Author: John Koetsier

It’s 2025. What’s working for iOS mobile measurement? What is going to give you the data you need to beat the competition?

Sometimes it seems easier to identify what’s not working. 

• SKAN is still hard
• IDFAs are still scarce
• Creative testing is harder
• Probabilistic has its own challenges
• And modeling … each major platform has released or is working on its own particular flavor of modeling iOS app advertising results, but it’s not always clear what goes into that

That’s why I recently discussed the future of iOS mobile measurement and marketing in 2025 with Jesse Lempiainen, co-founder of GeekLab, and Neils Beenen from Singular. (This is part of our 5-part Hack Gaming Growth event: see all the episodes in 1 place here.)

Hit play and keep scrolling for the highlights:

iOS mobile measurement: what’s working

There is good news. iOS mobile measurement still works … just not quite like it did, and not quite to the same level as it did. But there are developments that are hugely successful in restoring ad tracking, creative measurement, and conversion analysis.

Measurement methodologies that matter include:

• CAPIs (conversion APIs)
  CAPIs give you reliable deterministic data on conversions: usually conversions on a website. Snap, Google, Pinterest, TikTok, and Meta all have CAPIs. Learn more about CAPIs here …
• AEM (Meta) and Advanced Conversions (Snap) and Ads Conversion Modeling (Google)
  Platform and ad network modeling give you additional data on ad impact, but they have some challenges (see below).
• Advanced SAN from TikTok and similar programs from other platforms
  Advanced SAN and similar programs provide more data for MMPs like Singular to make accurate attribution decisions, while remaining privacy safe.
• SKAN/AAK
  Yes, SKAN and its new iteration, AdAttributionKit are imprecise, partial, and delayed, but they are also deterministic and have value, even if they are not enough on their own.
• Incrementality
  There are many ways to test for incrementality (see a 2025 guide to incrementality here) but they all have 1 goal: is my ad spend on Network X or Platform Y incremental. In other words, does it move the needle on growth, or does nothing change if I stop my spend there?
• Probabilistic
  Let’s be honest, probabilistic is often used by programmatic networks, and likely others as well.
• MMM (media mix modeling, or marketing mix modeling)
  MMM can be hard to implement, though it has its benefits. Marketers are moving more to incrementality testing than MMM.
• Data governance
  Data governance is simply enforced naming conventions for creatives and campaigns — Singular offers a platform which automates the process — that unlocks otherwise-hidden creative analysis capabilities.

The challenge for 2025 is for marketers to combine multiple attribution methods to navigate iOS mobile measurement in 2025. (Without going insane, or trying to boil the ocean, or stressing too much when varying means of measurement return slightly different results.)

The good news is that Singular does a lot of that for you in Unified Measurement. 

And by mixing the deterministic data that we still can get, and modeled measurement techniques like incrementality, and insights based on smart data governances, marketers can still optimize their ad spend effectively.

Deduplication becomes the key when platforms model conversions

AEM and Advanced Conversions and Ads Conversion Modeling and all the other variants of platform iOS mobile measurement modeling are important in 2025. Yes, they’re a little black boxy. Yes, it’s challenging to know exactly how the big mobile adtech players are modeling their results. 

But we are seeing more insights as a result of modeling that the big platforms are doing on conversion reporting, even if they’re a response to programmatic platforms using probabilistic methods and aimed at recapturing some of the installs and conversions that SANs won attributions on in the past.

The challenge is that when you’re looking at Advanced AEM or Ads Conversion Modeling from Google or Advanced Conversions from Snap, how do you know that they’re not all claiming the same installs?

“ Is there actually one way or one source of truth around how you can actually look at your performance data now?” asks Beenen. “That goes diametrically against the reality on the ground when it comes to measuring iOS signals because there’s too many sources available … the nitty gritty really goes into how you are doing your deduplication.”

Deduplication becomes key, and that’s where additional data from ad partners, like Advanced SAN from TikTok, is helpful.

“ We’re supporting AEM on TikTok, we’re supporting AEM on Meta as well,” says Beenen. “And if you then total the AEM up, of course we see a jump in the actual campaign performance.”

But it’s also important to look at data in your MMP dashboard, like Singular’s, to look for assists, showing you when ads on competing networks are working together to generate a result. And it’s important to check Singular’s attribution decisions, which are based on all the measurement data we can access. Singular sees what each network claims, but then makes an attribution decision that deduplicates all those claims, providing your best view of real, actual impact.

Ultimately you do want to bolster your iOS mobile measurement with incrementality testing as well. When done right, incrementality testing gives you extremely reliable data on what different ad partners and campaigns actually achieve. 

Because ultimately you want data from your ad partners — or at least what they’ll share in a privacy-safe way with Singular — but you don’t want them completely grading their own homework.

Creative measurement requires smart data governance

The challenge with all modeling is creative measurement and optimization, which you’d really like deterministic data on. One solution for that lies outside the platforms and networks themselves, and it’s data governance.

Seemingly unnecessary for smaller organizations, data governance is essential for all because campaigns and creatives easily reach into the hundreds, even on just 1 platform, even at small scale. For large organizations, we’ve seen 10s of thousands of creatives, which very quickly scales out of control if you try to manage it manually.

Data governance helps with iOS mobile measurement because when you manage creative names, campaign names, geo encoding, conversion outcomes, and any other metadata you want for optimization and reporting well, you can get high-quality first-party data on which creatives and CTAs and campaigns worked best, converted most frequently, and resulted in the highest ROI.

Even, to an extent, inside platform black boxes that might take your creative and mix it up in the blender to build thousands of ads from dozens of components. You’ll still know that a certain image and a certain string of text were involved in good — or bad — results.

Check out how here.

Conversion APIs (CAPIs) are increasingly important in iOS mobile measurement

When you think CAPIs, you think retail typically: a person clicks an ad, goes to a website — usually a mobile website, but not always — and buys a product. A web SDK in that retail site informs the ad network, and a conversion gets claimed.

But it can be used in gaming as well, via a mobile SDK.

And it’s another useful signal, says GeekLab’s Jesse Lempiainen:

“ That’s something that we’ve utilized quite a bit, both web2app campaigns that we partner up with Meta officially, and helping mobile publishers and developers do those web2app campaigns,” he says. “ We’re kind of staking a step back and measuring everything on a creative level and, and then utilizing these web technologies essentially … so conversion API to send events back to networks.”

Success that GeekLab is seeing there includes:

• Lower cost
• More daily new users in App Store Connect
• Retention rate going up

But don’t expect it to be perfect: Lempiainen estimates the data accuracy level at around 70%.

Which brings up a good point in iOS mobile measurement for 2025: don’t expect perfect. And you don’t need perfect either. You do need good directional data that you can make quick decisions on … and that you can get.

In fact, that’s enough to keep the growth machine moving in the right direction. 

(See what Jonathan Reich, CEO of Zedge, says about the perceived need for perfect data.)

More in the full podcast

Of course, there’s more in the full podcast episode. And don’t forget to check out all of the Hack Gaming Growth sessions as well.

Here’s what else you can expect:

• 00:00 Introduction to 2025 iOS Measurement
• 01:01 Current Challenges in iOS Mobile Measurement
• 03:50 Retargeting Strategies and Banner Ads
• 04:40 The Role of SKAdNetwork and AEM
• 05:47 Probabilistic Attribution and Industry Trends
• 09:06 Unified Measurement and Incrementality
• 19:24 Web to App Campaigns and Conversion APIs
• 24:36 Future of iOS Game Marketing in 2025
• 27:45 Conclusion and Final Thoughts

Get it wherever you get podcasts, or subscribe to our YouTube channel.

Of course, you already know: incrementality testing helps determine whether your ad spend is truly driving new user growth … or merely capturing users who would have installed your app anyway.

But how important is incrementality testing?

And how can you do it relatively painlessly?

In a recent Growth Masterminds I chatted with Jonathan Reich, the CEO of an app publisher with 750 million app installs: how they think of incrementality, how they measure incrementality, and tips for how you can do the same. The publisher is Zedge, which specializes in phone personalization with offerings like wallpapers, ringtones, and AI-powered customization tools so people can create their own one-off phone customizations.

Their flagship Zedge app has been installed over 750 million times, with 25+ million active users and 15 million reviews (!!). It’s primarily ad supported, so smart arbitrage on user acquisition is critical.

Which makes incrementality testing also critical.

Check out our chat here:

Why Zedge started incrementality testing

Initially Zedge grew organically. 

It’s in a cool niche — young smartphone owners are particularly interested in customization — and had such an early start from even before the smartphone era started that millions and millions of users just naturally downloaded the app and used it, thanks to strong ASO as well as SEO.

Eventually, however, Zedge started paid user acquisition in order to accelerate growth even more.

And it clearly worked.

But there was always a nagging question …

“At first, wow, we’re seeing return on ad spend through the ceiling,” Reich told me. “But then the question that came up was, hold on a second … if we were to stop advertising, would we generate those installs and that revenue accordingly?”

And that’s where incrementality testing comes into the picture.

How Zedge does incrementality testing

There are a ton of different approaches to incrementality testing:

1. Holdout group
2. Geo-based
3. Intent to treat
4. Time-based
5. Ghost ads
6. Synthetic controls

Check out much more about all of these, how they work, and how to start in our recently-published definitive guide to incrementality in 2025. 

Zedge primarily manages checking ad partners and campaigns for incrementality in 4 ways:

1. Market-level testing
   In market-level testing, Zedge pauses advertising in select markets and measures the impact on both overall installs and — importantly — organic installs. Depending on the difference, the company restarts campaigns.
2. Budget fluctuations
   If everything in your marketing budget is steady-state, it’s almost impossible to know what’s incremental. So Zedge will dramatically increase or decrease spend to assess the impact on install volume and velocity. How? They’ll often use a “sine wave” approach, Reich says: fixed spend, small spend, big spend, small spend. Comparing the varying investment levels to what you actually know happens in your app — your own first-party data — gives you all the information you need to know about the incrementality of your campaigns.
3. Platform comparisons
   Zedge will sometimes shift budget between platforms to see changes in organic installs and paid ROAS.
4. INCRMTL
   Zedge uses INCRMTL (and Singular data) to assess incrementality on a regular basis.

“There are whole sets of different tests that we have undertaken over time in order to really test the limits,” Reich says. “And we’ve done so without breaking the bank.”

That’s important.

You want to test, but you don’t want to overspend on tools. And you don’t want to underspend on campaigns that actually are incremental to your growth and profitability.

Should you buy your name on the App Store?

There’s 1 really big question every brand has to think about at some point: do you buy your own name on Apple Search Ads and Google Play?

It’s a tough question to answer. And there are potentially massive consequences to getting it wrong.

There is a good side to the problem, but also a bad side.

“If you are first and second, someone’s going to click on you,” Reich says. “However, when you are paying for that and you’re stealing from your organics, that is bad.”

The good news: incrementality testing can give you an answer. And it’s relatively easy to get the answer if you’re in multiple geos.

Simply pause spend in select geos, measure what happens with install volume, and assess whether the price you pay for showing up at the top of your own searches is worthwhile.

Key incrementality testing learnings

Incrementality will teach you things about your marketing that are hard to learn any other way. Here are a few that Zedge learned.

1. Some channels simply won’t work for you
   And some partners simply won’t work for you. No matter how much optimization you do, some partners failed to provide incremental value for Zedge, and you’re likely to find the same thing. That’s OK. Not everything works for everyone. Just find the ones that do.
2. Different channels drive different outcomes
   For Zedge, some were better at driving users who monetized via ads. Others were better at driving subscribers. You won’t know which is which until you test.
3. Don’t overvalue precision
   Seeking ultra-precise attribution data can lead to analysis paralysis. Once you get direction-relevant insights, make changes.
4. Don’t go “cold turkey”
   Completely turning off UA will hurt overall growth. You can run incrementality testing without getting that extreme.
5. Don’t be impatient
   Incrementality takes time to measure effectively. Don’t make changes and poke them with a stick tomorrow. Let them breathe.

Finding success is great. But finding failure is also progress, because you’ve eliminated a possibility.

“We have found that there are certain DSPs that just don’t work for us,” Reich says. “No matter how we test, no matter whether we scale up, scale down.”

That’s why you test before you scale.

How to test a new ad partner

So how do you test a new channel or ad partner? In a word, carefully.

1. Start small
   Give your new ad partner an initial test budget of a few thousand dollars over a couple of weeks.
2. Expand gradually
   If your early results with the new ad partner or channel are promising, increase spend in stages. As you do …
3. Monitor multiple factors:
   1. ROAS trends … up/down/same?
   2. Impact on organic installs … increasing, decreasing?
   3. Quality of new users … engagement, retention, monetization?

You have to test new partners to find the nuggets of gold. But you have to do so carefully.

“We’ll take a small test budget, dip our toe in the water, and begin to take a look at the results,” Reich says. “And then once we validate that something is helping us, the question is, how fast do you press down on the accelerator pedal?”

Much more in the full podcast.

As Zedge learned, incrementality is more art than science.

While data is crucial, success comes from balancing your quantitative analysis with your strategic intuition. The challenge: marketers have to be capable of accepting uncertainty.

Remember, the goal is optimization, not perfection.

“When you start out, you just think, hey, this is all digital,” Reich says. “We can measure anything. We’ll have answers instantaneously. And it ain’t like that.”

Check out the full episode wherever you get your podcasts, or on our YouTube channel. Here’s what you’ll find:

• 00:00 Introduction to Incrementality Testing
• 00:44 Meet Jonathan Reich, CEO of Zedge
• 01:15 The Origin Story of Zedge
• 05:24 Marketing Strategies and Organic Growth
• 07:26 Challenges and Insights in Incrementality
• 09:48 Testing and Optimization Techniques
• 14:44 Key Learnings and Best Practices
• 20:04 Common Pitfalls and Final Thoughts