Post-IDFA user acquisition: what I’m seeing and where we’re going

“Everything is supposed to be getting easier!”

I was talking to a mobile marketer the other day and he told me that the overall stream of technological progress is that the complex becomes simple, the hard becomes easy, and the impossible becomes doable. But, he added, there seems to be an opposite flow in mobile marketing and specifically mobile user acquisition.

That’s a little hard to argue with.

iOS 14.5 and the IDFA apocalypse, iOS 15 and Private Relay, plus recent changes from Google in “data-driven attribution” and Facebook deprecating Advanced Mobile Measurement … 2021 has been a year of changes for growth marketers. And guess what: we’re not finished yet. There’s more change coming, more learning, and probably more disruption.

And yet, there are some things I’m seeing (and some things Singular is working on) that might just help reverse that current and bring some simplicity back to mobile measurement.

What we’re seeing post-IDFA

In 2021, everyone is re-evaluating their marketing mix. In general, user acquisition teams are allocating their marketing spend in six different ways:

  1. Flight to quality
    Known good channels with repeated proven success in the past
  2. Owned marketing
    Marketing with first-party data
  3. Organic social
    Social/viral marketing with organic reach
  4. Influencer marketing
    Measurement is hard everywhere? Let’s try influencer marketing, where this has always been an issue
  5. App Store Optimization (ASO)
    CTR is important but CVR is critical: let’s get better at converting the traffic we do get into actual installs
  6. Still fully measurable channels
    There’s a reason Android had a big spend surge right after iOS 14.5 dropped

Add it up and we’re still seeing spend aggregate to major networks, but with lower budgets. More budget is going to the Liftoffs, ironSources, and Unitys of the world where contextual targeting and SKAN both operate. Brand budgets are bigger since performance is harder. Owned and earned media have become more key to success. Cross-promotion between owned apps and partners is growing.

And, of course, there’s a significant shift to mobile web, where you can quickly and cheaply test multiple approaches and where you are operating on first-party data principles.

With the outflow of performance dollars to Android, CPMs rose there. But we also saw clueful and aggressive competitors doubling down on iOS and “getting a discount” on less expensive campaigns.

The IDFA and mobile measurement

The result of iOS 14.5 and App Tracking Transparency is pretty clear: the IDFA is no longer really relevant for performance marketing calculations. It’s still an important resources for deeper insights, but it has lost its place as the primary driver of measurement and truth.

Instead, marketers are turning to three alternatives:

  1. SKAdNetwork
    Even top brands and very good performance marketers have had issues making SKAN work, but once you get it right, SKAdNetwork provides the best source of data on iOS user-acquisition advertising campaigns. It’s simply the future and where you need to invest, including as iOS 15 gets widely adopted.
  2. Web-to-app flows
    Where it works and in certain verticals, we’re seeing significant action in web-to-app flows. Marketing on the web can be much cheaper. It can make you more flexible. It can enable more experimentation. You can deliver more information, and you can collect first-party data before sending someone to the App Store (or Google Play, in the case of Android).
  3. Fingerprinting
    Some marketers are trying to do fingerprinting or probabilistic attribution. We don’t support or recommend this, and we’ve been very clear that it’s not a good solution. It’s only even possible, of course, for non-SANs traffic, so it’s automatically limited to a quarter or so of the UA market. And we think it will get technically harder and harder to do as Apple continues to close the loop on privacy.

Ultimately, marketers need to rely more and more on aggregated data, whether in SKAN, from Facebook, via Tiktok view-through, or some kind of cohort modeling or media mix modeling approach. And especially you need to get good at understanding and using SKAN. We have many examples in our data where marketers who get it achieve impressive growth results with SKAdNetwork.

It does work. It also does, sometimes, need some debugging.

Note:

We can help with that.

Making SKAN performant because … it is the future

Start simple. Ensure you’re set up and integrated right. Build the infrastructure you need on the backend, and make sure your partners are receiving postbacks. Use the reporting you get to establish CPA.

Then evolve.

Use your existing first-party data to analyze what correlates to your high-value users within the first 24 hours. Is it one event? Do you need to mix a few pieces of data together in an SKAdNetwork postback to get more information? We’ve seen just in-app purchase revenue or ad monetization revenue work. As you get more sophisticated, you want to build towards getting a good pLTV prediction within the first day after install. If you can’t now, your app may need some retooling to enable this, particularly in the onboarding phases.

Then implement the conversion models you’ve built with your MMP’s tools. That’s simpler, quicker, and more reliable than reinventing the wheel.

Some of what we see working:

  1. Hyper casual games
    Using ad monetization revenue as the key conversion value can ensure you leverage SKAN really well (and yes, Singular offers that model right out of the box)
  2. Gaming apps
    Focusing on IAPs works extremely well as conversion values
  3. Subscription apps
    Subscription-based apps are more likely to use a number of events or a mix of IAP revenue and events for conversion values
  4. Fintech, services, e-commerce, travel
    Here we see a lot of experimentation with web-to-app funnels

Important: beware of privacy thresholds.

If you don’t hit enough volume per campaign, all your hard work is down the drain. The result is what looks like crazy CPIs for paid acquisition while the blended organic/paid still looks normal. In general, that means at least 30 installs per campaign per day. But because Facebook uses so many of your SKAN campaign IDs for internal tracking, target, and measurement, Facebook recommends 180 installs per campaign per day.

Put simply: unlocking privacy thresholds is the key to get a complete and usable view of your metrics with SKAN.

Ultimately, we see more and more clients being able to use SKAdNetwork and scale with it. Often, they’re big brands which are less likely to go against Apple’s policies and therefore more incentivized to make SKAN work. Bigger companies also clearly have more to benefit from cross-promotion and are more likely to have robust pLTV models in place. And bigger companies are also more familiar with running brand campaigns.

But smaller companies test faster, adapt faster, and can unlock significant growth with few channels, so they have their own advantages.

Making mobile marketing simple again

I’m not sure we can quite achieve this, and I’m not 100% sure we ever had it even when IDFAs were free. But mobile marketing certainly was simpler.

The future is simpler too, but it’s more complex to get there.

That future involves building out varying views of reality and integrating them intelligently into a single source of truth. At Singular, we’re looking at marketing performance from known and aggregated spend data, from deterministic last-click measurement, from probabilistic aggregated results data, from first-party data, and from other sources. All of those have their unique perspective on what is actually happening as marketers market, whether putting dollars to work or investing in organic promotion. Each of them has value.

But then they also need to coalesce into a single source of truth to provide a simpler modeled view of reality.

Sure, marketers need to always be able to dive deeper, to investigate raw data, to analyze assumptions and insights from the various perspectives … but they also need a reliable data-driven way of simply getting a real-time scorecard on their progress.

Everything changes.

That’s the one unchanging constant.

And the marketers I’m talking to understand that and live that. They’re continuing to build, to test, to invest, and they’re continuing to deliver results. And we’re continuing to support them in every way possible.

Traditional MMPs are dead. Welcome to next-generation attribution

“Ask not for whom the bell tolls. It tolls for thee.”
 – John Donne (paraphrase)

There’s a reason the first mobile measurement platform was initially called Mobile App Tracking. Traditional MMPs were built to follow the activity on your phone. Life was very simple:

  • See an ad. UDID
  • Click an ad. Deeplink
  • Install an app. Fingerprint
  • Complete level 3. GAID
  • Buy a power-up to storm the castle and rescue the fairy prince/ess. IDFA
  • Click on a cross-marketed ad and install another app. IDFV

Trackers powered the mobile growth machine by collecting impression, click, install, and conversion data and revealing it to app developers, publishers, and marketers. Now that’s changing. Today with the ongoing deprecation of freely available device IDs and the introduction of advanced privacy regulations and frameworks, that world is disappearing.

Those MMPs that cannot survive in a more complex world of different data from more sources with less certainty will disappear along with it.

 

A new world of marketing measurement

While you could trace the new world of marketing measurement in mobile back a decade or more (right back to when Apple and Google dropped hard-coded device IDs and switched to advertising IDs to boost users’ privacy), it really started in earnest in 2019. That’s when Google announced that app installs resulting from iOS search activity would be modeled, not granular, and … not shared with third-party measurement tools. Now, to give marketers an accurate picture of what they were actually achieving we had to blend data from multiple sources to provide a holistic view of performance.

This spring iOS 14.5 turned the dial to 11, of course.

IDFA is now scarce, and while SKAdNetwork returns deterministic data about installs and conversions, none of it is tied to a device or person. Plus, while Android still offers an advertising identifier, the GAID, we see Google experimenting with FloC and other privacy-centric technologies that likely foreshadow additional changes and increased fragmentation of marketing data and best practices between the major mobile operating systems.

This comes close to eliminating the traditional role of the MMP.

Suddenly, a mobile measurement company must be more than a simple tracker to be useful to marketers.

If you are just an MMP, you see ad impressions. You measure clicks. You keep track of app installs. You cross-reference those with clicks and perform attribution. You get post-install events and calculate ROI and ROAS. As we all know, this is exactly the functionality that is being increasingly limited.

But it’s important to remember something critical:

This has always been only one chapter in a much larger story.

 

Looking at the bigger picture

For traditional MMPs, app installs happen by magic. They are created ex nihilo … one moment there’s nothing and the next POOF … a click, an install, and maybe a conversion.

That is not how a marketer sees app installs, however.

A marketer selects an audience, makes a campaign, designs creatives, chooses channels, picks partners. When the marketer gets a click, it’s not a surprise. An install is not a happy coincidence generated by random variables. A conversion doesn’t magically show up out of the blue. They happen as a result of all that planning … which a traditional MMP has zero visibility into.

Among the marketing measurement companies commonly referred to as MMPs, Singular is the only player with insight into all the key chapters of that growth story. And unlike other MMPs out there, Singular is well-positioned to help marketers optimize for growth at every stage of the journey … whether it’s mobile-centric, web-enabled, or multi-platform.

That’s because our goal has always been to connect as many data sources as possible: upper funnel and lower funnel, across channels and devices, campaign data and results, costs and ROI, bids and budgets … should I go on?

And, of course, our goal has always been to combine that data in interesting ways. The consequence is that Singular provides insights traditional MMPs can only dream of.

Want visibility into your creative performance? You’re going to need upper funnel data: campaigns, creative, impressions, clicks. Want to make them most effective for analysis? You’re going to need to combine that with lower funnel data: installs, actions, engagement, conversions. Want incrementality? You need every data source and piece of metadata, you need normalization and standardization, you need upper and lower funnel. Want cross-platform and cross-device? You need web as well as mobile.

This new reality is much more complex than the old one. It requires many different data sources.

If you’re merely a mobile SDK, you only see a fraction of the activity.

What you need is an MMP with a complete marketing data infrastructure — beyond just a mobile tracker — to empower growth marketers to see the entire marketing picture, get the full growth story, and capture the kind of advertising insights you need to optimize. You need tools that futureproof your efforts for a vastly changed (and continually changing) marketing ecosystem.

 

Different marketers have different needs

Singular isn’t the solution for everyone, and the next-generation attribution we’re building won’t fit every marketer’s need. But it is built for two specific types of marketers.

1) Mobile-first marketers: app is the majority of your product
Clearly, an MMP-style solution has typically been the solution of choice for mobile-first app marketers.

As an official MMP, this has always been a key part of our offering. However, we built it as a component of a complete mobile marketing infrastructure: collecting additional data sets and integrating directly with your internal systems.

Now more and more mobile-first growth marketers are seeing the benefits of this broader solution.

Non-MMPs can’t help: they can’t measure what you need to optimize. Traditional MMPs understand your business, but their hands are tied behind their backs in a changing world with less deterministic device identifier-based data. This marketers needs more than what a traditional MMP can offer: need extended visibility from data that can never be taken away … and security that comes with knowing you are protected as future change occurs.

In this emerging reality with Singular’s breadth of data sources and out-of-the-box cross-platform attribution, app marketers can own a full solution in one product.

2) Omnichannel marketers: app is part of your overall solution
Brand marketers in CPG or fintech or travel or other verticals have different challenges.

You need mobile measurement that only a traditional MMP can provide, but you probably also need additional tools, perhaps web-native and desktop, perhaps marketing clouds and CRM, to feed, run, manage and measure other components of your marketing activities.

There isn’t one solution that gives you everything, but Singular not only provides marketing measurement for mobile but also for web, or email, or anything else digital, and provides tools that neither traditional MMPs nor web-native marketing platforms can offer. And, of course, Singular integrates well into your suite of growth management tools.

 

Next-gen measurement: reinventing simplicity in an era of increasing complexity

What worked for traditional MMPs was simplicity: assigning attribution based on the last click, checking source apps and channels, viewing campaign ROI, generating insights about what to stop, what to start, and what to scale. In our past mobile marketing reality, this was not difficult.

That reality is changing.

And that level of simplicity is increasingly just not an option for marketers anymore.

Now you need all the data you can get. It comes in differing levels of granularity. It comes in multiple datasets from multiple partners. It includes campaign data like creatives, bids, and budgets. It combines attribution data like outcomes, installs, sales, conversions, achievements. It’s from the web. It’s from mobile. It’s SKAdNetwork, and it’s IDFA and GAID. Now you have to combine these data sets for both deterministic outcomes and probabilistic correlations to positive outcomes. Media mix modeling and incrementality are both paths to the same thing: understanding what builds value and generates desired action.

Singular is in the best possible position to help app and brand marketers achieve this. We’re the global experts in bringing it all together and making it fit. And we’d love to show you how we make the complex simple.

 

Talk with us

Interested in learning more about how you can futureproof your marketing growth with next-gen analytics and attribution?

Let us know. We’ll be in touch.

How Facebook could do look-alike audiences in SKAdNetwork and iOS 14

Look-alike audiences is an incredibly helpful tool in the mobile marketer’s arsenal, and it’s one that most user acquisition experts have written off in terms of the future of mobile marketing on iOS. Look-alikes (LAL) are largely powered by the IDFA, and the consensus is generally that IDFAs will be scarce. (Plus, don’t forget, you can’t just use first-party data like an email address as an identifier if you haven’t received App Tracking Transparency permission.)

But could SKAdNetwork power look-alike audiences?

In a privacy-safe way?

Without identifying specific users or devices? And without violating any of Apple’s policies?

That question recently popped up in the Mobile Attribution Privacy group on Slack, and a few of us took a bit of time to think it through. I want to be clear: I don’t have any special or private insight into what Facebook is or might be doing here. This is just speculation on my part as to how privacy-safe look-alike audiences might be able to work under SKAdNetwork on Facebook.

The question: look-alikes via SKAdNetwork?

A marketer asked if anyone knew whether we will be able to create look-alikes based on SKAdNetwork specific events or not.

One reply suggested it should be possible to build look-alikes out of SKAN conversion values … as long as Facebook is able to decode them according to how you encoded the six-bit conversion values. A concern, however, is how Facebook would be able to link the conversions back to individual people without violating privacy and tracking people who may not have consented.

Here’s one way I think it could work, and how Facebook could do look-alike audiences in iOS 14. Interestingly, it’s sort of similar to how Apple Ads Attribution works.

First: how look-alikes work now

Let’s start here: in terms of how look-alike audiences work today, you send Facebook your app events plus users’ IDFAs for the audience you want to do look-alike targeting on. Typically, of course, you’ll pick your best users, your paying customers, or your most-engaged community members.

Once Facebook gets that data, they match it row by row to Facebook users.

Then they look at their universe of Facebook users — and it’s a large universe. WhatsApp has two billion users, Instagram has well over a billion, and the core Facebook experience itself is closing in on three billion.

Post IDFA, all this changes significantly.

Sure, for all your App Tracking Transparency users you get direct data points, and if Facebook gets enough data points from its own users agreeing to be tracked AND enough data points from other third-party apps whose users also agree to be tracked, the old methods still apply. There is likely to be some of this, but whether it’s 5% or 50% of the ecosystem … time will tell.

If it turns out to be more than 20%, that’s still a significant number of people, and a significant number of your app’s users. You’ll be able to do some testing to see how accurate modeling based off these users is for your non-consenting app installers, and that modeling could be very useful.

(I do wonder if more consent will come from your most valuable users, though. It might be the case that those who trust you enough to buy significant amounts of in-app purchases in your app/game should trust you enough to let you track them. Again … time will tell.)

But if you don’t get consent, then Facebook doesn’t know who the user is when they get the SKAN postback, so the standard LAL model simply wouldn’t work.

Now, here’s how LAL could work in iOS 14 via SKAdNetwork

There may still be hope for look-alike audiences, but I reiterate: this is just my guess. This isn’t insider information.

OK, here we go:

  1. Facebook knows which of their users engage with which ads in significant detail. They’ll have impression, video view, video complete, and click data, plus probably much more. That is definitely already a signal about “people who might be interested in XYZ.” (In addition, Facebook has a lot of first-party data on each user based on things people have explicitly told Facebook.)
  2. Facebook also knows which users they place in a specific SKAdNetwork campaign. That means that when a SKAN postback arrives, it could be one of the users that was in that campaign during the relevant time window.
  3. For Facebook, each ad campaign will have a huge number of users. But over time, if you randomize how you assign users to campaigns (so that it’s not always the same people in the same campaign ID), you could start getting a signal of the probability that a user in that campaign is a whale. Also, if advertisers encode “install day” into the six-bit SKAdNetwork conversion value, it would help Facebook narrow down the set of users who could potentially belong to that anonymous SKAN postback.
  4. These are just one of the many reasons why Facebook is reserving 91 out of the 100 possible SKAdNetwork campaign IDs for their own algorithms, only giving advertisers nine campaign options.
  5. So in theory, Facebook just needs to get a lot better at building a probability that a user would be a whale in a certain game as opposed to an IDFA-universe knowing that someone is a whale like today. This, of course, is where their data science army comes in.

All of this is pure speculation, but it is one conceivable method how Facebook could maintain efficacy — probably lower than today, but still decent — in look-alike campaigns. I’m also sure they have dozens of other methods and ideas that I can’t even imagine yet since I’m just looking from the outside.

It’s more than Facebook, of course

I’ve used Facebook in this example because the big social network is the most prominent example of look-alike audiences for most mobile marketers. But many if not all of the big platforms are preparing to run similar technology, based on what I’m seeing. Whether you call it differential privacy or federated learning of cohorts or forest-in-the-trees privacy, the basic concept is generally equivalent: group individual devices and/or people and make them individually identifiable.

Then target by cohort, not person.

Ultimately, however, the main question is: will Facebook and other platforms get enough quality data to keep look-alike audiences as an effective targeting mechanism or not? How effective? And how much more data will be required to make cohorts targetable versus individual.

Plus, of course, if they can … will they enable that feature either explicitly or implicitly? You could imagine a scenario where platforms are able to do this, but keep the “secret sauce” of how it works behind the curtain.

Look-alike audiences are uncertain

But iOS 14.5 and increased privacy are not.

If you’d like some help preparing for iOS 14, SKAdNetwork, and all the massive amount of change the mobile marketing ecosystem is experiencing, we’d love to help.

Book some time. We’re here to talk.

pLTV in iOS 14: How to make SKAdNetwork conversion postbacks totally rock

The biggest lie about SKAdNetwork is that you can only return one post-install conversion value out of a pool of 64 potential options. Or that it’s basically a mono-channel conveying only one type of message.

Conversely, the biggest opportunity to win with SKAdNetwork is to deeply understand what it actually provides … and optimize the hell out of every last bit of it.

I mean that absolutely literally.

SKAdNetwork is, of course, Apple’s privacy-safe mobile attribution framework. And while it has definite limitations in granularity and post-install events — and does, literally, have only 64 discrete post-install conversion values — the secret to optimizing mobile marketing after Apple turns SKAdNetwork fully on relies on using every iota of an unsigned 6-bit entity that Apple calls “conversion-value.” Only then can you optimize your SKAdNetwork conversion models.

Why?

Because understanding the value of acquired users is critical to estimating ROAS and building LTV models. And it has to be predictive.

pLTV: Predictive life-time value

You simply must get good at predicting LTV in SKAdNetwork, because SKAdNetwork gives you only one chance to get post-install conversion data on acquired users, with very strict timing constraints. If you can’t predict the estimated value of newly acquired users quickly — most likely within one to three days — your ability to optimize ad campaigns will be critically impaired.

Fortunately, there’s a very good way to do all of this. And while it does require some level of sophistication in strategy, Singular makes it incredibly easy in execution.

Ultimately, it’s all about bits and bytes.

While these days we tend to think of data storage in terms of gigabytes and terabytes, a bit is the smallest atomic unit of storage in an electronic system. It is literally either a 1 or a 0. On or off. Something or nothing. One bit then, gives you one of two potential pieces of data.

Each additional bit gives you twice as much, so six bits gives you a whopping 64 potential values. (Geek note: since nerds count from zero and not one, the 64 values start at 0 and end at 63.)

bit and values for SKAdNetwork

That’s precisely why many tend to think of SKAdNetwork as providing 64 values, and think of one value being a specified LTV range, for instance, and another value being a different range. Or, each value could be mapped to an actual or inferred action by a user.

Here’s a critical insight.

SKAdNetwork: 64 values versus 6 bits

You’re actually not limited to 64 values. You are limited to six bits. And that’s a totally different thing, because each bit is essentially its own complete universe, enabling smart advertisers to walk and chew gum at the same time, while also playing the banjo and maybe even splitting the occasional atom just for fun.

In other words: you can measure multiple things in parallel.

Now you’re cooking with gas.

How you want to think about and use your precious six bits is entirely up to you, so you can align it precisely with how your app works, what your business model requires, and — perhaps most importantly — what you know about the relationship between D1, D2, or D3 events in your app and revenue events farther downstream.

In this example, you can use two bits for cohort data — allowing up to three days — and four bits for up to 16 different revenue buckets.

potential uses of SKAdNetwork bits

In this example, you’ve got the same two bits for up to three days of cohort data, but you’re adding a potentially deterministic on/off signal for a conversion event — in this case a sign-up — and leaving three more bits for up to eight different revenue buckets.

Here you’re really amping up the potential by getting cohort data, the sign-up event we’ve already mentioned, a predictive event based on user behavior you’ve seen in-app, and still four separate revenue buckets. (None, little, lots, whale?)

And those are all simple examples of how to use the available bits to drive your SKAdNetwork conversion models.

Bits, bytes, and your app’s critical events

There’s nothing too complex happening here, but the amount of data you’re getting back from one lonely post-install conversion postback is already starting to look fairly robust. Now it’s time to think about your app and how you’re going to use your six bits.

  • What are the early signals of eventual value?
  • How can you optimize your app experience to surface them as soon as possible?
  • What should you test in terms of usability and customer journey to get those predictive signals in the first few days?

What you return in your SKAdNetwork post-install conversion values doesn’t have to be actual events. It could be predictions. “Predictive,” after all, is not about the data that a platform supplies you. It’s about how you use that data, and match it up with what you alone know about your users and customers. In SKAdNetwork, you get to decide what each bit does: is it deterministic (X happened, or Y did not happen), or is predictive? And your app can relay data to your servers and query for a predictive score … which you can then update Singular with for reporting via SKAdNetwork.

If you find even one event that has predictive value, then you can create a model. If you can create a model, you can test it.

Once you’ve tested it — and now is a good time because you can look at SKAdNetwork data side-by-side with IDFA data — you know your user-level predictive capability.

And that’s an app marketer’s superpower.

You may not know it per-user by platform, but you know it in your app. So you can calculate your predictive accuracy, you can improve your model over time, you can grow your confidence in your predictions, and ultimately, you have greater power to optimize marketing investments. Tip: you can do entirely risk-free prediction testing by making predictions for conditions in the past. Predict X results for Y conditions for a 30, 60, or 90-day old cohort, and check your historical data. Fine-tune your predictions, and use them live when you’re ready to make future predictions.

When LTV prediction is hard

Even if you can’t find events with strong predictive capabilities, there’s still huge value.

For instance, maybe it’s hard to predict LTV in two or three days, or even seven days. But you can report deterministic values. And maybe you can commit to something weaker than a strong LTV prediction: a tiering of users into groups, perhaps.

In that scenario, you use two bits to assign users to weakly predictive tiers.

This isn’t a hard prediction or a specific event. It’s not a revenue level. It’s simply a grouping based on initial usage, and you can refine it as you go. But now you can generate reporting, assign a value to each tier, and measure predicted ROAS. Now you’ve got a system and process, and now your job shifts to simply using data science and machine learning to continually improve that system and process.

1% improvement each day makes you 40X smarter over the course of a year.

That might be a bridge just a bit too far (or two bits … sorry). But better is better, and whatever your improvement rate is, if you can engineer in continuous improvement at that rate … you’re going to get to the promised land.

Or, better ROAS.

Whichever you prefer.

Get more details on Singular’s SKAdNetwork solution here. Or, request a demo here

iOS 14 and MMPs: Where we stand right now

No IDFA? No problem. Continue to drive growth with Singular’s best-in-market SKAdNetwork solution. Click here to learn more!

Two weeks have passed since WWDC 2020, and the mobile marketing ecosystem is still evaluating what the future will look like. Various players in the industry, including MMPs, networks, publishers, and advertisers, have analyzed the changes and suggested different paths for our collective future.

At Singular, we strive to take an active role in shaping the future of our industry.

That’s why we’re the first MMP to announce full support for SKAdNetwork and have already released open-source code for ad networks, publishers, and advertisers to start getting used to SKAdNetwork and its capabilities.

We understand we are not able to control all the variables, and many unknowns are still to be solved. But we are working with partners, other MMPs, and Apple to potentially innovate additional solutions. I want to take this opportunity to recap the current situation and get up to speed with the latest developments.

SKAdNetwork

Apple released SKAdNetwork over two years ago, but it was still relatively unknown until WWDC this year.

We’ve written quite extensively about what SKAdNetwork is, but as a quick reminder,  SKAdNetwork is a privacy-preserving framework — a chunk of code — developed by Apple for mobile app install attribution. It runs on-device to measure conversion rates of app install campaigns without compromising users’ identities.

The main advantages of SKAdNetwork:

  • Aggregate-level accurate and potentially fraud-free attribution
  • Supports measurement of all ad networks, including self-attributing networks or SANs (Google, Facebook, Twitter, Snap, etc.)
  • Developed and promoted by Apple, so likely to be the standard for mobile attribution on iOS
  • Significantly improved as of version 2

The main limitations of SKAdNetwork:

  • The number of tracked campaigns is capped at 100, including all granularity levels from campaign to creative (although publisher and country are supported separately)
  • Limited to a single conversion value metric for each app install
  • No support for long cohorts
  • No view-through attribution
  • No user-level data
  • Partial real-time attribution feed

As things currently stand, SKAdNetwork is likely to become a leading methodology for mobile measurement on iOS, mostly because it is aligned with the privacy-centric standard Apple is promoting. This is why we’ve been preparing for its adoption for over a year, and why we announced a SKAdNetwork solution to support our customers and the ecosystem the day after WWDC.

However, it will be a gradual process and we are also working to promote some key improvements that would help it accommodate the full needs of marketers.

Proposed solutions for user-level attribution

Maintaining user-level attribution is definitely the path that would minimize disruption of the ecosystem, supporting the measurement methods which are used today, including granular attribution and user-based cohort analysis.

There are a few interesting suggestions in the industry that would potentially preserve user-level attribution.

We are analyzing these methods and actively participating in discussions with industry leaders and other MMPs to find creative solutions. Obviously, if Apple accepts a satisfactory solution, the entire industry would follow and Singular would be a key part of that.

However, there is a big “if” behind all these methods.

The essence of user-level attribution is combining user-identifiers such as IDFV from the advertiser app with marketing data generated on the publisher app. While Apple didn’t provide an explicit statement about methods that do this, this is clearly not in the spirit of Apple’s privacy guidelines for the IDFV:

Let’s take a look at two of the main suggestions so far …

1. Fingerprinting and probabilistic matching

Fingerprinting has been an alternative method for matching a touchpoint (a click or impression) with an app install. This method uses identifiers such as IP address, user-agent, device type and other metadata about a device to connect each touchpoint. Every MMP has used this method as a default in cases where IDFA isn’t available (for example, an app install driven by a mobile web ad, or a device with Limit Ad Tracking on, hiding the IDFA).

This method can work quite well, but since it’s not 100% accurate, it’s considered a probabilistic method. MMPs have been improving this mechanism for a while, and while we see more room for innovation and claims about improved accuracy, there still exists the question of Apple’s opinion on fingerprinting.

While probabilistic matching is relatively non-invasive from a privacy standpoint, Apple has taken a very active stance against it as part of the ITP framework, which was designed to bring privacy to the Web, and their IDFA changes seem to be following that same framework. More so, in one of the WWDC videos they seem to consider “fingerprinted ID” as a type of tracking that should not be allowed if the user didn’t explicitly opt-in.

Some of the most powerful players in the mobile marketing ecosystem are the SANs, the self-attributing networks. There is an existing precedent for Google using fingerprinting to match their UAC search campaigns which originate in a browser with app installs (if Facebook and others follow suit, that could establish fingerprinting as an acceptable method). However, while Google uses this method to attribute users and build cohorts internally, MMPs and advertisers do not get access to the raw data and can’t use it for deduping attributions.

The main advantages:

  • Fully integrated with the existing ecosystem
  • User-level attribution
  • Decays quickly, so more privacy-safe than persistent unique identifiers

The main limitations:

  • Apple does not accept this method, so this is not a compliant solution
  • SANs may not support it or they may prevent deduplication of attributions with other networks
  • IP-based methods are more exposed to fraud
  • Accuracy isn’t perfect, though a satisfying level of accuracy can be achieved

2. Privacy-preserving attribution with supply-side consent

Right now, almost everybody is trying to decipher what Apple meant by this paragraph in the App Store privacy policy:

Some believe that means you can access the IDFA as long as it stays on the device, but that’s technically inaccurate based on the current iOS 14 beta (and yes, we checked the code). Others have come up with interesting ideas to run code on the device itself as part of the mobile measurement partner SDK (software development kit) inside the advertiser app. The idea is that this would send data which is sufficient for accurate IDFA-based attribution without actually sending the IDFA itself.

The main incentive for these suggestions is to solve inaccuracies when using IP and probabilistic matching. That’s potentially attractive.

But there is a challenge: all those methods share a core assumption that user-level attribution is permitted even without the advertiser receiving consent from the end user.

One way to do this is to calculate an attribution hash (a cryptographic transformation) on the IDFA and IDFV and send the hash to the MMP. This would ensure that the MMP and the advertiser would not learn the user’s identifier. If this happened, attribution could still be completed by the MMP using the publisher-side clicks, which do include the IDFA, thanks to “supply-side” consent (the end user approving it on the publisher app).

How this would work:

  • You are in NewsCo’s app
  • You approve NewsCo’s use of your IDFA
  • You see an ad you like for an app install for CoolNewGame
  • You click on the ad
  • Newsco records a click that is associated to your IDFA and sends it to the MMP
  • You install CoolNewGame
  • The MMP SDK sends an attribution hash to the MMP’s server
  • The MMP matches the attribution hash to the click

There are other options as well. Another potential solution involves making changes to SKAdNetwork to supply IDFV in its notifications if the user opted-in on the supply-side. And yet another uses advanced mathematical constructs such as Zero Knowledge Proofs to perform attribution while preserving privacy.

All these suggestions rely on Apple accepting the concept of single-side consent and assume that supply-side consent is sufficient for attributing user-level installs on the demand-side. 

This is both a significant leap of faith and an interesting philosophical question since the user didn’t approve tracking on the demand side.

Additionally, each of those suggestions would require some code change on Apple’s side. As an example, calculating the attribution hash requires direct access to the IDFA without consent, which is currently not supported by Apple. This requirement is also quite naive and prone to malicious abuse. If supported, it would require Apple to provide additional functionality to support the solution without actually giving direct access to the IDFA, most likely by providing the attribution hash directly via the AppTrackingTransparency framework.

The main advantages:

  • Compatible with the existing ecosystem
  • Perfect accuracy, similar to IDFA-based matching

The main limitations:

  • Requires substantial policy change by Apple
  • Requires some application/code changes on Apple’s side
  • Provides weaker guarantees for maintaining the end-user privacy

Summary

There’s a lot going on and much of it is very technical.

We’re committed to supporting SKAdNetwork. At the same time, we are constantly thinking about new solutions and analyzing every suggestion by every player in the industry. This is a time of massive change, and we must be agile in response to new ideas in order to provide the best possible solutions for our customers.

With all of that in mind, we have to stay in compliance with Apple’s new rules and regulations. At Singular, we’re investing in multiple options in order to ensure that our customers will benefit from the highest quality of data possible, both in the short and long term.

Join the conversation

Have questions? You’re not alone. Our entire industry needs to adapt to these new privacy enhancements in iOS 14, and we have to do it fast. We invite you to join our community coalition, Mobile Action Privacy (MAP), on Slack to connect with other thought leaders, ask questions, and share ideas.