SKAdNetwork 4 (postbacks)
Apple sends the ad network a signed, delayed, anonymized report that an ad led to an install.
In one line
Can tell an ad network an ad led to an install, in a delayed report
Apple sends the ad network a signed, delayed, anonymized report that an ad led to an install.
Gives ad networks proof an ad drove an install without a device ID. Reports are coarse, delayed and may be dropped for small campaigns.
What it can and can't do
What the user does: You tap an ad, install the app, then use it for a few days
It can
- Send a signed report that an ad led to an install
- Send up to three reports over weeks
- Show a small campaign number (2 to 4 digits)
It cannot
- Give your name or device ID
- Report the same day; delays and low counts remove detail
- Work on Android; it uses the install referrer
How it differs between platforms: iPhone only (iOS 11.3 and later; iOS 16.1 for the newest version). Android uses the install referrer.
On Android and iPhone
| Android | iPhone | |
|---|---|---|
| Status | Not supported | Supported |
| First version | Not available | iOS 11.3 |
| Permission or entitlement | Not listed | SKAdNetworkItems (Info.plist, network IDs); app calls SKAdNetwork.updatePostbackConversionValue |
| Protection level | none | none |
| How access is granted | not-exposed | none |
| What the user sees | Not listed | Not listed |
| How long it lasts | Not listed | Apple signs postbacks; app and ad network never see device IDs |
| How to take it back | Not listed | Not listed |
| Store policy | None listed | App Review: no fingerprinting; ATT not required for SKAN |
| API | Not listed | SKAdNetwork, SKStoreProductViewController, SKOverlay |
| Notes | Android uses Install Referrer instead. | SKAN 4 (iOS 16.1): three postbacks, hierarchical source ID (2 to 4 digits), coarse and fine values, web-to-app. |
Data you get
What comes back: Signed postback to ad network
Shape: object
How often: up to 3 postbacks per install, each delayed
| Field | Type | Unit | Range | Precision |
|---|---|---|---|---|
source-identifier | string | digits | 2 to 4 | set by crowd anonymity |
conversion-value | integer | None | 0 to 63 (fine, first postback) | Not listed |
coarse-conversion-value | enum | None | low, medium, high | Not listed |
fidelity-type | integer | None | 0 view-through, 1 StoreKit-rendered click | Not listed |
Technical limits
- Postbacks to the ad network only (app developer does not get it directly)
- Randomized delay per window
- Crowd anonymity tiers cut data on small campaigns
- No user or device identifiers
The story
What it is
Apple's system that tells an ad network an ad led to an install, in a delayed, anonymous report.
For you
You tap an ad, install and open the app. Days later Apple sends the ad network a signed note that one install came from that ad, with no name or device ID.
For makers
It gives ad networks proof an ad drove an install, without an ID. Reports are coarse and delayed and may be dropped for small campaigns, so makers choose their budgets with less detail than before.
Where you see it
| Type | Example | Apps |
|---|---|---|
| games | A game studio finds out a video ad brought 500 installs and ties them to a coarse revenue band. | Zynga, Playtika |
| shopping | A retail app confirms an ad brought installs, but cannot see which person. | Temu |
| ads | An ad network gets signed proof so it can bill the advertiser without a device ID. | AppLovin, Unity |
| social | A social app learns which campaign brought sign ups after seven days. | Snapchat |
| analytics | A marketing team compares SKAN data with its own sales data to see the gap. | Singular |
| security | Signatures stop fake networks from claiming installs they did not drive. | None named |
Risks
- Coarse and delayed
- Small campaigns lose detail
Android and iPhone in short
- Android: Android has no SKAN. Google used the install referrer and advertising ID.
- iPhone: SKAdNetwork 4 added three postback windows and coarse values.
What the platform says
- Apple: SKAdNetwork. A privacy preserving way to measure ad driven installs without revealing user or device level data.
Good uses and abuse
Legitimate uses
- Privacy-safe install measurement
Known abuse
- Abused by conversion-value encoding to infer behavior
Scores
| Score | Value | Meaning |
|---|---|---|
| Sensitivity | 2 of 5 | How revealing the data is |
| Utility | 4 of 5 | How useful it is to honest apps |
| Friction | 1 of 5 | How much the user has to do |
| Fingerprint bits | Not rated | Rough identifying entropy, when known |
History
- 2018-03, iPhone iOS 11.3: SKAdNetwork 1.0
- 2019-06, iPhone iOS 13: Version 2: view-through attribution
- 2021-04, iPhone iOS 14.5: Version 3 plus ATT, becomes the default measure for non-consented users
- 2022-10, iPhone iOS 16.1: SKAN 4: multiple postbacks, hierarchical IDs, coarse values, web-to-app
- 2024-03, iPhone iOS 17.4: AdAttributionKit introduced as the successor framework
Related capabilities
Related
- SKAN conversion values (fine and coarse). Can send the ad network a small score about what you did after install
- SKAN crowd anonymity tiers. Can't get much detail when too few people installed from an ad
- SKAN postback windows and timers. Can only get install reports days later, in three windows with random delays
- AdAttributionKit (iOS 17.4). Can tell an ad network an ad led to an install, on iPhone, with delays
Same idea on the web
- Attribution Reporting API (retired). Can't measure which ad click led to a purchase in Chrome anymore
Sources
How well checked: Checked against real behavior.