Apps are big business, and measuring what’s really driving their growth has been a big headache for a long time.

As an investor in or operator of a company in the app space, you’re probably used to these pains, but do you know why they so regularly recur?

Well, one source of pain you may have gotten familiar with is the subtly shifting measurement protocols that govern privacy regulation compliant mobile device and user tracking.

On the Apple side, it’s happening again!

Instead of a new version of SKAdNetwork (SKAN), that terminology is being pushed to the side in favor talking about the “AAK,” the AdAttributionKit. Will this be the end of your mobile growth attribution headaches, or the start of entirely new ones?

Dr. Daypart is here to answer your questions and prescribe a preventative tonic…after some relevant recent updates!

Apple’s StoreKit Ad Network (SKAN) will remain the basis of their mobile ad privacy framework, but the more sophisticated product suite it undergirds, the Ad AttributionKit (AAK), will come to the forefront.

This is not a seismic change-some product updates to AAK will better serve advertisers interested in measuring their re-engagement and deep linking campaigns. There will be more attribution customization options.

That’s basically it-mobile ad measurement challenges on Apple aren’t going anywhere soon, though, and so it would behoove our dear users to read the details on where they stand.

At the very least, this could help you resist any anxiety others may try and instill in you around these changes, and maybe to inoculate your colleagues with your

Attribution is the act of taking a valuable action that occurred, and assigning credit for it to some kind of marketing activity, or to positive product word of mouth, or to pure chance…really to anything, you just want to know why something happened.

It’s very important! As our recent podcast guest Eric Tillbury said, “the difference between a million dollar ad tech company and a billion dollar ad tech company is a tracking URL.”

We’ll come back to what a tracking URL is later, but long story short: it’s important for attribution.

Attribution has two fundamental halves that determine the value it creates: its ability to detect value actions taken by a user, and its ability to hand out credit “accurately.”

App growth campaigns have arguably been more interested in attribution than most advertising initiatives for a long time, due to the significant amount data they gather on user value relative to other industries.

Think about a rideshare app, and how many data points it can gather on a customer before they even pay for their first ride, and how much information on long term customer value it can gather on a user over years, or even decades.

That’s one of the pinnacle examples of a rich user data profile, but any phone app angling to be utilized on a weekly or monthly basis by anyone who downloads it generally builds quite the data profile.

In addition to the data naturally generated by users, apps tend to have an incredible ability to run experiments with their user interface, or other customer facing elements of their service, at scale and speed.

With all this value piled up on the “observed user actions” side, you can see how incredibly powerful this whole “attribution” thing could be with equally rich and high-fidelity signal environment to properly determine what drove a user to take that first key action with the app.

Right here is the grand joke about mobile attribution-because this data can reveal so much about users so quickly, many parties have privacy concerns specific to mobile app tracking.

For their part, Apple has introduced multiple privacy measures, albeit possibly to diminish the value of their Silicon Valley rivals’ ad products-a theory alleged in regulatory proceedings but not yet found as a matter of intent by any court or competition authority-and also to enhance the value of their own ad products.

Both Apple and Google assign each device a unique, resettable advertising identifier usable for advertisers-Apple’s identifier is known as the Advertising Identifier, or IDFA. Android Devices have Google Advertising Identifiers, or GAIDs.

As of the publication of this article on July 5th, 2026, Google do not require any kind of user opt-in for GAID based tracking outside the EEK, UK, and Switzerland.

GAIDs are used by mobile ad products and mobile measurement providers to track user-level behavior as it pertains to interacting with advertisements, and subsequently performing valuable actions within advertiser apps.

While these IDs are anonymized, there are a number of ways they could be used that have rankled privacy advocates.

While users can easily reset their GAIDs on their phone at any time, the user is not prompted to do so at any time, and there are no standard education or information modules users are proactively presented with explaining what GAID is and why they might ever want to reset it.

When it isn’t reset for a long time-let’s say 3-4 years, the rough timespan over which a user will keep a smartphone in the US-a lot of activity can accrue that can paint a very clear picture of a user’s behaviors and, possibly, elements of their identity.

A persistent mobile ID also makes it easier to build cross-device IDs for users, through device fingerprinting or other methods, by doing the entire job of consistently identifying a user’s smartphone with one simple parameter.

For now, this is how Android app user tracking works, and it keeps things relatively simple on that side, in contrast with the mobile world Apple hath wrought over the last several years.

Google did have plans akin to Apple’s when it came to the Android mobile environment, specifically a major project called the Android Privacy Sandbox. However, in tandem with Google’s announcement that their long-threatened default blocking of third party cookies in Chrome wasn’t happening, they also stopped work on all of the projects that comprised the Android Privacy Sandbox.

SKAN has a long history, but when it first came out back in 2018, it landed about like Android Privacy Sandbox did.

When Apple first introduced the mouthful that was StoreKit Ad Network, without any serious benefit or consequence of adopting or not adopting the framework, it was a fairly minor concern for mobile apps and advertisers.

Two things brought it to the forefront. One was that at some point people decided to start abbreviating it as SKAN, which made it easier to type and talk about.

The second, arguably more important reason was Apple gave everyone a reason to care about SKAN: it announced its App Tracking Transparency (ATT) policy. This is a framework in which an app asks a user for permission to track them.

If the user opts out, their IDFA (the Apple unique advertising identifier) will not be accessible in that app, and so they will not be deterministically tracked at the user level.

Shortly after, we got the little number that broke the hearts and minds of a generation of mobile advertisers: 14.5

This iOs release implemented ATT requirements for every app, which meant that every single user was either going to be prompted to opt in or out of tracking in every app, unless they had already enabled blanket anti-tracking settings for their device, which is automatically enabled in cases such as child accounts.

Even in the year 2025, where the numbers are up after years of apps getting better at asking users nicely, global ATT opt-in rates are around 35%, which means roughly 65% of the time, iOs users are not letting apps use their deterministic identifier, though opt-out rates vary widely by app category, region, demographic, and other factors.

For good measure, to kick tracking while it’s down, Apple introduced Intelligent Tracking Prevention, a Safari feature that masked a user’s IP address and inhibited common methods to otherwise track a user across the internet.

How could anyone complain, though? They gave the world SKAN back in 2018, and SKAN would be our salvation! Well, in theory, for a time.

SKAdNetwork utilizes several core mechanism to preserve user privacy:

The particulars of how much data one could get due to Crowd Anonymity is a whole complicated subject, but the short version is: advertisers needed to manage their ad campaign, test design, and measurement setups to make sure they could get enough SKAN data on any given bucket of ads.

Adding to that trouble, the delayed post back system was a bit of a doozie. You could set an attribution window of up to 30 days to record in-app actions as part of your SKAN data.

However, the way it worked is that none of this data would be sent until either the end of the 30 day attribution window, or after 24 hours of a given user not generating any conversions in your app.

For example-if a user downloads your app and starts taking valuable actions immediately, but does not stop taking valuable actions for more than 24 hours in the first 30 days of having the app, and you have a 30 day attribution window set-you would not get any of that post-install conversion data in your SKAN reporting for over 30 days.

Later versions of SKAN introduced multiple post back windows, where the advertiser could get up to 3 post backs for days 0-2, 3-7, and 8-35 which alleviated things a bit.

Another major drawback of SKAN versus standard attribution systems was its inability to be used in the measurement of re-engagement campaigns, as it was built only to record results of the initial app installation.

Advertisers not only had to adjust to these changes, but also were still able to take advantage of a tracking system on the Android side that was free to use deterministic IDs without consumer opt in within most regions, and provided much more granular, near-realtime data.

The ad and measurement tech ecosystem adapted, through SKAN integration, probabilistic modeling, and accessible MMM tooling, reducing but not eliminating the disruption for advertisers.

Features to help parse SKAN data became big chunks of mobile measurement platforms:

Even the mighty Meta built in modals to help the smallest of advertisers understand where they stood with SKAN and the granularity available to them:

Large advertisers running sophisticated measurement programs had to consider and manage SKAN for any internal data initiatives and would need to build SKAN into any custom measurement tech they utilized, but for many advertisers, they eventually let the major ad platforms and Mobile Measurement Partners(MMPs) handle it for them.

The reader may have heard of Mobile Measurement Partners, such as AppsFlyer and Kochava. These are technology platforms meant to enable and enhance marketers’ ability to measure the impact of campaigns and initiatives on the growth of their apps.

Speaking of major technology platforms handling this issue-the release of iOs 14.5 caused a whole paradigm shift in terms of how ad platforms managed measurement entirely, notably at Meta.

Apple’s ATT is what accelerated Meta’s use of modeling and probabilistically reporting conversions, in tandem with their drive to vastly expand advertiser adoption of their Conversions API (CAPI), a server-side event data connection tool that also helped their efforts to battle third party cookie blocking on the web side.

Arguably, Meta emerged from this with a stronger competitive position versus smaller advertising platforms that do not have user base and engineering resources to put solutions of this size together.

That being said, these measurement changes arguably incentivize all parties to develop better capacity around probabilistic conversions and alternative methods of user tracking like device fingerprinting, which may or may not have user consent in all cases and could conflict with privacy laws in some jurisdictions.

AAK, launched around the same time time as SKAN, largely just builds on top of things that were already “working” for the old privacy systems.

The conversion reporting system remains largely the same, with Crowd Anonymity governing data granularity available, and the same limited number of total postbacks from the same post-install day windows as SKAN.

Firstly, there are more ways to display ads and record impressions. SKAN was highly primitive in this regard, recording only the initial impression, and then sending a signal if the ad remained open for 3 seconds for a “view.” AdAttributionKit allows use of the iOs design language to more carefully define events like “clicks” and “views” in an ad.

Secondly, AAK has a “conversion type” that will allow advertisers to provide a deep link for re-engagement and separate tracking for these campaigns, arguably the largest gap that existed in the old SKAN framework.

A re-engagement campaign is an effort to bring lapsed, or low value users, back to an app or a specific part of an app. Deep links are a way to send them directly to a specific screen in the app, such as a new feature you are trying to get them to adopt, or an offer to incentivize them to return to regular app usage.

The development of AAK makes it clear how important the re-engagement measurement capability is-one of the major new items in last year’s AAK developer update was specifically for conversion windows on re-engagement campaigns. One of the other major releases was more support for customizing attribution rules.

Apple basically kept the core privacy preservation elements of SKAN, its Crowd Anonymity and carefully controlled postbacks, and is advancing the framework that will give advertisers a few things they want that never seemed to really conflict with those privacy needs.

This also means that The Tracking Wars, and its winners and losers, are fundamentally unchanged.

MMPs can offer the same handling-this-as-a-service, Meta can keep modeling conversions, Google can keep not imposing this on Android, and smaller parties can keep working scrappy (and hopefully not extralegal) solutions.

As an attorney, if your client is using any mobile data or technology that claims to help with Apple app user tracking degradation, be sure to have someone dig into the details-the move from SKAN to AAK hasn’t really changed how the anonymization works at its core, so beware any solutions that sounds to good to be true.

As an investor, don’t let your portfolio’s growth teams fall into the hole of over-focus on Android user acquisition if it continues to simply be more measurable and looks a little better on paper.

Similarly, don’t let them go overboard on re-engagement efforts with questionable incremental revenue contributions just because the measurement capabilities for these campaigns got better on iOs.