SKAdNetwork is Apple's StoreKit framework that validates installs driven by ads, sending the ad network a cryptographically signed postback that holds no user or device data, as Apple's SKAdNetwork documentation describes it. Its short name is SKAN. The main benefit of SKAdNetwork is that an ad network learns which campaign brought an install without tracking the person, and Apple says apps don't need App Tracking Transparency consent to use it.
Apple attributes an iOS app install by recording the ad impression on the device, waiting for the install and the first launch, then sending a postback after a delay, in what is called SKAdNetwork attribution. SKAN 4 postbacks carry a conversion value the app sets, in up to three postbacks over 35 days, with more or less detail depending on the size of the crowd. AdAttributionKit, Apple's newer framework from iOS 17.4, takes over this work for the App Store and alternative app marketplaces, and shares ad network IDs and impressions with SKAdNetwork.
This page covers how SKAdNetwork attributes an install, who does what, the time windows, conversion values, SKAN 4's three postbacks and crowd anonymity tiers, which ad wins, how App Tracking Transparency relates, how AdAttributionKit relates and reopens apps through Universal Links, and why SKAdNetwork cannot credit a link, which is where mobile app attribution, deterministic vs probabilistic attribution and a mobile measurement partner come in.
How does SKAdNetwork attribute an install without user data?
SKAdNetwork attributes an install without user data by keeping the ad impression on the device and letting the device, not the ad network, report the install later in a postback Apple signs. SKAdNetwork attribution runs in five steps, in the order Apple's "Receiving ad attributions and postbacks" lists them:
The ad network signs an ad.
A source app or a Safari web page shows the ad, and StoreKit records the impression.
The person installs the advertised app.
The person launches the app, and the app updates its conversion value.
The device sends a signed postback to the ad network after a delay.
Each of those steps belongs to one of three participants.
Who does what in SKAdNetwork?
Three participants share the work in SKAdNetwork, and Apple's SKAdNetwork documentation gives each its own jobs:
The ad network registers with Apple to get an ad network ID, signs the ads it serves, and receives and verifies the postbacks.
The source app lists the ad network IDs in its Info.plist file and shows the ads those networks sign.
The advertised app updates the conversion value when the person first launches it and, optionally, names a server URL in its Info.plist to receive a copy of the winning postback.
Every part of that exchange has a deadline.
Which time windows must an install meet to send a postback?
An install must meet five time windows to send a postback, one for each event from the ad to the first conversion update, per Apple's "Receiving ad attributions and postbacks":
Event | Window |
|---|---|
StoreKit-rendered ad shown | 30 days to install the app |
View-through ad shown | 24 hours to install the app |
Attributable web ad tapped | 30 days to install the app |
App installed | 60 days to launch the app |
App launched | 60 days from install to send the first conversion update |
Apple puts the minimum time between the ad impression and the ad network receiving a postback at 24 to 48 hours, cut to 5 to 10 minutes during testing with a downloaded profile. What the postback says about the person's actions depends on the conversion value.
What is a conversion value in SKAdNetwork?
A conversion value in SKAdNetwork is a number the app sets after launch to say what the person did, which Apple returns in the postback only when crowd anonymity allows it.
The fine conversion value is "an unsigned 6-bit value" from 0 to 63, per Apple's updatePostbackConversionValue(_:coarseValue:lockWindow:completionHandler:) reference, the same range Adjust and Tenjin give. The coarse conversion value is low, medium or high, per Apple's SKAdNetwork.CoarseConversionValue constants. Apple leaves the meaning to the app: "It's up to your app to decide what the conversion values signify in your app, both the fineValue and the coarseValue."
How does SKAN 4 send up to three postbacks?
SKAN 4 sends up to three postbacks by splitting the 35 days after the first launch into three conversion windows, each closed by its own postback, per Apple's "Receiving postbacks in multiple conversion windows":
Conversion window | Days after first launch | Value it can carry | Delay before sending |
|---|---|---|---|
First | Days 0 to 2 | Fine or coarse | 24 to 48 hours |
Second | Days 3 to 7 | Coarse only | 24 to 144 hours |
Third | Days 8 to 35 | Coarse only | 24 to 144 hours |
Three postbacks need iOS 16.1 or later and an ad signed with SKAdNetwork 4. Calling the update with lockWindow set to true finalizes the current window early, so its postback is prepared at once and sent after the same random delay.
How does crowd anonymity limit what a postback contains?
Crowd anonymity limits what a postback contains by placing each install in one of four postback data tiers, and the smaller the crowd, the less the first postback says. From the highest tier down, per Apple:
Tier 3: the source identifier with up to four digits, the fine value, the source app ID or source domain, and the country code when the country's crowd is large enough.
Tier 2: the source identifier with up to four digits and the fine value.
Tier 1: a two digit source identifier and a coarse value.
Tier 0: a two digit source identifier only, and only the first postback.
Apple sets the tier from the crowd size of the app or domain showing the ad, the advertised app, the country of the install and the source identifier. Tier decides the detail, and a second rule decides which network gets the detail at all.
Which ad wins the install when several ad networks qualify?
When several ad networks qualify, one ad network gets the winning postback, marked did-win true, and up to five other networks get one nonwinning postback each, with no conversion value or source app ID. Apple's rule from SKAdNetwork 2.2 is that the most recent impression with the highest fidelity-type wins, so a StoreKit-rendered ad, fidelity-type 1, beats a more recent view-through ad, fidelity-type 0. From iOS 15, the advertised app's developer can opt in to receive an exact copy of the winning postback, sent at the same time as the ad network's.
Does SKAdNetwork need App Tracking Transparency consent?
No, SKAdNetwork does not need App Tracking Transparency consent: Apple's SKAdNetwork documentation says "Apps don't need to use AppTrackingTransparency before calling SKAdNetwork APIs, and can call these APIs regardless of their tracking authorization status." Adjust's glossary records the history: with iOS 14.5 in April 2021, SKAdNetwork became the only way to attribute users who had not opted in to tracking. Apple also says the APIs have no effect when called from an App Clip, or from an iPhone or iPad app running on a Mac or on visionOS.
How does AdAttributionKit relate to SKAdNetwork?
AdAttributionKit, available from iOS 17.4, measures ad conversions on the App Store and on alternative app marketplaces, while SKAdNetwork works only with the App Store, per Apple's "Understanding AdAttributionKit and SKAdNetwork interoperability". Apple's SKAdNetwork page now tells developers to use AdAttributionKit for app ad campaigns. Four interoperability rules decide how the two frameworks share an install:
Only one impression can win a conversion, whether it comes from AdAttributionKit or SKAdNetwork.
Click-through ads always take precedence over view-through ads, and within each group the most recent impression wins.
The system considers a maximum of six impressions for any conversion.
Ad network IDs work in both frameworks, and SKAdNetwork's conversion update calls are mirrored into AdAttributionKit.
Adjust reports that at WWDC24 Apple expanded AdAttributionKit in place of an expected SKAN 5. The expansion also reaches people who already have the app.
How does AdAttributionKit reopen an installed app through a universal link?
AdAttributionKit reopens an installed app through a universal link when the person who already has the app taps an ad, which Apple counts as a re-engagement conversion. In iOS 18 and later, Apple's AdAttributionKit "Receiving ad attributions and postbacks" says a publisher app can pass a re-engagement URL, which the system opens only if the advertised app is installed and the URL is a registered universal link for it, as set up in Universal Links. The system appends the query parameter "AdAttributionKitReengagementOpen" so the app knows AdAttributionKit opened it, and strips known tracking parameters before the app receives the URL. Re-engagement creates no nonwinning postbacks and comes only from taps, not from view-through ads.
Why can't SKAdNetwork credit a single link or creator?
SKAdNetwork can't credit a single link or creator because it only credits ads signed by an ad network registered with Apple, and its postback names a campaign source identifier, never a person, so an install from a referral link, a creator's bio link or an SMS gets no postback at all. Those installs need link-level attribution, matched on the link's own click, as compared in deterministic vs probabilistic attribution. Smler's mobile app attribution credits such installs per link, to the last Smler link clicked inside a 90 day window.
What does an MMP do with SKAdNetwork postbacks?
An MMP collects the SKAdNetwork postbacks every ad network receives for an advertiser, maps conversion values to in-app events, and reports the campaigns side by side, which Adjust's glossary describes as helping clients "get the most of the data they receive". Adjust and Branch both sell conversion value setups, like Branch's SKAN Magic Set Up. The developer's own copy of the winning postback, available from iOS 15, is the same data the winning network gets, so a mobile measurement partner adds collection, decoding and reporting, not access.
How do you test SKAdNetwork postbacks before launch?
You test SKAdNetwork postbacks before launch with the StoreKitTest framework and SKAdTestSession, which Apple's "Receiving ad attributions and postbacks" names for unit testing ad impressions and postbacks. To check the time windows on a device, install Apple's testing profile, which cuts the postback delay from 24 to 48 hours to 5 to 10 minutes.