December 11, 2025
When Apple introduced App Tracking Transparency (ATT), the industry expected disruption. What few anticipated was just how dramatically Apple’s own design choices would reshape the advertising landscape. iOS campaigns became harder to optimize, attribution became unstable, and much of the deterministic signal that powered performance marketing evaporated overnight.
For many teams, this raised a practical, unavoidable question:
Is relying on SKAN enough — or do we need third-party measurement to regain clarity?
A new large-scale study involving more than 11,000 iPhone users finally gives us data that helps answer that question. And the answer lies in what we can call the consent gap.
The study tested two different consent prompts:
The result was stark. Apple’s softer language — “Turn On Personalized Ads?” — produced almost double the consent rate of the ATT prompt, even though both prompts describe essentially the same type of data use.
If your current iOS consent rate is around 15%, you’re not underperforming; you’re sitting exactly where the system is engineered to push you. This is the heart of the consent gap: the distance between what users might agree to in a neutral environment and what they choose when Apple frames the question negatively.
Another striking finding from the study is that even people who say they like personalized ads become far less likely to opt in when confronted with the ATT prompt. The prompt itself suppresses consent by an additional 15.1 points among this group.
And the language of “tracking” does more than discourage — it distorts understanding. Users shown the ATT prompt were:
None of that is true for standard advertising use cases, yet the prompt consistently steers users toward these misconceptions. The result is a consent rate that reflects fear, not preference — and this misunderstanding disproportionately affects third-party advertisers, not Apple.
This gap is not just a UX issue; it’s a measurement problem.
With only 13–15% of users consenting, the overwhelming majority of your iOS traffic is invisible to traditional MMPs (AppsFlyer, Adjust, Branch). Instead, all measurement collapses onto SKAN’s aggregated, delayed, privacy-preserving postbacks. SKAN becomes your only source of truth — not because it is the best tool, but because the consent gap starves everything else.
SKAN can be enough in certain circumstances. If your campaigns optimize toward broad objectives, operate on large budgets, and don’t depend on early-stage LTV modelling, SKAN’s population-level signals may be adequate. But for subscription businesses, behaviour-driven funnels, or any environment where creative, audience, and event-level nuance drives performance, SKAN alone simply doesn’t provide enough visibility.
This isn’t SKAN’s failure. It’s the predictable consequence of the consent gap.
For some marketers, SKAN provides all the signal needed. For others, it leaves major blind spots. The dividing line is shaped almost entirely by consent volume.
SKAN tends to be sufficient when:
SKAN falls short when:
Most fitness and wellness apps fall into the second category. Your product needs early signals to tune its recommendation engine and campaign structure. And that is precisely what disappears when consent sits at 15%.
This study also shows what’s possible if you improve consent, even modestly. Apple demonstrates that a more neutral or positive prompt can yield 25% opt-in — nearly double the current industry norm.
Even if you don’t reach Apple’s 25%, moving from 15% to 20% has measurable impact:
A move from 15% to 25% produces an even sharper transformation, delivering nearly 66% more signal, stabilizing performance, and restoring much of the optimization intelligence that ATT restricted.
The point is simple:
The consent gap isn’t fixed — and closing it is one of the few remaining levers on iOS.
The study makes one thing clear: Apple’s prompt design is the primary reason third-party consent is so low. Low consent isn’t evidence of user rejection, distrust, or weak value exchange. It is the direct result of the system’s framing.
If your consent remains at 13–15%, SKAN will inevitably be your dominant measurement tool. Whether SKAN is “enough” depends on your business model and how much precision you need.
If you can improve consent into the 20–25% range, however, SKAN becomes just one part of a healthier ecosystem. You regain deterministic event streams, MMPs become meaningful again, LTV modelling stabilizes, and campaign optimization becomes less guesswork and more engineering.
The strategic takeaway is simple:
SKAN isn’t the problem. The consent gap is.
And the marketers who learn to close that gap take back performance, signal, and control.
Source: https://www.bu.edu/dbi/files/2024/09/ssrn-4887872-ATT.pdf