution rarely costs nothing, even if you’ve already selected an MMP. For an existing app, expect to pay around €3,000 to €10,000 (excluding tax) to set up event tracking, integrate or trigger the SDKs, test postbacks, and produce initial usable documentation. For a new app with paid user acquisition planned, the effort is generally included in the analytics package, but it must be budgeted for.
MMP subscriptions vary significantly depending on the number of installations, the modules, and the terms negotiated. Some plans start with affordable tiers, while others become more cost-effective as volumes increase. With this budget, it’s sometimes better to invest initially in simple, reliable tracking tools and then add an advanced MMP when media spending justifies multi-network analysis.
In terms of timelines, allow two to four weeks for a thorough integration into an app that’s already structured—and even longer if the business events aren’t clear or if multiple ad networks are involved. You’ll also need to factor in SKAdNetwork’s natural processing time: you won’t be able to validate everything in real time. Test cycles must account for conversion windows and random delays.
The main risk isn’t just technical. It’s making poor marketing decisions based on incomplete data. Cutting a campaign short, over-prioritizing a channel that tracks conversions better, comparing iOS and Android as if the rules were the same—these mistakes can quickly end up costing more than the integration itself.
The pitfalls that non-technical people don't notice until it's too late
First pitfall: confusing compliance with performance. Simply displaying the ATT prompt is not enough. The text, the timing of the prompt, and the lead-in screen all influence the acceptance rate, but Apple strictly regulates what can be requested. An aggressive or misleading prompt can increase the risk of rejection.
Second pitfall: overlooking privacy thresholds. If your campaigns are too fragmented by audience, country, creative, or micro-budget, Apple may hide certain details. The result: the more you segment, the less you see. For an SME, grouping campaigns together at the outset often yields better insights than highly granular targeting that’s impossible to interpret.
Third pitfall: letting each service provider choose their own metrics. The media agency wants to optimize quickly, the developer wants to limit changes, and the executive wants to see revenue figures. Without a shared decision-making process, the conversion matrix becomes a shaky compromise. From the agency’s perspective, the natural approach is to start with the business model and then work backward to measurable events—not the other way around.
App security also deserves a place in this discussion. Every third-party SDK can introduce dependencies, permissions, or network traffic. Before release, a review similar to this Mobile App Security Checklist Avoid pitting acquisition against technical robustness.
When the Perfect Measure Becomes a Bad Idea
The temptation is strong to rebuild the old world: to attribute everything, connect everything, and optimize everything down to the individual user. This usually leads to a dead end. Apple, Google, and the standards ported by the IAB Tech Lab in 2025 are moving toward privacy-preserving attribution, with formalized use cases for environments without third-party cookies.
A more sustainable strategy combines several imperfect but consistent layers: SKAdNetwork mobile tracking for iOS, the Attribution Reporting API for Android, anonymized product analytics, consent-based CRM data, and financial analysis. No single tool is sufficient on its own. Together, they provide a solid enough picture to make decisions.
The scenario where the obvious solution is the wrong one: launching a major user acquisition campaign before you’ve stabilized the onboarding process. You’ll get installs—maybe even postbacks—but you’ll mainly be measuring churn. With a limited budget, optimizing the app’s early stages can yield better results than adding an ad network.
Defining the scope of this type of project early on helps avoid most unpleasant surprises: choosing events, Apple and Google restrictions, SDK security, and analyzing campaign financials. Above all, an outside perspective helps ensure the right decisions are made before media budgets start rolling out.
FAQ on SKAdNetwork Mobile Tracking
Does SKAdNetwork completely replace the IDFA?
No. SKAdNetwork does not replace the IDFA exactly: it provides aggregated and limited attribution, without individual tracking. If the user accepts ATT, the IDFA may still be available, but you should not base your entire measurement strategy on it.
How long does it take to get the SKAdNetwork results?
The first signal covers days 0 through 2, and the subsequent postbacks cover days 3 through 7 and 8 through 35. Postbacks 2 and 3 may arrive 24 to 144 hours after the end of their respective windows.
Does SKAdNetwork work on Android?
No, SKAdNetwork is an Apple framework related to the App Store. On Android, Google is promoting the Privacy Sandbox Attribution Reporting API to measure installs and conversions without cross-app advertising identifiers.
Do you need an MMP to track your app campaigns?
It isn't required for a small launch, but it's often useful when multiple networks, countries, or campaigns are running simultaneously. MMP simplifies consolidation without removing the limits imposed by Apple or Google.