Kochava Foundry's Grant Simmons Talks Closing the iOS Measurement Gap in a Post-ATT World

Disclaimer: The opinions represented here are those of the individual and do not necessarily represent those of their current or former employer.*

Five years after Apple's App Tracking Transparency changes, most growth teams still talk about post-ATT measurement as a problem they have already solved. The data tells a different story. Adoption of SKAdNetwork and its successor, AdAttributionKit, remains shallow, the signal that does come through is often misread, and a small group of advertisers has quietly opened a measurement gap on everyone else. To understand who is winning and why, we spoke with Grant Simmons, VP of Kochava Foundry, the strategic measurement group that runs advanced attribution and incrementality work for growth marketing teams across mobile apps. After expert-witness work on the industry's largest ad fraud cases and hundreds of SKAN consults, Grant has a clear view of what separates the operators who get iOS measurement from the ones still guessing.

Watch the full interview below, or read on for a selection of key takeaways.

Treat SKAN Adoption as a Competitive Edge

Most teams assume the rest of the market has caught up on iOS measurement. Kochava's own client data says the opposite. What looks like a slow but shared industry migration is actually a widening split, with a handful of advertisers compounding a measurement advantage every quarter while the majority remain stuck on partial implementations. The longer that split persists, the harder it gets for the laggards to close it, because they have no clean signal to benchmark themselves against in the first place.

"We sent a survey out to all of our clients who we saw as actively marketing on iOS devices and just said, how steeped into iOS are you? And the answer is one in five feel confident with their iOS measurement. One in five. … Only roughly a third haven't implemented SKAN at all, and roughly a quarter haven't even heard of AdAttributionKit. Fully implemented and actively optimizing it, all on board, that's 11% of the file."

Start by placing your team honestly on that curve. If you have not implemented SKAN, or you have postbacks coming in that no one has revisited, you are in the majority, not the exception. Grant is blunt about why the whole industry has moved so slowly: the bottleneck is the publishers, not the advertisers. "They are not incentivised to implement increasingly sophisticated versions of SKAN," he notes, "because generally the larger DSPs have figured out workarounds." AdAttributionKit, the current version, accounts for less than 3% of the conversions Kochava sees, and useful features like deferred deep linking that arrived in version 4 still sit unused. The practical move for advertisers is to pressure SDK and publisher partners for current AdAttributionKit support and to treat full implementation as a way to out-measure competitors who are still waiting.

Win Back Deterministic Signal by Prompting for ATT

Deterministic attribution did not disappear after ATT, but the habit of asking for it did. Many teams wrote off the consent prompt years ago, when early opt-in rates were low and the prompt felt like a lost cause, and they never went back to check whether that math still held. It doesn't. User willingness to opt in has shifted a lot since 2021, and the teams still operating on outdated assumptions are passing up a deterministic signal that is far more available now than they think.

"Folks should be prompting. Most folks don't. Only about 8% of the apps we see are actually presenting the app tracking transparency prompt. What's interesting is the opt-in rates of folks who are presented has steadily gone up. When the prompt first launched, roughly about 35% of folks accepted. We crossed the Rubicon in September of 2024, where if presented with the prompt, 50% take it, 50% don't. If you look at last month, almost 70% of folks accept the prompt now."

When nearly seven in ten users now say yes, the prompt is no longer a marginal source of truth. It is the cheapest deterministic data available, and it feeds cleaner attribution and stronger retargeting downstream. The fix is simply to ask. Show Apple's prompt at a point in the journey where the user already understands the app's value, so the request lands in context instead of cold at first launch. Pair that consented data with Apple Search Ads, one of the few places you can still tie a conversion to a specific device and keyword, and you have a deterministic backbone most competitors have abandoned.

Engineer Your Conversion Values and Use the Full Bit Protocol

SKAN doesn't hand you raw events. It gives you a tiny, fixed budget of signal: a conversion value you set at install time, with only 64 possible states to describe everything a user does afterward. How you spend that budget decides whether your aggregated data tells a clear story or collapses into noise. This is where the gap between sophisticated and unsophisticated teams is widest, because most never use the full range available to them.

"We've probably done about a hundred of these, and I've only seen maybe three, four brands use all sixty-four bits. You can also make a heck of a lot more combinations than sixty-four with eight bits worth of data. So you can use those first bits to do other things with. Hold the session open to look for registrations, to do geo adornment, to do audience adornment. We built some fraud flags in there. There's lots of flexibility in there."

Treat conversion-value design as a measurement project, not a one-time toggle. Map your key events and the timing windows around them, then back-test three or four candidate models with regression analysis before you commit, the same workflow Kochava runs in a SKAN consult. The goal is to close null buckets so more of your postbacks carry a usable value, and to spend any leftover bits on geo, audience, registration, or fraud signals rather than leaving them idle. Conversion values can only carry so much, though. Where SKAN's aggregated data runs out, Adikteev's 360-degree iOS approach and its compliance-focused identification system, Adaptive ID, work to recover the post-install signal those values leave behind.

Stop Trusting Any Single Attribution Method

Post-ATT, the signals you do get disagree by design, not by accident. Last-touch credits whoever was closest to the install. A media mix model infers contribution from aggregate spend and pays no attention to individual users. They are measuring different things, so when they tell you different stories about the same channel, the conflict is structural, not a sign that someone's data is broken. An operator who picks one of them and trusts it is making budget decisions on a partial picture.

"If I were to sum it up, it's that no single attribution method gives a full picture now. Full stop. Your last touch is telling you one thing, your MMM is telling you something different. Let's say you isolate Google, and Google says, I'm claiming a hell of a lot, and the MMM says, yeah, but it's not really contributive. Your arbiter between those two things is incrementality testing."

The fragmentation is structural. The networks that win in SKAN are the self-attributing ones, Snapchat, TikTok, Meta, Google, and InMobi, and several of them model conversions you will never see as an advertiser. As Grant puts it, they "grade their own homework." That is precisely why an independent arbiter matters: when last-touch and your media mix model disagree about a channel, you need a neutral test to break the tie. Run last-touch for speed and MMM for the macro view, and treat them as complementary lenses rather than competing scoreboards. For independent, ID-less measurement that does not rely on the networks' own numbers, this is where Adaptive ID fits, giving teams a source of truth outside the walled gardens.

Make Incrementality the Tiebreaker

Every other method on this list describes what happened. Incrementality is the only one that estimates what would have happened anyway, and that counterfactual is the whole game. If a user would have installed or converted without ever seeing your ad, that spend bought you nothing, no matter how confidently last-touch or a network claims the credit. Measuring that gap is harder and slower than reading a dashboard, which is exactly why most teams skip it, and why the ones that don't can settle arguments no other method can.

"We take an exposed universe and we build a synthetic control from it. So we express, what would the user have done without an ad treatment? Once you can capture that, you can express what they did with an ad treatment. We do a lot of incremental testing. It's probably about where half the revenue comes from within my group."

Operationalize this with holdout and control groups rather than treating it as a quarterly report. Build a synthetic control from your exposed universe, measure true lift, and only scale spend on channels that prove they add value beyond the baseline. The payoff is concrete: when Product Madness moved its Heart of Vegas retargeting from a standard Day 7 ROAS target to an incrementality-first approach with holdout groups, it verified a 12% increase in retention among churned players on Day 14, proof the spend was preventing churn rather than cannibalizing organic users. That is the difference between scaling on faith and scaling on evidence.

Check out Adikteev's iOS Incrementality Playbook to see how you can measure true lift and optimize your marketing spend.

The Measurement Edge Is Operational

The post-ATT winners are not the teams with the most exotic tooling. They are the ones who prompt for consent, engineer their conversion values, triangulate their sources, and let incrementality settle the hard calls. As Grant Simmons and the team at Kochava Foundry make clear, the gap between the top 11% and everyone else is operational discipline, not technology. Close that gap now and you compound an advantage every quarter while the rest are still guessing.

Looking to drive more performance from your campaigns? Adikteev helps app marketers scale with advanced retargeting and incrementality-driven solutions. Get in touch to learn more.