“Rizz” may be the word of 2023, but for us, app marketers, it has always been SKAN. And we believe Apple SKAdNetwork will also be the top term in 2024.
This is no surprise, since SKAN remains a mobile attribution lifeblood to those running ad campaigns on iOS. Being a sophisticated framework, SKAdNetwork has no intention of resting on its laurels and, therefore, keeps evolving.
Right now the mobile industry is still adopting its 4th edition, while a better, newer version is already on the way. During WWDC23 Apple announced the arrival of SKAdNetwork 5.0. The exact date of its release is still unknown, but anyway, it’s one more reason to learn more, isn’t it?
Our article provides an in-depth exploration of SKAdNetwork, its benefits for publishers and advertisers, the details of its functioning, as well as its limitations and challenges.
SKAdNetwork (StoreKit Ad Network) is a framework, a set of software and protocols created by Apple for privacy-safe mobile attribution for user acquisition campaigns. SKAdNetwork, or SKAN, reports deterministic aggregated attribution data for mobile app marketing campaigns. That means marketers get insights into their marketing in a privacy-safe manner: they don’t get device-level data on what individual users are doing.
Let’s imagine you have a favorite game on your iPad, and sometimes you see ads for new apps. Intrigued, you download the advertised app, and this is where SKAdNetwork enters the game.
In simple words, SKAdNetwork tells the app developers if you decided to get a new game after seeing their ad. But it doesn’t tell them anything about you personally — like your name or where you live.
Professionally speaking, SKAdNetwork, also known as StoreKit Ad Network or SKAN, is a type of direct install attribution that dominates the App Store. The framework has a privacy-centric approach at its core: it reduces developers’ access to user-level or device-specific data.
Apple introduced this framework with iOS14 in 2018. Its emergence fluttered the industry: app publishers had to adjust to new privacy-preserving terms in order not to jeopardize their monetization efforts.
They weren’t the only ones who embraced the change. Mobile measurement partners had to adapt as well, introducing their solutions (for example, AppsFlyer SKAdNetwork and Singular SKAdNetwork integrations) and updating SDKs. These tools are aimed at helping you verify conversion data and have lots of useful features (predictive analytics, customizable reports, fraud protection, etc.).
As it was mentioned above, Apple SKAdNetwork is a framework that measures conversion rates without disclosing the identity of users. It detects two types of user engagement, views and StoreKit renders. Now, Let’s figure out how SKAdNetwork functions.
We’ll start with listing the elements, or, let us say, participants of the ad attribution process. There are five of them:
Basically, ad networks sign up with Apple to get a special ad network ID and, of course, use the API — while publishers set up their apps to show ads from these networks.
These are the main players. Time to see how they interact with each other.
To truly comprehend the inner workings of SKAdNetwork, it’s imperative to deconstruct its flow.
Here is the description of the journey ad impressions go through to secure a successful ad attribution:
In essence, the postback system in SKAdNetwork serves as the bridge that connects ad networks, app developers, and advertisers, facilitating the attribution of app installs and providing data for campaign optimization, all while prioritizing user privacy.
Please note there can be a considerable attribution window between tapping an ad and an install. It can take up to 30 days.
If an app publisher uses a third-party MMP, the latter can get postbacks directly if the ad network chooses them as an endpoint. Alternatively, the postback information can be shared with an MMP of your choosing through dedicated dashboards and APIs.
User privacy is at the heart of everything Apple does. SKAdNetwork is all about that — ensuring that users’ sensitive data stays secure.
There is a specific data threshold, or tier that’s used by the device to understand how much user information the postback can include. SKAN 4.0 divided this tier into crowd anonymity tiers.
The postback data tier (0, 1, 2, and 3) depends on the crowd size linked to the following elements:
Apple designates a tier for each install. This move helps decide on which data should be limited in the postbacks sent by SKAN. Here’s the list of affected postback fields for SKAN 4.0:
The source identifier (SID) is a unique four-digit code that helps you recognize the ad campaign to which the app install is attributed. It can be found only in SKAdNetwork 4.0. Previous versions of the framework had the campaign identifier instead.
With the source identifier, you can see which of the ad campaigns is the most successful one.
The conversion value is a solitary number transmitted by a recently installed app to let the ad network know that an install has been attributed to it. There are two types of the conversion value, fine and coarse.
Fine conversion values are a 6-bit value, providing up to 64 combinations (0 to 63). These values aim to gather insights into user behavior.
Coarse conversion values come into play when the conversion volume is not sufficient to meet the privacy threshold for fine conversion values. They can be categorized as low, medium, or high, serving as aggregated buckets for predefined CV values. And, optionally, the conversion value can also be null.
The source app ID is a unique identifier for the app that showed the ad — while the source domain identifies the domain displaying a web ad.
The adoption of SKAdNetwork in the mobile advertising ecosystem offers several benefits for both advertisers and app publishers. Here are key advantages of using SKAN for the parties involved.
Take a look at the advantages that SKAdNetwork offers to app publishers:
And what about advertisers? Well, for them, SKAdNetwork can be useful in a number of ways:
Ultimately, what the industry knows and has experienced with SKAN is that it limits the way marketers are used to doing their jobs. Historically SKAN obfuscated lots of data with privacy thresholds for those that didn’t properly scale, and SKAN severely limited marketers’ insights past the first 24-hour window after install. But between SKAN 4’s improvements and Singular’s Advanced Analytics and cohorted SKAN data, we believe SKAN reporting in 2024 will drastically improve and be much more familiar to use
While SKAdNetwork is an effective framework, it has some limitations meaningful for app publishers and marketers, some of them include:
|For SKAN 2.0 and 3.0, segmented data is presented only at the campaign level, with only 100 campaigns available. SKAdNetwork 4.0, however, offers a significant improvement in granularity by enabling you to set up 10,000 campaigns.
|SKAdNetwork provides less granular data compared to traditional attribution models. The use of conversion values, while privacy-centric, may limit the depth of insights available for optimizing ad campaigns.
|Room for fraud
|Since attribution takes place within SKAdNetwork, fraudsters will focus their activities here.
|Data manipulation results in poor performance and financial losses.
|Limited analytics on ROI/LTV/ROAS
|Even though since the SKAdNetwork 4.0 launch advertisers got more information, installs, conversions, and post-install events are still measured in a very restrictive way.
|The limited data available in SKAdNetwork postbacks may pose challenges for advertisers in optimizing their campaigns effectively. Balancing campaign performance with privacy considerations requires innovative strategies.
|A minimum postback delay of 24 hours.
|With such a delay, it’s hard to optimize ad campaigns promptly.
|Absence of re-engagement attribution support
|No in-built tools and analytics.
|There is no help from SKAN in returning churned users.
The limitations above, naturally, pave the way for certain challenges for app marketers.
These challenges even caused a peculiar reaction in the mobile industry — the iOS ad spend dropped when iOS14.5 adoption hit the mainstream in 2021 (noteworthy, Android ad spend was showing steady growth back then), reports Singular.
The best way to address these challenges is to partner with a mobile measurement partner that cooperates with major ad networks (Facebook, Twitter, Google Search Ads, Snapchat, and Tiktok). A good MMP will also ensure that there is no fraud affecting your data.
With SKAN 4 we’re given more data, more signals early on, and more late funnel data which is huge. Adding coarse values to the first postback will provide early indicators for marketers even if they didn’t fully reach the volume required to pass crowd anonymity. Plus, with two more postbacks, SKAN will provide directional insights for what’s coming later in the funnel which wasn’t available previously. This is game-changing because it will allow MMPs like Singular to model cohorted data up to D30.
During WWDC22, it was revealed that SKAdNetwork is set to undergo modifications with the upcoming 4.0 version. The official rollout took place on October 24, 2022. Right now the industry is getting accustomed to this SKAN edition.
Here’s a brief overview of the main SKAdNetwork 4.0 features.
Campaign IDs were replaced by hierarchical source IDs, allowing up to four digits in the postback. The digits are determined by the number of installs and the user privacy level for a specific campaign, centered around the concept of “Crowd Anonymity.”
Crowd anonymity in SKAN 4 is defined by the quantity of installs generated from the marketing campaign.
SKAdNetwork 4.0 receives up to three postbacks for specific conversion windows: 0-2 days, 3-7 days, and 8-35 days.
The framework offers both fine-grained and coarse-grained conversion values. The fine-grained option provides detailed data, while the coarse-grained version is designed for campaigns with a lower number of app installs, offering limited data.
SKAdNetwork 4.0 brings about the attribution of web-based ads, expanding its capabilities beyond app-based attribution.
These modifications aim to enhance privacy, provide more nuanced attribution options, and extend SKAdNetwork’s applicability to web-based advertising scenarios.
It’s a little bit early to speak about SKAdNetwork 5.0 features as the version hasn’t been released yet.
However, some details about the framework have been circulating for quite a time, so here are some assumptions we can make about this edition:
Other details and features will be unveiled later.
There is one.
It’s the IDFA itself.
In fact, SKAN was created as an alternative to the IDFA.
Here’s a quick reminder of what the IDFA is.
Apple’s Identifier for Advertisers (IDFA) is a unique, persistent identifier assigned to each Apple device. The purpose of IDFA is to enable advertisers and developers to track user activity across apps and websites to deliver targeted advertisements. Advertisers also have to seek user consent through a prompt to get access to the IDFA since the introduction of iOS 14.5 and App Tracking Transparency (ATT).
To help you understand the differences between the IDFA and SKAdNetwork attribution models, we’ve prepared this short comparison table:
|Matching and Granularity
|1 to 1 matching with full granularity on various campaign variables.
|Attribution is group-based, associated with a unique ID, offering cluster-level insights.
|1. User clicks on an ad, and multiple variables are recorded.2. MMP records IDFA data.3. User installs the app. 4. MMP matches and attributes the conversion.
|1. User clicks on an ad, and an ID with up to 100 values is tracked. 2. User installs the app, triggering a 24-hour timer. 3. Postback is sent if there are no updates in 24 hours. 4. DSP forwards postback without IDFA or user-level data.
|IDFA Postback includes: Device ID (IDFA), Transaction ID, Additional device data.
Contents depend on the version of SKAdNetwork you use. Later versions offer more sophisticated options.
A SKAdNetwork 4.0 postback includes: Source Identifier; Conversion Value; Source App ID; Source Domain.
For beginners, it’s only natural to wonder about a possible Apple SearchAds-SKAdNetwork connection.
However, the only thing that unites them is the name of their inventor, Apple. They can work alongside each other, but not together. In other words, Apple Search Ads and SKAdNetwork don’t depend on each other. They’re separate entities.
The Apple Search Ads attribution is deterministic and has nothing to do with SKAN. The channel takes a token from a device to measure the performance of an ad campaign (takes a token from a device), and ATT has no power over Apple Search Ads attribution.
Attribution applies only to users using iOS 10 or later. Apple Search Ads promptly reports a download or redownload as it happens and continues to do so for up to 30 days. An install (or re-install) must occur either from an App Store listing or as a result of an Apple Search Ads impression for attribution to be valid.
For enhanced reporting and optimization concerning your Apple Search Ads efforts, consider using third-party tools such as SplitMetrics Acquire. With this platform, you’ll be able attribute Apple Search Ads data in one Ads Manager dashboard without copy-pasting endless tables.
SKAdNetwork is a critical tool in the mobile advertising ecosystem, allowing app marketers to measure and optimize campaigns for iOS apps while respecting user privacy.
Both advertisers and app developers need to carefully study the framework and navigate its intricacies, leveraging its benefits while addressing its limitations to ensure effective mobile attribution.
In that regard, the best course of action here is to stay informed and follow the news about future SKAdNetwork editions.
As they say, knowledge is power.