SKAN 4.0 is a structural overhaul of Apple's privacy-preserving attribution framework, not an incremental update. It introduces three postbacks instead of one, hierarchical conversion values (coarse and fine-grained), crowd anonymity tiers that gate how much data you receive, and web-to-app attribution. Despite being available since iOS 16.1, adoption has been glacially slow: according to analysis on the Mobile User Acquisition Show, Meta reported only roughly 5% SKAN 4.0 adoption among ad networks as of late 2024. Understanding the practical differences between 3.0 and 4.0 matters because Apple has since rebranded the framework as AdAttributionKit (AAK), but the underlying architecture, including the 64 conversion values, three postbacks, and crowd anonymity thresholds, remains unchanged.
Page Contents
SKAdNetwork 3.0
SKAN 3.0, released alongside iOS 14.5 in 2021, is the version most UA teams have actually operationalized. It delivers a single postback per install, fired after a timer window of 24 to 63 hours (depending on how long conversion value updates extend the timer). You get one 6-bit conversion value, which encodes up to 64 possible states. According to Singular's SKAN benchmarks report, most advertisers use fewer than 20 of those 64 values because mapping revenue ranges and engagement events into 6 bits is operationally complex. The postback goes to the ad network only, not the advertiser directly, and includes the campaign ID (limited to 100 campaigns per network per app).
Pros
- Near-universal adoption: virtually all ad networks support 3.0 postbacks
- Well-documented optimization playbooks exist after 3+ years of industry use
- Single postback simplifies data pipelines and conversion value decoding
- Campaign ID limit of 100 is manageable for most mid-size advertisers
Cons
- Single postback means zero visibility into post-install events beyond the timer window
- 6-bit conversion value forces painful tradeoffs between revenue and engagement signals
- No web-to-app attribution: Safari-originating installs receive no SKAN signal
- Null conversion values reported at roughly 30-40% of installs per AppsFlyer's data
Best for: SKAN 3.0 remains the practical workhorse for teams that need reliable, well-supported attribution across all ad networks today. If your optimization loops depend on first-postback signals and your MMP has mature 3.0 decoding, you are already operating on this version whether you realize it or not.
Need help scaling your mobile app growth? Talk to RocketShip HQ about how we apply these strategies for apps spending $50K+/month on UA.
SKAdNetwork 4.0 (and AdAttributionKit)
SKAN 4.0, available since iOS 16.1 (October 2022) and architecturally mirrored in AdAttributionKit, introduces three sequential postbacks, hierarchical conversion values, crowd anonymity tiers, and web-to-app support. The first postback can carry a fine-grained (6-bit) conversion value if crowd anonymity conditions are met. Second and third postbacks arrive at days 3-7 and days 8-35 after install respectively, but carry only coarse values (low, medium, high). According to Apple's developer documentation, crowd anonymity operates across four tiers (0-3), and the data you receive is directly proportional to the install volume a campaign generates. This means smaller campaigns get less data, which can be a dealbreaker for niche targeting.
Pros
- Three postbacks theoretically provide 35-day post-install visibility
- Hierarchical conversion values let you measure engagement AND revenue without tradeoffs
- Web-to-app attribution covers Safari ad clicks for the first time
- Source identifier expands to 4 digits (up to 10,000 campaign combinations) at the highest tier
Cons
- Only ~5% ad network adoption per Meta's reporting, severely limiting practical utility
- Second and third postbacks carry only coarse values (low/medium/high), limiting signal quality
- Crowd anonymity tiers mean low-volume campaigns receive null or minimal data
- Timer windows extend up to 35 days, making real-time optimization nearly impossible for later postbacks
Best for: SKAN 4.0 is theoretically superior for subscription apps and games with longer monetization curves that need post-install signals beyond day 1. In practice, it only delivers meaningful incremental value if you run high-volume campaigns on networks that have fully implemented 4.0 support. Most teams should prepare for this architecture (since AAK mirrors it) but not rely on postbacks 2 and 3 for optimization decisions today.
Side-by-Side Comparison
| Feature | SKAN 3.0 | SKAN 4.0 / AAK |
|---|---|---|
| Number of postbacks | 1 | 3 (day 0-2, day 3-7, day 8-35) |
| Fine-grained conversion values | 6-bit (64 states), always included | 6-bit, only in first postback IF crowd anonymity tier 2-3 |
| Coarse conversion values | Not available | Low / Medium / High in all 3 postbacks |
| Crowd anonymity tiers | Not applicable (binary null or value) | 4 tiers (0-3) gating data granularity |
| Campaign ID capacity | 2 digits (up to 100) | 2-4 digits (up to 10,000) depending on tier |
| Web-to-app attribution | Not supported | Supported via Safari Private Click Measurement |
| Null conversion value rate | ~30-40% per AppsFlyer data | Higher at low tiers; first postback comparable at high tiers |
| Network adoption rate (2025) | ~95%+ of major networks | ~5% per Meta's observed adoption data |
| Timer window (first postback) | 24-63 hours | 24-63 hours (unchanged) |
| Postback recipient | Ad network only | Ad network + developer (copy) for first postback |
| Practical optimization window | Day 0-1 events only | Theoretically day 0-35, practically still day 0-1 |
| MMP support maturity | Fully mature across AppsFlyer, Adjust, Singular, Kochava | Partially implemented; second/third postback decoding inconsistent |
Verdict
Choose SKAN 3.0 as your operational baseline if you need reliable, well-supported attribution across every major ad network today. The single postback with 6-bit conversion values is a known quantity, and every MMP has mature tooling to decode and model around it. Your entire measurement framework can be built on first-postback data with predictive modeling filling the gaps. Choose SKAN 4.0 as your architectural target, especially if you are running high-volume campaigns (thousands of daily installs per campaign) where crowd anonymity tiers will unlock the expanded source identifiers and fine-grained values. Subscription apps and mid-core games with day 7 to day 30 monetization curves stand to gain the most from second and third postbacks, even with coarse values. A "high" coarse value on the third postback confirming a subscription renewal is still a meaningful signal. The honest practitioner's take: do not restructure your entire measurement stack around SKAN 4.0's second and third postbacks. According to the Mobile User Acquisition Show's analysis of AAK, these later postbacks remain "largely ineffective" for most advertisers because coarse values compress too much signal and the long timer windows make them useless for real-time bidding decisions. Focus your energy on optimizing campaigns with limited SKAN data from the first postback, invest in predictive LTV models, and use incrementality testing as your source of truth for channel-level decisions. RocketShip HQ's Weighted Anomaly Scoring framework, which weights metric changes by business impact using the formula abs(% change) x sqrt(spend), helps here: a ROAS drop on a $5K/day campaign matters far more than a similar drop on $200/day, and that prioritization logic works regardless of whether you are on 3.0 or 4.0.
Frequently Asked Questions
Do I need to update my MMP to use SKAN 4.0?
Yes, but your MMP likely already supports it. AppsFlyer, Adjust, Singular, and Kochava have all shipped SKAN 4.0 and AdAttributionKit support. The bottleneck is not your MMP; it is ad network adoption, which sits at roughly 5% according to Meta's observed data.
What are crowd anonymity tiers and how do they affect my data?
Crowd anonymity tiers (0 through 3) determine how much data Apple includes in your postback. At tier 0, you get almost nothing (null values). At tier 3, you unlock fine-grained conversion values and up to 4-digit source identifiers. According to Apple's documentation, tier thresholds are based on install volume and are not publicly disclosed, but industry consensus places tier 2-3 thresholds at hundreds to thousands of installs per campaign.
Can SKAN 4.0 attribute installs from web ads in Safari?
Yes, SKAN 4.0 adds web-to-app attribution via Safari's Private Click Measurement. This fills a significant gap because SKAN 3.0 provided zero attribution for installs originating from web-based ad clicks. For advertisers running mobile web campaigns alongside app install campaigns, this is arguably the most immediately useful 4.0 feature.
Why does Google UAC reporting sometimes not match my MMP's SKAN data?
Google UAC uses its own modeled conversions alongside SKAN postbacks, creating discrepancies with your MMP's SKAN-only view. The Mobile User Acquisition Show covered this gap in detail: Google may claim installs that your MMP cannot verify through SKAN because of timer expiration, null conversion values, or different attribution logic. Cross-referencing both sources and using incrementality tests is the only reliable way to reconcile.
Should I map coarse conversion values for the second and third postbacks?
Yes, but keep expectations low. Map coarse values (low, medium, high) to broad engagement tiers: for example, "low" = opened but no engagement, "medium" = completed onboarding, "high" = made a purchase or started a trial. According to Singular's SKAN benchmarks, most advertisers are not yet receiving enough second and third postbacks to make statistically significant optimization decisions.
Is AdAttributionKit replacing SKAN entirely?
Apple rebranded SKAdNetwork as AdAttributionKit (AAK) starting with iOS 17.4, but the underlying architecture is the same: 64 conversion values, 3 postbacks, crowd anonymity tiers. Think of it as SKAN 4.0 with a new name and minor API changes. If your stack supports SKAN 4.0, the migration to AAK is incremental, not a rebuild.
How does SKAN attribution differ from web attribution models like last-click?
SKAN is fundamentally different from web attribution because it is deterministic at the campaign level but probabilistic at the user level. There is no click ID, no user-level path, and no multi-touch attribution. The differences between mobile and web attribution mean that web frameworks like Google Analytics' last-click model simply do not translate. You must build a mobile-native measurement framework that combines SKAN signals, modeled conversions, and incrementality.
What is the biggest practical win from SKAN 4.0 right now?
The expanded source identifier is the most immediately useful feature. At crowd anonymity tier 3, you get 4-digit source identifiers instead of 2-digit, which lets you encode creative ID, placement, or audience segment alongside your campaign ID. This dramatically improves campaign-level optimization without waiting for the second or third postback.
Looking to scale your mobile app growth with performance creative that delivers results? Talk to RocketShip HQ to learn how our frameworks can work for your app.
Not ready yet? Get strategies and tips from the leading edge of mobile growth in a generative AI world: subscribe to our newsletter.
Related Reading
- Privacy-first attribution and measurement for mobile apps (comprehensive guide)
- What Is AdAttributionKit and How Is It Different from SKAN? (2026)
- AppsFlyer App Retargeting Report: Benchmarks and Post-ATT Strategies (2026)
- AppsFlyer Mobile Ad Fraud Report: Fraud Rates and Protection Benchmarks (2026)
- How Has ATT Changed Mobile Advertising? (2026)
