Full user-level iOS tracking on TikTok still only works for the share of users who opt in to Apple’s App Tracking Transparency (ATT) prompt. Disclosed benchmark sources currently put that share anywhere from about 35% to about 70%, depending on methodology.
For everyone else, TikTok fills the gap with SKAdNetwork (SKAN) aggregated postbacks, iOS real-time conversion reporting through a Mobile Measurement Partner (MMP), and modeled conversions.
None of these fully replaces device-level tracking. Combined, they still give you enough signal to run and optimize campaigns, which is why most advertisers now build a layered setup instead of chasing one fix.
This guide covers what ATT actually restricts, what current opt-in data shows (including where two major benchmark sources disagree), which TikTok-specific tracking tools still work, and how to set them up.
Figures below reflect TikTok’s documentation and third-party benchmark data confirmed live as of September 2026. Both change often enough to recheck every few months.
Table of Contents
Get Up to $6,000 in Free TikTok Ad Credits
TikTok offers ad credit incentives for new advertisers, helping you test campaigns with a lower upfront cost.
Key takeaways
- Two disclosed-methodology sources report very different ATT opt-in numbers for essentially the same measure (the acceptance rate among users shown the prompt): Adjust’s Q2 2025 data puts it at 35%, up from 34.5% in Q2 2024, while Kochava Foundry’s 2026 survey puts it at 69.7%. Treat this as a real range, not a single settled figure.
- TikTok supports iOS real-time conversion reporting through seven Mobile Measurement Partners (AppsFlyer, Adjust, Airbridge, Branch, Kochava, Singular, Tenjin) and its own App Events SDK version 1.5.0+, confirmed in TikTok’s own documentation, last updated August 2026.
- SKAN 4.0, TikTok’s default privacy-safe fallback for non-consented users, is only now reaching majority adoption industry-wide, six years after ATT launched, according to Kochava Foundry’s 2026 survey of iOS app marketers.
- SKAN’s crowd anonymity thresholds mean lower-volume campaigns can get little to no postback data at all, a real limitation Kochava Foundry’s research specifically calls out as hitting mid-tier app developers hardest.
- Apple’s next-generation framework, AdAttributionKit (AAK), has been available since March 2024 but shows negligible adoption as of 2026, so it’s not yet a setup worth building around.
- No single method covers all your iOS traffic. The realistic setup layers consented device-level data, SKAN aggregated data, and modeled conversions, then treats each with the right level of confidence.
What ATT actually restricts
Apple’s App Tracking Transparency framework, enforced since iOS 14.5 in April 2021, requires apps to ask permission before tracking a user’s activity across other companies’ apps and websites using their device identifier (IDFA).
If a user declines or is never asked, TikTok can’t attach that person’s ad click to what they do inside your app the way it could before ATT. Everything else in iOS measurement since 2021 exists to work around that one restriction.
What ATT opt-in data actually shows in 2026
There’s no single, agreed-upon ATT opt-in figure right now, even among sources that disclose their own methodology, and it’s worth knowing that before you anchor a strategy to any one number.
Adjust’s data, drawn from its own measurement platform across app categories, puts the industry-wide opt-in rate (measured among users shown the prompt) at 35% as of Q2 2025, per Adjust’s ATT opt-in rate benchmarks. That’s up slightly from 34.5% in Q2 2024 and 34% in Q2 2023.
It also varies heavily by category. Sports apps lead at 50% and hyper casual gaming sits at 43%, while education apps have climbed from a 2023 low of 7% to 14% in 2025.
Kochava Foundry’s 2026 survey of iOS app marketers puts the same basic measure, the share of users who accept when shown the ATT prompt, at 69.7%, according to its 2026 iOS attribution research. That’s climbed from about 37% at launch, past the halfway mark in September 2024.
These two disclosed-methodology sources measure a similar concept but land nearly 35 points apart. Likely drivers include different app samples and survey periods, Adjust’s multi-year vertical averaging versus Kochava’s single 2026 marketer survey.
Neither source publishes enough overlap with the other to fully reconcile the gap.
Use both as directional benchmarks, not a single settled truth, and prioritize your own app’s actual data once you have it.
Kochava’s mid-2026 attribution data adds separate context on where iOS measurement volume actually flows. Roughly 10% of iOS attributions are consented, device-level data, and 11% flow through SKAN.
Another 36% flow through Apple Ads, predominantly brands bidding on their own brand terms, with the remainder split across other aggregate and modeled approaches.
A rising opt-in or acceptance rate doesn’t automatically raise your own consented population if your app isn’t prompting users to begin with. That’s a separate, app-side decision, disconnected from how well Apple’s prompt performs once it appears.
What still works for TikTok advertisers
Consented, device-level tracking
For whatever share of your users opt in, TikTok still gets full event-level attribution the way it did before ATT. Treat the 35-70% range above as a rough industry reference, not a specific prediction for your app.
This population is small relative to your full iOS traffic, but it’s genuinely deterministic, not estimated. That makes it useful as a seed for extrapolating patterns to your non-consented traffic.
TikAdTools’ broader guide to TikTok’s attribution model covers the same principle: treat any confirmed data source as a baseline for the modeled estimates covering the rest of your traffic.
iOS real-time conversion reporting through an MMP
TikTok’s iOS real-time conversion reporting delivers near-real-time performance data at the campaign, ad group, and ad level for iOS 14.5+ campaigns. It supports goal-based bidding strategies like Target Cost per Result and Target ROAS.
TikTok’s documentation, last updated August 2026, confirms this is generally available through seven MMPs: AppsFlyer, Adjust, Airbridge, Branch, Kochava, Singular, and Tenjin, plus TikTok’s own App Events SDK (version 1.5.0 or higher).
The App Events API alone does not support this specific feature yet, so an API-only setup stays limited to standard, non-real-time reporting. That’s a distinction worth knowing if you’re also weighing the TikTok Conversion API for server-side tracking.
Each MMP needs its own configuration to unlock this. AppsFlyer needs Advanced Data Sharing (ADS) enabled, and Adjust needs a configured probabilistic modeling attribution window. The setup isn’t automatic just because you’re connected to one of these partners.
Confirming your base TikTok Pixel or SDK implementation is solid first makes this whole stack more reliable. TikAdTools’ Pixel Setup Helper walks through that base setup before you layer MMP-specific configuration on top of it.
SKAdNetwork (SKAN), TikTok’s default fallback
For non-consented users, TikTok’s default measurement path is SKAN, Apple’s privacy-preserving framework. It reports aggregated, delayed conversion data rather than individual events.
SKAN 4.0, released October 2022, added three conversion windows and up to five postbacks per install.
Per Kochava Foundry’s 2026 research, SKAN 4 is only now reaching majority adoption across the industry, six years into ATT. The bottleneck has consistently been publisher-side implementation, not advertiser demand.
If you’ve reached TikTok’s SKAN campaign limit and your app already qualifies for real-time reporting, TikTok lets you turn off SKAN attribution to launch more iOS campaigns.
Doing so means you lose SKAN data in both TikTok Ads Manager and your MMP reporting, a real tradeoff, not a free unlock.
AdAttributionKit (AAK): not yet worth building around
Apple’s next-generation replacement for SKAN, AdAttributionKit, has been available since March 2024 and adds capabilities SKAN never had, including reengagement attribution and stronger anti-fraud measures.
Kochava Foundry’s 2026 survey found adoption remains negligible despite that. A meaningful share of surveyed marketers hadn’t even heard of it, and very few had an active testing plan.
Publisher-side adoption has historically been the bottleneck at every past SKAN version transition, and the same pattern is repeating with AAK. It’s worth watching, not worth prioritizing yet.
Modeled conversions
Where neither consented tracking nor SKAN data is available or sufficient, TikTok and its MMP partners fill the gap with modeled, statistically estimated conversions.
The general industry approach is probabilistic modeling calibrated against whatever consented and SKAN data does exist, though TikTok doesn’t publish its own exact modeling methodology. Treat these figures as directional, an estimate rather than a measurement.
A realistic 2026 iOS measurement stack
- Implement an ATT prompt if you haven’t. A meaningful share of marketers still hadn’t built a custom pre-prompt experience as of Kochava Foundry’s 2026 findings, relying only on Apple’s bare default dialog. A well-designed pre-prompt that explains the value exchange first is standard practice now, not a growth hack.
- Check your app’s eligibility for real-time reporting directly in TikTok Ads Manager. When you create an iOS 14 dedicated campaign, TikTok flags whether your app has sufficient data signals for real-time conversions. Don’t assume eligibility either way without checking.
- Connect an MMP that supports TikTok’s iOS real-time reporting (AppsFlyer, Adjust, Airbridge, Branch, Kochava, Singular, or Tenjin), then complete that MMP’s specific configuration step, not just the basic integration. TikAdTools’ Pixel setup guide for ecommerce covers the parallel web-side setup if you’re running both.
- Confirm your SKAN setup is actually correct, not just present. A third of the marketers Kochava Foundry surveyed hadn’t implemented SKAN at all, and another meaningful share had it in place without being sure how it was configured.
- Don’t expect SKAN postbacks on low-volume campaigns. Crowd anonymity thresholds mean a campaign without enough conversion volume may return little or no SKAN data. Budget your test size accordingly, especially for niche or early-stage apps.
- Treat your consented population as a sample, not your whole picture. Use it to sanity-check what your modeled and SKAN data are telling you about the rest of your iOS traffic, rather than optimizing purely on the smaller consented slice.
- Recheck this setup periodically. SKAN version adoption, AAK adoption, and TikTok’s supported MMP list all continue to shift, and the two opt-in benchmarks above may converge or diverge further.
Common mistakes to avoid
Treating one opt-in benchmark as settled fact is a frequent error. Adjust and Kochava Foundry, both disclosed-methodology sources, currently disagree by roughly 35 percentage points on essentially the same measure.
Anchor decisions to your own data once you have it, not to whichever published number you happened to see first.
Assuming SKAN and real-time MMP reporting are interchangeable is another mistake. SKAN covers non-consented users through Apple’s framework, while real-time reporting through an MMP covers users where enough signal exists to model with confidence.
TikTok’s Ads Manager reports these two paths separately for exactly that reason.
Ignoring MMP-specific configuration steps after connecting one is a common gap. Being connected to AppsFlyer or Adjust doesn’t automatically enable TikTok’s real-time reporting, each partner has its own toggle that has to be turned on separately.
Expecting SKAN to always return data is a mistake that hits smaller and mid-tier apps hardest. Crowd anonymity thresholds withhold postback data below certain volume levels, so a low-spend campaign can come back with sparse or no SKAN reporting even when it’s configured correctly.
Frequently asked questions
Does ATT tracking still work at all for TikTok ads in 2026?
Yes, for the share of iOS users who opt in, though disclosed sources disagree on exactly how large that share is, 35% per Adjust’s Q2 2025 data, 69.7% per Kochava Foundry’s 2026 survey. For everyone else, TikTok relies on SKAdNetwork’s aggregated reporting and modeled conversions instead of individual-level tracking.
Why do Adjust and Kochava report such different ATT opt-in rates?
Both disclose their methodology, but they measure different app samples over different periods, Adjust averages multiple verticals across several years, while Kochava’s figure comes from a single 2026 marketer survey. Neither source publishes enough overlap to fully explain the roughly 35-point gap, so treat both as directional rather than exact.
Which MMPs support TikTok’s iOS real-time conversion reporting?
Seven, per TikTok’s own documentation as of August 2026: AppsFlyer, Adjust, Airbridge, Branch, Kochava, Singular, and Tenjin. TikTok’s own App Events SDK (version 1.5.0+) also supports it, though the App Events API by itself does not yet, so an API-only setup stays limited to standard, non-real-time reporting.
Is SKAdNetwork enough on its own for TikTok iOS campaigns?
It’s the default fallback, but SKAN 4.0 reports aggregated, delayed data rather than individual events, and its crowd anonymity thresholds can withhold data entirely from lower-volume campaigns. Most advertisers pair SKAN with an MMP connection and modeled conversions rather than relying on SKAN alone.
Should I turn off SKAN attribution to get more iOS campaigns on TikTok?
Only if you’ve already reached TikTok’s SKAN campaign limit and your app qualifies for iOS real-time conversion reporting. Turning off SKAN attribution removes SKAN data from both TikTok Ads Manager and your MMP reporting, so it’s a real tradeoff, not a free unlock.
Should I bother with AdAttributionKit (AAK) instead of SKAN right now?
Not yet, for most advertisers. AAK offers real improvements over SKAN, including reengagement attribution, but adoption remains negligible as of 2026 because publishers, who control the other side of the integration, haven’t widely implemented it.
Conclusion
Full iOS tracking on TikTok hasn’t come back, and even the benchmark data disagrees on exactly how much of it exists today. What’s changed since ATT launched is that the workarounds have matured: real-time reporting through seven supported MMPs, a maturing SKAN 4.0 implementation, and modeled conversions that fill in the rest.
Build your measurement stack in layers rather than searching for one method to replace what ATT restricted, or one benchmark to treat as final. Use your own consented, device-level data as the check on everything else your stack is telling you.
This tracking layer sits on top of the same TikTok ads structure covered elsewhere on this site, campaigns, ad groups, and ads still work the same way regardless of how much of your iOS traffic is consented.
Sources