SKAdNetwork (SKAN) is Apple's privacy-preserving attribution framework for iOS app ads. The device, not the ad network, decides which ad gets credit for an install under Apple's rules, and the winning network receives an Apple-signed postback after a deliberate delay. That postback carries a short source identifier the network maps to its campaigns and a conversion value your app sets, never user or device identifiers. On Meta, SKAN is one of two campaign attribution methods for iOS 14+ app promotion campaigns; the other is Meta's own Aggregated Event Measurement (AEM). Below: postbacks, conversion values and windows, how Ads Manager reports SKAN, AEM versus your MMP, schema design, and budgeting when data arrives days late.
What SKAdNetwork is and why it exists
Before App Tracking Transparency, mobile attribution leaned on the device advertising identifier, matching clicks to installs person by person. Since iOS 14.5, apps must ask permission before tracking, and many people say no. SKAdNetwork is Apple's privacy-safe alternative: it works whether or not someone granted tracking permission, and it tells you that an ad led to an install, and roughly what happened next, without revealing who the person was.
The flow is simple. Meta signs the ads it serves. If the person installs within Apple's attribution window and opens the app within 60 days of installing, the device records the attribution and the app updates a conversion value as the person engages. The device then sends an Apple-signed postback to the winning network, plus an exact copy to the developer if the app is set up to receive one, which is how many mobile measurement partners (MMPs) get SKAN data. Up to five losing networks with qualifying impressions get a postback marked as not winning, with no conversion value.
The rules for who wins are Apple's, not Meta's. A tapped ad beats a viewed ad even when the view is more recent, and within each group the most recent impression wins. A click-through install must happen within 30 days of the tap and a view-through install within 24 hours. Because every network using SKAN is judged by the same on-device rules, it is the closest thing iOS has to a neutral referee. The price is precision: data arrives late, aggregated and sometimes stripped of detail.
Postbacks, conversion values and windows
The conversion value is the only post-install signal SKAN carries, so it is what you design. The fine value is a number from 0 to 63, giving 64 states to encode events, revenue ranges or both. The coarse value, added in SKAN 4, is simply low, medium or high. Your app, through the Facebook SDK, an MMP SDK or your own code, updates these values as the person engages, and whatever value stands when a window closes is what gets sent.
SKAN 3, still used for some traffic, sends one postback. Each increase in the value restarts a 24-hour timer, and once the timer runs out the device sends the postback 0 to 24 hours later. SKAN 4 replaced this with three fixed conversion windows measured from first app launch: days 0 to 2, 3 to 7 and 8 to 35. Each window can produce its own postback, so one install can report up to three times.
What each postback contains depends on Apple's crowd anonymity tier, which Apple bases on crowd size and other factors: broadly, the more installs that share the same ad details, the more detail survives. In the two highest tiers, the first postback carries the fine value and a source identifier of up to four digits. Tier 1 gets only a coarse value and a two-digit identifier, and Tier 0 gets a single postback with a two-digit identifier and no conversion value. Second and third postbacks only ever carry a coarse value and a two-digit identifier.
Do the arithmetic: a first postback typically lands three to four days after first launch, the second between days 8 and 13, the third between days 36 and 41. Apps can lock a window early to get a postback sooner, but Meta advises against lock windows: they are not supported in Events Manager and can cause under-reporting and weaker performance.
- Window 1 (days 0-2): sent 24-48 hours after close, fine or coarse value
- Window 2 (days 3-7): sent 24-144 hours after close, coarse value only
- Window 3 (days 8-35): sent 24-144 hours after close, coarse value only
Which SKAN version Meta uses right now
As of November 2026, Meta's Help Center still says it is gradually introducing SKAN 4.0 support and that it may not be available to every advertiser. Ads Manager supports multiple versions: SKAN 4.0 when the ad is shown on iOS 16.1 or later inside a Meta app built with iOS SDK 16.1 or later, and an earlier version otherwise. Expect your data to mix both.
On the advertiser side, SKAN 4.0 is a toggle. In Events Manager, go to Data sources, select your app, open the Settings tab and find the setup tasks for iOS app events; in the Apple's SKAdNetwork section, click See extra settings and switch version 4.0 on. If you use the Facebook SDK, it must be version 16.2.1 or later. Meta warns that a switch in either direction can take at least 24 hours to apply and that frequent switching can hurt performance, so decide once.
Apple's newer AdAttributionKit, available from iOS 17.4, interoperates with SKAN: only one impression can win a conversion across both frameworks, and SKAN conversion value updates are mirrored into AdAttributionKit. Meta's Help Center guidance for iOS app campaigns still describes SKAdNetwork only, so ask your MMP how it handles AdAttributionKit postbacks rather than rebuilding your setup on assumption.
How Meta reports SKAN data
Meta's Help Center is explicit about the limits. SKAN has delays of at least 24 hours, and results are dated by when Meta receives them, not when the install or purchase happened. Postbacks arrive aggregated at campaign level, so Meta may use statistical modeling to fill in ad set and ad results, unless the campaign has a single ad set or ad. Creative-level reporting is not supported when an ad has more than one creative element.
Attribution settings work differently too. SKAN campaigns do not support Meta's 1-day, 7-day or 28-day click or view windows; results follow SKAN's own window, and the Compare attribution settings tool splits click-through from view-through. Ads Manager includes the first SKAN 4.0 postback with a fine value. First postbacks with only a coarse value, and all second and third postbacks, stay out of Ads Manager; Meta reports later-window coarse values through the Ads Insights API.
Three more details matter in daily work. If you use the Facebook SDK and Meta can automap coarse values, first-window events are decoded and reported as fine values. Ads Manager annotates conversions that came from Apple's SKAdNetwork API and flags results that include statistical modeling, so read the footnotes before the numbers. And A/B tests on SKAN campaigns run only at campaign level, not at ad set or ad level.
SKAN vs AEM vs your MMP
Every iOS 14+ app promotion campaign on Meta uses one of two campaign attribution methods, chosen when you build it. Aggregated Event Measurement is Meta's own: it processes events with identifiers removed, differential privacy added and data aggregated across users. Meta is gradually extending it to the app promotion objective, which lets eligible apps run app promotion campaigns without configuring SKAN events at all; you can still configure them if you want a cross-platform view.
In reporting, the two behave very differently. AEM is close to real time, uses 1-day, 7-day or 28-day click for installs and 1-day or 7-day click for app event and value optimization, and supports breakdowns such as age, gender, country, platform and placement. SKAN lags by at least a day and follows Apple's windows. In an AEM campaign you can still see SKAN side by side: open the Columns drop-down, choose Compare attribution settings, then Additional reporting options.
Your MMP sits across both. It typically receives copies of the winning SKAN postbacks for your app, whichever network won, decodes them with your schema and compares Meta with every other channel on the same rules. Since October 9, 2024, Meta has also sent AEM reporting to MMPs, with integration timing varying by partner, so check which Meta numbers your MMP dashboard is showing. When AEM and SKAN disagree in Ads Manager, Meta's Help Center lists the usual causes, most of them structural rather than signs of broken tracking.
- Delays: AEM is near real time, SKAN lags at least 24 hours
- Scope: AEM measures within Meta's technologies, SKAN across ad networks
- Windows: AEM uses 1-, 7- or 28-day click for installs, SKAN uses Apple's own windows
- Events defined differently in each setup, or not eligible for AEM
- Large gaps right after a campaign starts or stops, so allow a few days
- Low conversion volume, which inflates percentage gaps
- Your SKAN setup method differing from your primary AEM integration
Designing a conversion value schema
First decide what sets conversion values: the Facebook SDK, your MMP's SDK, or your own app code if you send events through the Conversions API or App Events API. Pick exactly one. Meta's developer documentation warns that updating values through multiple channels can cause race conditions, and that decreasing a value, which SKAN 4.0 allows, lets lower-priority events overwrite higher-priority ones and loses signal. Some MMPs offer value decrements as a feature, so confirm it is switched off. One writer, values that only move up.
Then configure it in Events Manager, which offers three routes: Use recommendations, where Meta proposes events and bundles from your last 28 days of data; Import from partner app, which pulls in the schema built on your MMP's platform; and Customize, where you set every event, bundle and value range yourself. You can configure up to 63 individual or bundled events, prioritized 1 to 63 with 63 the highest, and only the highest-priority event in a window is reported. Events without a conversion value become inactive for campaign and ad set reporting.
The design rules are mostly about timing. Put the events closest to revenue in the top slots, but do not rely on them alone: Meta notes that many lower-funnel events may not happen inside Apple's conversion window, so add upper- and mid-funnel events such as registration or tutorial completion to keep signal flowing. A bundle counts only when every event in it happens within the same window. If you only optimize for installs, Meta says no event configuration is needed.
For value optimization, turn on Maximize value of conversions and build a value set: at least four consecutive, non-overlapping purchase ranges per event and currency, each using one of the 63 slots. Meta recommends narrower ranges and considering 1-day purchase values, and for customized setups it suggests at least 8 ranges when value optimization takes 50 to 70% of your budget and at least 16 above that. For SKAN 4.0, configure coarse values too. With three or fewer fine-value events and the Facebook SDK, Meta creates coarse values automatically, as in the ladder below; a value set alone takes four or more slots, so value-optimized schemas need coarse values set by hand.
- Top priority: subscription purchase, coarse value high
- Second priority: trial started, coarse value medium
- Third priority: registration completed, coarse value low
Reading delayed, aggregated results
Because SKAN results are dated by arrival, a day in Ads Manager pairs that day's spend with installs driven by spend several days earlier. When you raise budgets, SKAN cost per install looks worse than reality for a few days; when you cut, it looks better. Read SKAN on trailing multi-day windows, compare equal periods, and never make a budget call on one day of SKAN data. Where AEM reporting is available, use it for same-day direction.
Next, know what you are not seeing. Ads Manager's SKAN numbers are mostly first-window results, roughly the first two days after first launch for SKAN 4.0 traffic. Day-7 purchases, trial conversions and other activity up to day 35 arrive only as coarse values in second and third postbacks, which Meta exposes through the Ads Insights API. Anything after day 35 is outside SKAN entirely and has to come from your MMP or your own revenue data. For subscription and gaming apps, treat SKAN ROAS as an early indicator and calibrate it against cohort data.
Privacy thresholds create gaps too. When a campaign's install volume is thin, Apple withholds conversion values and the postback reports a null. Meta's Help Center says that without at least 88 app installs per campaign per day you will probably see null conversions, and suggests combining ad sets and campaigns, raising budgets or moving up the funnel to more frequent events. Ad set and ad results in a SKAN campaign may be modeled, so make structural decisions at campaign level.
Budgeting and testing under SKAN
Everything above points one way: fewer, bigger campaigns. Meta's SKAN campaign setup guidance centers on liquidity: Advantage+ app campaigns, Advantage+ campaign budget, broad targeting, as many placements as possible and a 48-hour conversion window for faster feedback. Its Help Center adds that each iOS 14+ campaign has one optimization type, that an Advantage+ app campaign is limited to one ad set with a fixed optimization choice, and that you can create up to 24 iOS 14+ campaigns per app; past that, new campaigns stay pending until you turn one off. Confirm the cap in your account before planning structure.
Budget backwards from the thresholds. Multiply your expected cost per install by the 88 daily installs Meta cites for a rough daily floor per campaign: at a hypothetical $4 cost per install, that is about $350 a day. Meta also ties exiting the learning phase to about 50 conversions per week. If you cannot fund that across several campaigns, consolidate until you can. Scale in steps, wait a full postback cycle before judging a change, and if you use cost caps, schedule ad sets to run at least three days, as Meta advises.
Testing needs the same discipline, since SKAN A/B tests run only at campaign level and frequent edits push campaigns back into learning. Meta's consolidation guidance is to test creative in campaigns for Android or iOS 13.7 and earlier, then move proven creative into your iOS 14+ campaigns, with up to six creative options per ad set. Keep event definitions identical across AEM, SKAN and the Facebook SDK. Our mobile apps and app install playbook covers optimization events and creative in more depth.
Ad account access is the last piece. Meta asks you to check which ad accounts run app promotion campaigns in the app's Advertising settings, and each ad account needs access to the business portfolio that has permission to advertise the app. If you spend $100k+ a month across several accounts, Power Ads supplies agency ad accounts created in its own business portfolio and shared into yours via partner access, with 4% deducted from each top-up and no monthly or per-account fees. Power Ads supplies accounts only and does not build or optimize campaigns, so the SKAN schema stays with your team.
Key takeaways
- SKAdNetwork is Apple's on-device attribution framework: it credits installs to ads and returns delayed, aggregated postbacks with no user-level data.
- As of November 2026, Meta still describes SKAN 4.0 support as a gradual rollout, so check the toggle in Events Manager and expect a mix of SKAN versions.
- Ads Manager shows only first postbacks with fine values, dated by arrival; later coarse postbacks live in the Ads Insights API and your MMP.
- Use AEM for near real-time direction, SKAN as the cross-network check and your MMP to tie both to cohort revenue.
- Build one conversion value schema with a single writer, no decreasing values and no lock windows, mixing revenue events with earlier funnel events.
- Consolidate campaigns, aim for the 88 daily installs per campaign Meta cites, and judge changes on multi-day windows, never on yesterday.
FAQ
What is SKAdNetwork (SKAN)?
SKAdNetwork is Apple's privacy-preserving attribution framework for iOS app ads. The device decides which ad network gets credit for an install, then sends that network an Apple-signed postback after a delay. Postbacks contain a short source identifier and a conversion value set by the app, but no user or device identifiers, so results are aggregated and delayed rather than real time.
Does Meta support SKAN 4.0?
Yes, with caveats. As of November 2026, Meta's Help Center still describes SKAN 4.0 support as a gradual rollout that may not be available to everyone. You turn it on in Events Manager under the Apple's SKAdNetwork settings, and if you use the Facebook SDK it must be version 16.2.1 or later. SKAN 4.0 applies when ads show on iOS 16.1 or later in Meta apps built with iOS SDK 16.1 or later; other traffic uses an earlier SKAN version.
How long does SKAN data take to appear in Meta Ads Manager?
At least 24 hours, and often several days. Under SKAN 4.0 the first conversion window covers roughly the first two days after the app is first opened, and Apple adds a random 24 to 48 hour delay before sending the postback. Meta then reports results on the date it receives them, not the date of the install, so daily figures pair one day's spend with earlier installs.
Should I use AEM or SKAN for iOS app campaigns on Meta?
Meta lets you choose either as the campaign attribution method for iOS 14+ app promotion campaigns. AEM, if your app is eligible, gives near real-time reporting, 1-day, 7-day or 28-day click attribution settings and demographic breakdowns, and needs no SKAN event configuration. SKAN gives Apple-signed, cross-network data your MMP can compare across channels. You can also keep SKAN events configured and view both side by side with Compare attribution settings.
Why don't Meta's SKAN numbers match my MMP?
Some gap is normal, and most of it is structural. Ads Manager counts only first postbacks with fine values, plus coarse values Meta can decode when the Facebook SDK automaps them; it dates results by when Meta receives them and may model ad set and ad results. Your MMP decodes the raw postbacks with its own schema, may count coarse and later-window postbacks, and may use different date logic. Since October 2024 it may also show Meta's AEM numbers, which use different windows. Compare at campaign level over several days.
What is SKAN optimization?
SKAN optimization means designing your conversion value schema and campaign structure so Apple's limited signal stays useful. On Meta that means one writer for conversion values, priorities that put revenue events on top while keeping earlier events in the window, value sets for value optimization, coarse values for SKAN 4.0, no lock windows, and consolidated campaigns with enough daily installs to avoid null values.
Sources
- facebook.com/business/help/188126096313109
- facebook.com/business/help/670955636925518
- facebook.com/business/help/584603712214119
- facebook.com/business/help/1356268495231843
- facebook.com/business/help/3499572073626051
- facebook.com/business/help/721422165168355
- facebook.com/business/help/331612538028890
- facebook.com/business/help/2750680505215705
- facebook.com/business/help/332131285040809
- facebook.com/business/help/193250612476055
- facebook.com/business/help/288812062988970
- facebook.com/business/help/378857770047703
- facebook.com/business/help/595608511524017
- facebook.com/business/help/581721639581617
- facebook.com/business/help/651033805513936
- facebook.com/business/help/1329822420714248
- developers.facebook.com/docs/app-events/skanv4.0
- developer.apple.com/documentation/storekit/receiving-postbacks-in-multiple-conversion-windows
- developer.apple.com/documentation/storekit/receiving-ad-attributions-and-postbacks
- developer.apple.com/documentation/adattributionkit/adattributionkit-skadnetwork-interoperability
