Author: John Koetsier

There’s a bit of a perfect storm hitting mobile apps, mobile advertising, and growth marketing in general. Three massive factors are changing everything:

• Privacy
• Economic downturn
• Post-covid rebound

These three forces, Eric Seufert says, “present immense challenges to performance marketers at the moment.” The worst part: they combine with each other in unpredictable ways to not just make mobile marketing harder, but to also render recent experience less useful as a predictor of success.

The 3 massive converging changes in 2022

First is privacy.

Basically everyone in marketing and mobile user acquisition knows that the future of both measurement and optimization is privacy-focused, more aggregate than granular, more fuzzy than deterministic, and more multi-modal than it’s been in the past. That’s happening by law, and it’s happening by Apple and Google edict, with App Tracking Transparency, SKAdNetwork, Privacy Sandbox for Web, and Privacy Sandbox for Android.

At the same time, macroeconomic indicators aren’t good. 

Adtech firms including Unity and AppLovin are laying off people, Meta and Twitter are freezing hiring, Tencent and ByteDance are axing thousands, and long-time top companies like Netflix are trimming headcount. Leading mobile-focused companies like Noom and Klarna and many others (Ritual, Numbrs, even Niantic, the studio behind Pokémon Go) are also freezing projects and/or also laying off employees.

To add to the chaos, the digital/mobile boost that Covid provided for the tech industry in general and the mobile industry specifically is going away as people resume normal lives and prior patterns. Some of what mobile apps gained will be lost. Not everything that people did in lockdown or under restrictions will remain as life patterns and mobile usage change. And when you add economic pressure to a reversion to previous norms, it might be harder to justify a nightly habit of ordering delivery.

Or buying up in the mobile game you no longer spend 10 hours a week on.

And those kinds of changes have impact.

So what do you do? What’s your plan?

Well, most mobile marketers will focus on continuing to fly the plane where it’s going, and there’s a lot to be said for that strategy. The plane needs to stay in the air, after all: you’ve got growth targets, and they’re not going to hit if you dream (or worry) about the future and forget about now.

And yet …

How should a scaled-up performance marketing organization deal with the emerging realities of privacy? Probably by keeping one hand on the wheel of now (ok, mostly two) but probably also by casting an occasional eye forward and planning to be well-optimized and well-adapted for a world that is rewriting the rules of growth.

Of course, it’s not simple. 

Because at the same time as privacy changes make it harder to find new customers, the one-two whammy of macroeconomic changes and post-Covid rebound might just result in fewer potential new customers being available, period. So you need a plan to continue to acquire valuable customers effectively even when it’s harder.

Here are 12 places to focus and ensure your marketing and advertising future, whichever ways the privacy path goes, and whatever happens with the economy or people’s changing habits.

1. Biggify the organic footprint

Organic won’t replace paid advertising unless lightning strikes and you become the next Pokemon Go. But it will be additive, and quality organic marketing tends to participate in a virtuous circle with paid marketing, offering an amplifier effect.

That starts with your own branded content, but it’s infinitely more scalable if you architect your product and its core loop to make it easy, appealing, and rewarding for your players, customers, or users to share their own, generated via your platform.

“Content is fire. Social media is gasoline.”

– Jay Baer

If you do the right things in content — whether your own or your customers’ — unexpectedly good things can happen.

The good news is that done well, organic marketing doesn’t just amplify paid, it brings in “free” customers and users on its own, making it a valuable part of a recession-proofing strategy. But as anyone with any experience knows: these users aren’t free. It costs time and money to build and deploy organic marketing strategies. When it hits, however, these can be the cheapest (and best) new customers you’ve ever had.

2. Become viral-worthy (funky/interesting/odd/amazing)

Organic is great. Organic plus viral is amazing.

Planning to go viral is a dumb business strategy, mostly because aside from a few very talented, connected, and tuned-in influencers, it’s incredibly hard to make something go viral. It happens all the time, and the characteristics of what goes viral often show similarities, but making your weekly blog post catch social fire is ridiculously hard.

And yet, the potential rewards are great:

“Advertising brings in customers, but word-of-mouth brings in the best customers.” 

– Jonah Berger, author & marketing professor

So design your app, structure your gameplay, record your thoughts, publish your content, and enable your users to publish their experiences in ways that maximize your likelihood of achieving virality. Then hope for the best … while potentially upping the odds by seeding it with friendly influencers.

3. Target with creative

I’ve heard two top marketers in the last month say they were “targeting via creative” after never hearing that phrase in a decade or more. 

Post-privacy, when you can’t narrowcast to a tightly defined behaviorally-created audience, build creative that answers questions, as Activision senior growth manager Thanasi Chalkiadakis recently told me. Creative that answers questions like how does this game work, or how does this app help me, or who else uses this service tend to bring in better-qualified users, he says.

“Build creative that answers questions.”

– Thanasi Chalkiadakis, senior growth manager, Activision

Answer the right questions, and you’ve targeted even ads that can’t really be targeted, at least not the way mobile marketers have become accustomed to over the last decade. And likely you’ve done it cheaper than truly behaviorally-target ads of 2020 … which is a good recession-proof strategy.

Also, target via creative by showing the kinds of people who form your best customers. Maybe they’re into sports, or maybe they’re big football fans, or maybe they’re over the age of 50. Whatever: target with your creative as much (or more than) your audiences, your contextuals, and your channels, and you’re more likely to get your messages in front of people who you want in your app.

4. Ensure you have clean data and good ingestion protocols

This is always good advice, but if you’re going to have less data, or more aggregate data, or more diverse sources of data, you really, really need ensure it’s the best quality it can be. That means ensuring you import clean, standardized, normalized data into your internal BI systems as much as possible.

Everyone knows this matters.

But so many don’t invest the time or energy in ensuring it’s done and it keeps getting done, as partners and technologies and schemata change and evolve.

So it also means ensuring you have defined processes in place for handling that data, enriching that data, and putting that data to work. Easy access to actionable insight is hugely important for growth professionals, and making sure you have that is a job that never really ends.

(Shockingly, I know, Singular can help with this … see Singular ETL.)

5. Test contextual targeting

It might seem super-lame and you might be wishing for good old behavioral targeting mechanisms to come back, but in some scenarios contextual targeting is going to be what you’re left with.

“Because IDFA is more or less gone, you can only target a very small fraction of iOS users using identity. So, you know, that ship has kind of sailed at this point. So your best bet at this point is to work with a vendor that can really look at a lot of different signals, including contextual signals, and really use machine learning to target the user. You can’t rely on unique identifiers to target users on iOS anymore, you really have to be using probabilistic and machine learning based approaches to target users.”

 – Anurag Agarwal, VP of product at Moloco

There are definitely some challenges. Some studies suggest there’s a 100X difference in click-through rate between contextual and behavioral targeted ads while only costing about 2X as much. 

That said, however, it’s actually not as horrific as modern marketers who never had to operate contextually might think. That 100X difference compares retargeted ads with regular old display ads on the web, which is an unfair comparison.

Retargeting — where possible — is always going to have significant advantages in terms of awareness, brand, and propensity for conversion compared to a basically invisible banner ad.

Picking the right units that invite engagement is going to lead to much better results.

And by picking the right context, you can infer behavioral intent even when you literally know nothing specifically about an ad viewer. After all, someone watching a rewarded ad unit in a game is … at least … someone who plays games. By knowing the game, the style, perhaps even the level of the game the player is seeing an ad at — all things a supply-side partner might be able to offer without tracking and without device identification — all help narrow down exactly what kind of player this person might be.

6. Engage microinfluencers

You might not have touched influencer marketing with a stick three years ago, or even two years ago.

Now with tracking challenges elsewhere and the continuing strength of Tiktok, that’s changing. More and more performance marketers are turning to platforms that automate influencer marketing campaigns with predictable — even if they’re not fully trackable — returns. You can go it alone, but it’s a lot of work and you won’t really have access to the metrics that will make it scalable.

“Don’t get distracted by the big follower counts. You might have a 2 million follower TikToker who’s charging you $3,000 for a post, but if you actually look at their videos, they’re only getting between like, you know, 10,000 views or something like that.  Where the sweet spot is is you want to flip that ratio.  You want to find smaller influencers who are newer, who have much larger engagements.”

 – Aidan Quest, JetFuel

Performance marketers have traditionally hated influencer marketing due to its inherent lack of trackability. Intelligently done, however, you can track results with incrementality, by time-specific analysis of installs, by integrated coupon codes or offers or specific landing pages … the sky is the limit.

7. Widen your data gathering

Three years ago mobile marketers could live dumb, lazy, happy, and successful lives. IDFAs and GAIDs were free for all, and as long as you had the infrastructure to collect, process, and share them, you had pretty much all the information you needed to run user acquisition at scale.

(OK, the first line of that paragraph is an exaggeration.)

Now there’s so much more that growth marketers in the mobile space have to learn, know, and add to their ever-more-complex data models for driving growth.

1. IDFA (you still get some)
2. GAID (for now)
3. SKAN
4. Privacy Sandbox on Android (soon)
5. Cost/campaign data from partners
6. In-app data
7. Store data
8. Broader economy data (for incrementality/MMM)
9. Broader political/social/entertainment data (for incrementality/MMM)
10. And probably more

With less certain data to drive both measurement and optimization, you need more data to power fuzzy models for where to up the ante and where to cool your jets. It’s getting harder, and the new deterministic-but-not-granular measurement models from Apple and Google are also harder to learn, harder to implement, harder to operate, and harder to optimize.

A wider funnel for data just might also open your digital eyes on who could be a customer or user, and where they might be found.

But if you do this, you also need to …

8. Invest in data science

Fuzzy data from multiple sources is harder to distill into action than simple last-click attribution data. Get comfortable with truthiness rather than truth.

And invest in data science that delivers directionally-accurate insight versus what you might have expected previously: a 100% clear and obvious answer to the marketer’s perennial question: what do I do next?

Along the path, study and become comfortable with confidence intervals, because they just might become your new best friends.

9. Invest more in onboarding

When you could find a new user by throwing a $5 bill at Facebook and you had tens of millions in venture capital overflowing out of your bank account, you could afford to not worry too much about leaky buckets or churn. It wasn’t smart, but sometimes it was expedient and made sense in a very dot-com era burn-the-world-down kind of way.

Not anymore.

And that means UA experts now need to be product experts too.

“No one should know what happens [from] D0 to D30 quite like someone in UA, because you need to know what you’re showcasing in the app,” says Thanasi Chalkiadakis, who leads growth for the mobile version of Call of Duty for Activision. “You need to know what your big KPIs are. So you should know those tutorial moments, those gotcha moments, all of that. You should have all of that baked out, but past D30.”

Onboard users with care. Personalize the experience for them as you teach them your app and they teach themselves to your app, adapting your user experience to the best possible flow for each persona that fits.

At the same time …

10. Adopt lifecycle marketing

Adopt lifecycle marketing and start thinking about the value of a customer, player, or user who sticks around for 10 months, not 10 days. Maybe even 10 years.

That’s going to be different for different apps — especially hypercasual apps — but our new privacy and economic conditions privilege different kinds of apps, and different kinds of growth strategies.

When you could cherry pick AAA new users out of the crowds, maybe hypercasual made sense. Maybe a churn-style growth mechanic made sense. But when you have to use the fish net to capture whole crowds of hundreds of users of relatively unknown value, maybe you have to focus more on identifying the ones who will stick and nurturing them for the long term.

Which, of course, also means product has to change.

11. Become expert at each platform’s preferred ways of attribution

I get it. You hate SKAdNetwork. And you’re not going to love Privacy Sandbox for Android when it comes out.

They’re complicated. They’re not as good, for mobile growth purposes, as IDFA and GAID and the existing industry infrastructure. And they’re tough to figure out.

Here’s the reality: they’re what you’ve got. Apps and marketers that are leaning in are finding ways to generate high performance returns. You can too, and by design, the platforms are more and more ensuring that SKAN and, in the future Privacy Sandbox, will be the primary ways to get deterministic data on installs.

12. Open Pandora’s box (get more creative)

Every marketer that is successful has learned over time to find an edge. Find something that tipped the scale in their direction.

Keep working on that.

One way: super-creative marketing campaigns that demand attention along with the complex math and modeling that evaluates the results of that attention.

Now it’s time to empty the tank. Most creative one wins, whether it’s the privacy game or the bigger piece of a shrinking pie.

Everyone can market in a growing economy with defined processes for success. Now it’s time to see who can outcompete in a difficult economy with challenging new processes that we’re still inventing.

Here’s your chance to shine.

13. Finally: check out our upcoming SKAN 4 webinar

SKAN 4 will soon be the way to measure marketing impact and optimize campaigns on iOS and optimize. It’s a big improvement in a lot of ways over the existing SKAdNetwork, but it will also add a ton of complexity.

I’m doing a deep dive into SKAN 4 with Singular CEO Gadi Eliashiv: we’ll be talking about how user acquisition and growth will be impacted by SKAN 4, and we’ll be sharing how SKAN 4 could impact your data collection and growth strategy.

What is the future of user acquisition measurement?

There are so many things we take on faith in life. When we step out of bed in the morning, we take it for granted that our feet won’t sink through the floor. When we hit the gas on green, we have some level of trust that another car won’t sideswipe us. When we train for a job, we believe that we’ll develop the skills and knowledge needed to perform our roles.

But not really in mobile marketing. Not really in user acquisition. Not anymore.

Over the past few years, that bedrock faith in the nature of reality and how things work in the mobile growth space has been shaken multiple times. Buckle up, because the next few years with SKAN 4.0 and Privacy Sandbox on Android promises to continue the rapid pace of change.

So what does our future hold?

We recently took some time to figure that out with some mobile all-stars in a webinar on the future of user acquisition. Our all-stars included:

• Sherry Y. Lin, Group Manager, Marketing Technology for Lyft
• Cody Christie, Director of User Acquisition & Marketing Technologies for Riot (who unfortunately had to cancel as he fell ill)
• Shamanth Rao, Founder & CEO, Rocketship HQ
• Nick Blake, VP, EMEA, Liftoff
• Gadi Eliashiv, CEO and Co-Founder, Singular

Here are some of the top insights from that panel.

1. 87% of mobile marketers are not digging into Privacy Sandbox just yet

Everyone has a lot to think about and work on. What we saw in the first months of SKAdNetwork on iOS is that many marketers avoided implementing SKAN as they ran ad campaigns based on fingerprinting or old marketing intelligence. But early adopters who got it right — like Rovio — reaped the lion’s share of rewards.

Even today, only 13% of marketers are actively checking out Privacy Sandbox. I get it: it’s in the far-off future. But so was SKAN at one point.

The insight here is simple: 

Many marketers can’t or won’t make the effort to adopt new technologies, even ones mandated by major platform owners. Be one of those who does go early, and you’ll be better adapted to changing conditions.

2. Targeting is a very significant difference between Apple’s SKAN and Google’s Privacy Sandbox for Android

Most marketers have been pleasantly surprised by Google’s Privacy Sandbox for Android. Most have also been happy about the announced changes to SKAN 4.0, while wanting more.

But there’s a massive and foundational difference between them: targeting.

“[Google’s] solutions address both attribution and targeting, whereas SKAN only addresses attribution,” Lyft’s Sherry Lin said on the webinar.

Clearly, we can have a debate about Topics API in Privacy Sandbox for Android, how good it is, how fine-tuned or not it is … but the fact remains that there’s a privacy-safe interest-based behaviorally-acquired targeting mechanism in PSA and there’s no equivalent functionality in SKAdNetwork today or the SKAN 4.0 specification for tomorrow. 

(Plus, there’s even retargeting functionality built into Android Privacy Sandbox via Fledge.)

As things stand, therefore, Android marketing will have a leg up on iOS until Apple does something similar (if it will).

3. SKAN 4.0 is not a complete rewrite

It’s been hard enough for marketers to re-instrument for ATT and SKAN over the past year. The good news is that SKAN 4 is not only better, with more detail and more insight, it’s also an evolution not a revolution.

“Advertisers are not gonna have to blow everything up and start again … and we’ll get deeper insights than we currently have with SKAN,” says Liftoff’s Nick Blake. “So it’s a step in the right direction.”

There are always wrinkles, however, and while Apple will likely update devices at a pretty significant rate as per usual, there is now an ecosystem component to SKAN capabilities and functionalities, and partners will also need to upgrade their own tech stacks to support the latest versions. 

Until that happens, expect some speed bumps.

4. Privacy doesn’t have to suck (from a marketing perspective)

SKAN has been a long tough road. 

As of today, just 9% of mobile marketers that we surveyed during a recent webinar said their success with SKAN could be rated as “good.” 46% said it was passable — not exactly the kind of grade you want to bring back to your boss during bonus season — and 50% said “we really have work to do.”

Which, of course, is code for: it sucks.

But Google has shown that privacy isn’t necessarily doom and gloom for digital marketing. There is a way to make performance marketing work in a privacy-safe manner.

“My first reaction was like, ‘Wow, this doesn’t suck. It’s actually not that bad,’” says Singular CEO Gadi Eliashiv.

Of course, Google had a second-mover advantage, and we’re seeing the results already in SKAN 4.0, which addresses some of the points where Privacy Sandbox is better. But the good news is that PSA doesn’t suck, and SKAN is looking like it will suck less.

5. Multiple postbacks will be a wonderful thing

GAID and IDFA enabled extremely long-lived cohorts that gave marketers immense amounts of data about users, campaigns, ad partners, and more. That’s gone for good, essentially. 

But Privacy Sandbox enables multiple postbacks, and SKAN 4.0 will as well. And that’s huge for marketers to be able to understand, in a privacy-safe way, the value of their advertising dollars.

6. Platform privilege will still be a concern

Apple requires third-party advertisers to use SKAN, but uses its own internal attribution methodology for Apple Search Ads and gives campaigns there access to more data, thanks to first-party privilege. That’s controversial, of course, but old news by now. 

The question today is whether Google, as the author of the Privacy Sandbox on Android spec, will privilege itself in any way.

“I’m fairly certain that Google is going to privilege its own inventory, its own network,” says Shamanth Rao of Rocketship HQ. 

I haven’t seen it yet in the spec, but you almost have to assume that a platform is going to build something that works best with its tools and processes, no? So this is something to keep an eye on.

7. Timing matters

I said above that getting a headstart on competitors is important. Learning how to optimize growth strategies faster than them is a competitive advantage.

But what’s the right timing?

Going too early might be wasteful of resources, especially given the fact that specifications are guaranteed to change over time.

“When is the right time?” says Lyft’s Sherry Lin. “When is a product mature enough for us to go in there and test? If it’s too early, the product is not ready … [the key is] really knowing when is the right time to test and also making sure that we line up the right engineering and data science resources to do the evaluation justice.”

And that, of course, is a judgment call. Perhaps the best solution is to align yourself with the right partners who are investing massive resources to get it right … and lean on their expertise.

8. Retargeting will still be a challenge

While there’s a built-in mechanism on Android for retargeting in FLEDGE, there’s currently no line of sight to similar functionality on iOS.And while some in the gaming community — like Rovio in our most recent webinar — haven’t found retargeting fruitful, others like Lyft do.

“Half of our audience can’t be retargeted because per ATT, we can’t be sharing any of these identifiers, not even server-side ones like email addresses and phone numbers unless we get user consent,” says Sherry Lin. “So our retargeting program has taken a really big hit.”

There’s other privacy-safe options — Lyft is looking at Dstillery and seller-defined audiences — but it’s not quite like the old days of IDFA.

9. Incrementality will have a seat at the table

It may not be as fast as you like or as precise as you could wish for, but incrementality will be a part of the overall solution that marketers need, especially marketers with a well-known brand.

But there are limitations, and they’re not likely to go away soon:

“Incrementality is super useful, super important, especially for advertisers, again, that have high mindshare,” Lin says. “On its own, though, it’s not super helpful for performance marketers because the data comes in slowly.”

It’s slower, it’s expensive, and there are multiple non-obvious ways to mess it up, but done well, there’s value. Which is probably why only 13% of mobile marketers that we surveyed are actually using it:

The majority, 55% are not, and 32% are dipping their toes in the incrementality waters. That last number is likely to rise in the coming years.

10. Media Mix Modeling is really hard in chaotic times

Media Mix Modeling has some potential as well, and has been used for decades by big brands. But it relies on some level of stability in the world.

And that’s been hard to come by recently.

“No. I don’t think that would be applicable,” Rao said when asked if MMM would be useful during Covid. “Because again, I think there has to be some precedent.”

That’s tough news, because the turbulence of Covid times hasn’t really settled down at all, with economic and political chaos globally at the moment.

Still, there’s some potential here. And it might even hit a Singular product at some point:

“It’s not gonna replace your signal from SKAN or GAID, or Privacy Sandbox, but it’s something else that you use to inform decisions,” says Singular CEO Gadi Eliashiv. “And it’s actually something I foresee also becoming more prevalent in platforms like Singular and other platforms.”

11. MMPs are shock absorbers

We’ve had a ton of change over the last year. That’s only going to continue over the next few, as Apple iterates SKAN and Google implements and releases Privacy Sandbox on both web and mobile.

“Embrace the change and don’t get too comfortable,” says Liftoff’s Nick Blake.

That’s good advice, but it’s not easy to follow when your livelihood depends on managing and improving growth rates. You have to fly the plane, after all, while also fixing it. 

Good news?

There is a pillow for this space between a rock and a hard place, however.

“I kind of see us as … the shocks on the car,” says Singular’s Gadi Eliashiv. “So the road is really bumpy. And, you know, I don’t want to say the MMPs are the springs, but you kind of wanna make the road ahead for the marketers a bit more stable, right? And so our job is to make that simple, to simplify, to absorb all the changes in the tabulations, and make it a bit more simple.”

12. Fuzziness is your friend

Whether accurate or not as a measure of marketing impact, last-click measurement based on IDFA and GAID was hard data. That’s going away.

“Embrace fuzziness,” says Rao.

There are going to be many more sources of marketing data that you’ll need to take into account in the future, including:

• IDFA where available
• GAID until gone
• SKAN
• Privacy Sandbox
• Cost & campaign data
• In-app analytics and insight
• Store data (App Store, Google Play)
• And much more (see here for a more comprehensive list)

Melding that all together into actionable insight will be critical, but you’re not going to get 100% certainty on very much. Which means you’ll have to embrace fuzziness.

Watch the entire webinar yourself

You’ve read some of the highlights, but there’s much more insights in this packed webinar. It’s still available on-demand, so you can enjoy it at your leisure.

And if you’re looking for a new set of shocks for your mobile marketing, book some time with a Singular expert. We’d be happy to listen, learn, understand, and suggest.

What can we learn from $500 million in retail app user acquisition spend? 

• Mobile marketers in ecommerce verticals don’t spend a lot on re-engagement
• Retention is highest in summer and lowest in winter
• Organic vs paid ratios are significantly different on iOS and Android
• And much more …

Free lunch (or, shopping apps ad spend data)

I seek data like a hungry carnivore on the African savannah hunts wildebeest, so when I sniffed out a cleansed, normalized global dataset on shopping apps that I didn’t have to mercilessly beat out of some overworked and stressed out analyst, I howled like a hyena stumbling across an unexpected protein-pack airdrop from heaven.

Singular CTO Eran Friedman assembled the dataset, doubtless for private use in his dark arts of mobile growth, and you are the beneficiary.

Some notes before we jump in:

1. It’s a year-long, global dataset
2. It’s from 2021, the year of the change (in other words: ATT)
3. It shows seasonal adjustments as well as some of the impact of iOS 14.5
4. It highlights how high-growth apps with significant marketing investment are seeking growth, not necessarily the mobile ecosystem as a whole
5. There’s much more we’d like to learn and recent ecosystem data to review: keep tuned

Let’s dive in.

Retail apps don’t spend much on re-engagement advertising (except prior to holidays) 

Retail app marketers focus on new users, not re-engaging existing or lapsed customers. For most of the year, more than 80% of their paid marketing spend is directed towards hunting new customers, not farming existing users.

The only time it drops below 80% is — you guessed it — in December, when every shopping and store app wants to remind customers that they have purchased in the past, that there are deals now, and that they have just the right app to do it with.

There is a giant caveat here, of course. 

Unlike many other verticals like gaming, e-commerce apps are much more likely to have multiple personal identifiers for their customers. When you buy something with a credit card in an app, you’re almost guaranteed to provide your email address during the account creator or checkout process, and you are likely to be asked for your phone number so you can get SMS alerts on shipping status.

Which means that customer retention, re-engagement, and nurture has shifted, for a lot of shopping app growth teams, out of the paid acquisition team and into the engagement/retention/live ops/nurture team. Or at least is being targeted via owned channels, not paid.

Speculation: 
It’s possible that Android Privacy Sandbox, which will offer Fledge for built-in on-platform re-engagement, could boost this number on Android. Time will tell. Also worth noting: in a recent Singular webinar, one of the things top marketers want most from iOS is a privacy-safe way to do effective retargeting.

Spend distribution by month follows slightly different patterns on iOS and Android (but there are 2 caveats)

No, mobile advertising spend doesn’t peak in the holiday months of November and December. Actually, it peaks in the other kind of holidays:  the spring/summer months of May, June, and July.

In addition, mobile user acquisition marketers spend just a little differently on iOS and Android. Here’s the pretty chart with spend for each platform as a percentage of total spend for each platform:

It’s a little challenging to see the pattern in a chart like this, so here’s the data simplified in an overlapping area chart:

Now you can clearly see the head-and-shoulders pattern for Android, which is a bit more pronounced than for iOS, where the “head” is somewhat spread out and less pronounced. 

There are a couple of big caveats here, however.

See that divot out of the top of the iOS “head?” Apple’s privacy-focused ATT hit in summer 2021, and ad spend fled iOS to Android. That divot is pretty closely aligned to the initial impact of iOS 14.5 and subsequent iOS point releases that Apple released to more iPhone owners. And the growth out of the divot is likely some of the biggest advertisers starting to figure SKAdNetwork out.

Second, a single year doesn’t make a lasting, bankable pattern, so as previously mentioned, stay tuned for more data as we come to the end of 2022. We’ll have to see this pattern for the current year before we can chalk it up to a potentially durable platform difference between Android and iOS.

Retention is highest in summer, lowest in winter

As you know, retention is a sad, sad story in the mobile growth space, with the average app retention a few short weeks after download typically in the single digits. Interestingly in the shopping apps vertical, retention differs significantly from month to month, and season to season.

For instance, users you acquire in January have truly atrocious D7 retention rates:

• 8% on iOS
• 4% on Android

It gets better, however, with nearly double the retention rate on iOS in August and September, and nearly triple the retention rate on Android in September and October.

There’s always been a perception that iOS retention rates are better than Android, and according to the letter of the law, that’s true: every single month, iOS shows a higher D7 user retention rate. The big BUT here, however, is that the rates really don’t vary that much between the major mobile platforms. Besides January and February, when iOS literally doubles up on Android, for the other months it’s not that different. iOS sees 20-30% higher retention than Android, sure, but it’s not double or triple the rate.

Smart marketers correlate that difference with the cost of user acquisition difference and keep on trucking.

Interestingly, retention and ad spend levels are somewhat positively correlated: app publishers spend more in the summer months; retention levels are highest in late summer and early fall. On the one hand, that’s counterintuitive: more spend, more noise, more activity, and more competition in user acquisition would seem to decrease retention as retail apps poach each others’ customers. On the other hand, that very flurry of activity could keep apps top of mind and result in de facto even if unintentional engagement or retention campaigns.

The lion’s share of spend is on Android

Ad spend globally for iOS and Android has been fairly stable for the past few years at about 60% for Android and 40% for iOS, at least among Singular’s customer base which skews North American and European. That changed with Apple’s release of App Tracking Transparency in iOS 14.5:

Immediately post ATT Android ad spend share grew, and not because iOS spend dropped significantly. iOS spend decreased slightly from May to following months, but only about 5%. Android ad spend, however, ramped 23% from May to April, grew a bit more in June, and then stayed near that level for the rest of the summer. It was measurable, it was familiar, it worked according to the traditional rules of mobile user acquisition, and as such it sucked up ad spend (and acquisition costs skyrocketed).

Everything evened out somewhat to historical norms in platform ad spend as the year came to a close, likely due to significant brand spend that realized that despite measurement woes, people were still on iPhones, still using their iPhones, still installing apps, and still buying things.

Organic vs paid installs: iOS vs Android

We’ve seen a long-standing 80/20 run in general for paid versus organic installs for high-growth apps. Clearly, this discounts viral apps that blow up and generate 10s or 100s of millions of app installs in just a few months. But for most high-growth mobile-centric companies, paid acquisition has historically been about 80% of their app installs.

That looks to be the case recently as well, but the overall chart is deceiving:

The Android-only chart, however, shows a fairly significant difference:

And the iOS chart shows the opposite movement:

Clearly, the year ends trending to 75-25 on Android as a paid-organic split of user acquisition, and a fairly different 85/15 on iOS. Assuming that measurement stayed stable on Android (a fairly safe assumption) we’re seeing a real effect there, possibly caused by an organic boost following months of increased Android ad spend. On iOS, we see a three-month blip where organics decrease fairly instantly and alarmingly, followed by a gradual return to near (but not quite at) historical averages.

We’ll need to pull more data to find where this is going now.

Data, questions, and answers

As always, data answers some questions and poses new questions, for which we need new data. I will track down some data analyst (or hound Friedman) and answer some of those.

Plus pose additional others, I assume. If you have any, ping me on Twitter.

And, as always, if you’re looking for help in getting measurement right in these crazy and chaotic mobile growth times, book a demo and tell us what you need.

We seem to oscillate between cycles of expansion and contraction in what we think an app is, pulled by the demands of growth, the limitations of handheld devices, and the opportunities that blossom when apps successfully transition into platforms. Which is interesting to consider as we think about what TikTok’s long-developing but still nascent pushing into gaming means for app publishers.

And, of course, when we think about what an app is, and what a game is.

A decade ago, an app wanted to be simple. Plain. Brutally direct. Developers would compete with each other to remove every last potentially non-essential feature. Still to this day if you want to post your photos to Instagram in anything other than one at a time, you’ll need to download a separate app — Layout from Instagram — in order to format multiple photos in a pattern and satisfy your thirst for likes.

None of it really feels like social media anymore

Twitter is a news and opinion app

Snapchat is a messaging app

TikTok is an entertainment app

LinkedIn is a professional network app

Pinterest is an inspo app

IG and FB are… actually, I’m not even sure what they are anymore

— Matthew Kobach 🍕 (@mkobach) May 28, 2022

Then as screens got bigger and WeChat showed how to become a hardwareless platform, features started coming back and apps started getting bigger again.

Just look at Facebook

Today while Meta (the company) is working hard on inventing the future of digital existence while also just coincidentally owning the platform it runs on, Facebook (the app) is:

• Videos/movies (must beat YouTube)
• Reels (gotta win back those young-uns from TikTok)
• Audio chat rooms
• Video hang-out rooms
• Streaming (Twitch, where is Twitch?)
• Group discussions
• Live video
• Stories
• Marketplace (Craigslist? eBay? What are these things?)
• Pages (Geocities for brands, or is that too harsh?)
• Neighborhoods
• Shop (hello, Amazon)
• Events (what, Eventbrite did you say?)
• Dating (match that, Match)
• Games (sorta)
• Payments (PayPal/Venmo lite)

Essentially, everybody wants to be WeChat, baby! 

With instant messaging as the initial foundation and successive layers of social networking, gaming, payments, shopping, third-party services and more being added over the decade of its existence, WeChat makes not just $17-20 billion on its own social services, but likely billions more in tiny increments via small slices of the $400 billion transacted annually over its platform. Plus, of course, whatever amounts 900 million people using WeChat Pay send over the service.

Platform > App.

And then you’ve got TikTok

TikTok is hot off literally the most profitable quarter in mobile app history with $840 million of revenue in Q1. (OK, let’s not make Amazon choke on its Blonde Vanilla Latte #6 Starbucks in a sudden fit of laughter: this counts in-app purchases only, and it’s a very small fraction of the merchandise volume that continually gushes through the Amazon mobile app in even a seriously bad quarter.)

That number will hit $1 billion quarterly very soon: perhaps this very quarter.

And it’s built on the backs of 1.6 billion monthly active global users, a very respectable and still fast-growing fraction of the global population, who spend 20 hours a month in the app. (Except when, like me, they occasionally drown in guilt over the massive time suck TikTok has become when blog posts MUST BE WRITTEN and tearfully delete it, only to re-download a few weeks later when the guilt wears off and we are fresh with new resolve to not let this one app own our lives. This time, at least.)

TikTok is gearing up for gaming, but it’s not your daddy’s games TikTok is looking at.

Not your games either.

The kinds of games TikTok is looking to smush into dance/cat/crash/sports video clips aren’t just games for the sake of games (which would be lame, and accessible elsewhere, and a copy-paste waste of time, and likely a losing endeavor). Rather, they seem to be games that are entwined with the core loop of the TikTok app itself in unique and interesting ways.

One type looks to be a game to play while engaged in a livestream.

Which is a complete stroke of genius.

Livestreams can be cool, but only really if you’re unbelievably passionate about Random Influencer #9 who is streaming her cat playing with an octopus in Second Life. Sure, you hang around for a minute or two, but eventually, the siren call of the thumb scroll is too loud to ignore, and you’re on to the next hit of dopamine. Sayonara, octopus.

But if you could:

• play a game related to the livestream …
• maybe with the other watchers of the livestream …
• that even potentially engages with the creator who is hosting the livestream

 … that could be seriously cool. Kinda fun. Engaging. Sticky. And something that hooks you into the content and the community around that creator to keep you there in the livestream … while also providing part of the fun and interest and enjoyment and escape that you’re currently seeking in your 10-minute break from life.

TechCrunch says that could be with screensharing, live feedback that streams to the entire audience, and other interesting features. Other gamelike options include virtual gifting/softcore gambling in something like a 50/50 draw, perhaps where a creator gets 50% of the proceeds and a winner from the audience gets 50%. 

(After platform fees, of course.)

None of that is fully baked, announced, and released, but all of it sounds innovative and inspiring: at least in terms of creating new mashups of media for entertainment and commerce.

The big smush (AKA converged experiences)

And all of it is entirely on point with what we’re seeing elsewhere: converged experiences in shared spaces with multiple uses. Fortnite, famously, for games, with social media of a sort, fun, concerts, movies, and who knows what else.

Meta’s own Horizon Worlds, a Roblox-slash-Mindcraft for adults, fits the bill with options to play, build, create, explore, share, and more. And yep, the concerts and movies and workouts are there, plus Workroom, remote work collaboration that surprisingly doesn’t suck.

(And all of this, of course, is just part of what Meta Quest does.)

It’s all getting kinda … dare we say … metaversy.

Which is of course the directional flow of many apps today and means that not only are they adding different types of experiences but also in many cases adding platform-esque features that they hope to grow. Very few, however, have the cojones to take the next step in connecting divergent experiences from different publishers via access points that would anticipate a still-emerging but true multi-publisher multi-tenant multi-owner open metaverse. 

And, in fact, the technological framework to do so doesn’t really exist yet, unless you consider it to be the internet itself. (Which is not a weak argument, by the way.)

Back to TikTok: what does this mean?

TikTok is growing beyond a niche, but it’s doing so in measured ways that build on its existing user experience and brand promise.

That is inevitably competitive to not just the 800-pound-gorillas of mobile, because almost all of them are attention merchants packaging audiences for advertisers, but also social startups, games publishers, and other verticals. Time spent in one mobile ecosystem cannot be spent in another: there really isn’t a mobile equivalent of the second-screen experience that long-form traditional and smart TV has spawned.

TikTok is winning already, and it wants more.

It wants active time, not just passive time. It wants bigger pots of money for creators that attract more creators who attract more consumers. It wants more payment flow that can be tapped for platform shares that drive gross merchandising volume, so to speak, as well as in-app revenue.

As it does so and reinvents part of what it is, TikTok could also play a large role in game discovery. Think mini-games during livestreams that mirror full games that you can get via a tap. Think livestreams of those games that perhaps you can play along with your favorite gamer?

The virtual sky is the limit.

Smart app publishers will see opportunities in what one of the world’s largest social entertainment apps is building for growing their own impact. Others might see inspiration for unique directions they take their own gaming and social platforms.

• User acquisition on iOS is broken
• Overall return on investment is down 38%, and spend is still down by 25%
• But there is a solution

Privacy isn’t going anywhere, and rightfully so. App Tracking Transparency and SKAdNetwork are here to stay, and while they will likely add new features over time to help marketers optimize user acquisition campaigns, we are never going back to the device ID free-for-all that enabled permissionless tracking and measurement.

At the same time, the reality is that user acquisition on iOS is still largely broken.

Not only has it been challenging for marketers to adapt to the new realities, fundamental metrics and data that advertisers need to run, measure, and optimize campaigns are simply missing. We see the results every day in inefficient campaigns and suboptimal ROI.

Getting expert at SKAdNetwork

So what’s the solution?

Getting expert at using SKAN … and picking the right tool that makes it possible. That’s precisely why we wrote Singular’s brand-new SKAN guide. We call it High-performance UA on iOS using SKAN: How to make it work.

Singular recently released SKAN Advanced Analytics, a new set of features and capabilities that fix the fundamental problems inherent in SKAdNetwork. In a brutally concise way, the SKAN guide walks through each of 7 major problems marketers face when using SKAdNetwork to accomplish attribution.

They are:

1. missing data
2. randomized timing of postbacks
3. limited conversion data,
4. limited time for conversion value updates
5. limited campaign attributes
6. complex conversion value encoding
7. missing cohorts

The guide explains each problem, details the impact, and then outlines a simple and clear solution for every one.

The solution is using SKAN Advanced Analytics

The solution is using SKAN Advanced Analytics to enrich campaign data, decode conversion values to KPIs, estimate performance even when there’s missing data due to privacy thresholds, and provide usable cohorts for long-term LTV and ROAS calculation.

SKAN Advanced Analytics builds on sophisticated data science, modeling, and machine learning to provide reliable and actionable information. The results are impressive: 87% D7 revenue accuracy on average for clients who beta-tested it.

But don’t take our word for it:

“At Space Sheep Games, we depend on a consolidated view of revenue from in-app ads and IAP to understand ROAS. Partnering with Singular has been super valuable as they are at the forefront of the changes in the market.”

Rene Retz, CEO of Space Sheep Games

Growth marketers from Qiiwi Games, Rovio, Space Sheep Games and many more mobile publishers are finding ways to get the near real-time predictive data they need to optimize campaigns on the fly, and the sophisticated conversion models they need to capture all revenue, regardless of source, and calculate ROAS in ways they previously couldn’t.

“Our strategy has been to embrace the change in paradigm, turn this disruption into an opportunity to grow our business, and build a future-proof UA infrastructure. Singular has undoubtedly been instrumental in helping us pioneer the new ways of running acquisition.”

Kieran O’Leary, COO, Rovio

The result is performance mobile marketing at scale and speed that works for marketers and maintains full and complete compliance with Apple guidelines. In other words, privacy-safe marketing measurement that doesn’t suck.

How to get started

You’ll be able to scan it in a few minutes, recognize the challenges, and identify the solutions you need for your growth campaigns. Then you can request a demo and get personalized insight into how SKAN Advanced Analytics can help you deliver the growth you’ve been working so hard to achieve.

“Collecting and analyzing data from SKAdNetwork can quickly become a time-consuming pain. Singular’s SKAdNetwork suite has helped us improve significantly. We’ve been able to optimize data collection and BI models to match our needs and to accurately predict future revenues.”

Marcus Dale, CTO of Qiiwi Games

May 18, San Francisco: Singular announces SKAN Advanced Analytics to restore mobile marketers’ ability to run high-performance, predictable, and privacy-safe user acquisition campaigns on iOS.

Since Apple launched iOS 14.5 with SKAdNetwork for ad results measurement, mobile user acquisition spend is down 25% and marketers’ return on advertising investment is down 38%. SKAN increases privacy, but it has multiple challenges, including:

• Missing data due to privacy thresholds
• Randomized timing of attribution
• Limited conversion value reporting
• Lack of cohort metrics

As a result, it’s hard if not impossible for app publishers to accurately calculate LTV and ROAS. It’s difficult to estimate the value of their new users, making it challenging to optimize ad campaigns and confidently invest in marketing.

Singular is solving those problems with SKAN Advanced Analytics. Using only privacy-safe data that Apple provides via SKAdNetwork and aggregated, non-personal, campaign data from ad partners, Singular is applying data science and machine learning to fill the gaps.

Today we announce the latest update to SKAN Advanced Analytics, SKAN Cohorts, which returns visibility of critical KPIs that marketers currently lack due to SKAdNetwork limitations. The results in real-world tests are nothing short of astounding, reaching 87% D7 revenue accuracy on average for beta clients. 

“Our strategy has been to embrace the change in paradigm, turn this disruption into an opportunity to grow our business, and build a future-proof UA infrastructure,” says Kieran O’Leary, COO, Rovio. “Singular has undoubtedly been instrumental in helping us pioneer the new ways of running acquisition.”

“Collecting and analyzing data from SKAdNetwork can quickly become a time-consuming pain. Singular’s SKAdNetwork suite has helped us improve significantly,” says Marcus Dale, CTO of Qiiwi Games. “We’ve been able to optimize data collection and BI models to match our needs and to accurately predict future revenues.”

SKAN Advanced Analytics helps take the marketing landscape back to the pre iOS 14.5 days with improved reporting accuracy and visibility into cohorted metrics that have been unavailable for over a year. Marketers can now confidently rescale their iOS ad spend knowing accurate measurement is guiding their investment decisions.

Key solutions in Singular’s SKAN Advanced Analytics include:

• SKAN Modeled Metrics
  Using available data and historical trends, Singular models events and conversion values that Apple censors for privacy protection.
• SKAN Smart Conversion Management
  Singular provides 7 different conversion models to maximize data marketers get from the 6 bits of conversion data SKAN returns.
• SKAN Instant Campaign Optimization
  Singular provides near real-time predictive D7 LTV calculations that enable instant campaign optimization for ad partners.
• SKAN Cohorts
  Singular provides estimated cohorts that give marketers visibility into critical measures like revenue and ROAS for campaigns.

Singular’s SKAN Advanced Analytics is built on the first-ever and still industry-leading iOS user acquisition attribution solution that includes SKAN Advanced Reporting, enriching conversion data with aggregated cost and click data from ad networks. Along with powerful out-of-the-box features and easy set-up, Singular’s SKAN solution enables superior analytics and insight that winning mobile marketers need.

“We’re 100% committed to user privacy, and we’re also 100% committed to giving marketers the data and tools that help them drive massive growth,” says Singular CEO Gadi Eliashiv. “We’ve always believed those twin objectives are not in conflict, and now with our advanced AI and data science, we’ve delivered a solution that proves it.”

“At Space Sheep Games, we depend on a consolidated view of revenue from in-app ads and IAP to understand ROAS,” Rene Retz, CEO of Space Sheep Games. “Partnering with Singular has been super valuable as they are at the forefront of the changes in the market.”

More information: Singular’s SKAN solution

 

About Singular
Singular’s next-gen attribution and analytics powers marketers to grow faster by uncovering accurate, granular, and timely performance insights. World-class teams from brands like WB Games, Twitter, Lyft, Rovio, Airbnb, Activision, Homa Games, EA, LinkedIn and more use Singular to make smarter user acquisition decisions and analyze the impact of every ad dollar with full-funnel marketing analytics, best-in-class ad fraud prevention, and automatic loading directly into your BI tools.

• Google updated Privacy Sandbox for Android documentation
• It’s pretty much as we’ve blogged
• Google explicitly acknowledges a role for MMPs in attribution
• Google shares web-to-app conversion paths

Unless you’ve been under a rock or been totally heads-down on your own growth initiatives, you know that Google announced Privacy Sandbox for Android about three months ago, saying at the time that Android’s ATT moment — the deprecation of the GAID — was inevitable.

Now we have more details on exactly what that future of attribution on Android looks like.

MMPs and Privacy Sandbox for Android

Google has broken it down in a simple timeline, and explicitly adds the concept of a mobile measurement partner (MMP) quarterbacking the attribution process.

In the real world, advertisers are running hundreds or thousands of campaigns with dozens of partners, meaning that issues around attribution and incrementality can get complicated. Google’s simplified example shows two adtech partners — one of which could be Google, but not necessarily — both running campaigns for a mobile app publisher.

A smartphone owner taps an ad from both ad networks while also viewing an additional ad later from one of them. 

Interestingly, as we’ve talked about before, adtech providers can set the priority of what Google calls sources: clicks, views, and potentially other pre-install indicators. They are not the only ones, however. MMPs can also set priorities for sources.

---

Remember, in Privacy Sandbox for Android:

Sources = views and clicks

Triggers = installs, sign-ups, purchases

---

In the example above — and likely in many real-world scenarios, the ad networks prioritize clicks overviews, as does the MMP. (Of course this could change in the case of video ads, in-game ads, audio ads, or other formats that are not standard clickable in-app ads.) For each adtech partner, the most recent high-priority source gets credit, so in spite of the fact that Adtech A in the image above registers both a click on Day 1 and an ad view on Day 3, the click gets the credit.

Attribution credit …

As currently set up, Privacy Sandbox for Android assigns attribution credit to each source, even if there’s only one install. In other words, both Adtech A and Adtech B get attribution notifications from Privacy Sandbox. 

Google clearly states that both get credit for the one install:

However, only the MMP — which sees both sides — knows that both ad networks got those postbacks for the same install.

Since Google knows that everyone getting credit for everything is typically bad, ensuring an MMP is in the mix is a good idea to stop a new Privacy Sandbox form of click spam from getting credit for everything. And it’s also important to use deduplication keys:

“When an advertiser uses multiple ad tech platforms to register the same trigger event, a deduplication key should be used,” Google says. “The deduplication key serves to disambiguate these repeated reports of the same event. If no deduplication key is provided, duplicate triggers may be reported back to each ad tech platform as unique.”

While there’s no talk of multi-touch attribution (MTA) here, there’s clearly an emerging capability for an MMP, which sees all the sources and triggers, to provide a very nuanced view of an advertiser’s performance across all paid media partners, giving publishers much better visibility into how their ad campaigns deliver incremental impact. Expect some surprises here, and expect this to change how performance marketing campaigns are built and optimized.

Web-to-app as well as app-to-web and other combinations

Google also updated multi-channel app install customer journeys, saying that “all combinations of app- and web-based trigger paths are supported.” This of course makes perfect sense: while we’re primarily dealing with Privacy Sandbox on Android here, the entire effort is built on insight and technology that originates from where Privacy Sandbox was first seeded: the web.

That means …

• App to app
• App to web
• Web to app
• Web to web

… is all part of the plan from the beginning.

And that is precisely what we’ve been hoping for on the iOS side of the table, as we said in May of last year:

With Private Click Measurement, Apple gave us a tool to measure app to web journeys in SKAdNetwork. That’s great, but we also need tools to measure web-to-app journeys.

Getting it on Android makes it more likely that this will arrive on iOS, and since on iOS Apple — like Google on Android — controls the entire mobile operating system, it should be possible. Certainly in mobile Safari, but also in other browsers if a) Apple creates open APIs, and b) Google and other browser makers use them.

In any case, this is good to see from Google: there’s a clear path to high-quality privacy-safe advertising measurement and attribution on Android.

More on Privacy Sandbox for Android from Singular:

• Android’s SDK Runtime in Privacy Sandbox is a game changer
• Topics API: How Privacy Sandbox on Android connects people with ads
• FLEDGE in Privacy Sandbox on Android: reinventing the mobile ad network?

Also, here’s a new high-level video from Google for a quick introduction:

You might say it’s kind of the best of times and the worst of times in marketing measurement, with apologies to Charles Dickens.

Because knowing who a person is and which device from where is accessing a service is undergoing wholesale change from a marketing point of view in the age of privacy. IDFA is gonzo, GAID is on the way out, and as any web marketer has known for years now, third-party cookies are an endangered species.

Not shockingly, all of that is changing ad targeting, marketing attribution, and campaign optimization. So times are tough, in a way. But in another very real sense, it is apparently the golden age of measurement technology. Why? Because we are seeing a huge amount of innovation in the space.

• Apple: SKAN
• Google: Privacy Sandbox (web and Android)
• Brave (and others): Blockchain solutions
• IAB: Project Rearc
• And more … ID5, Unified ID 2.0 from The Trade Desk … and so on

Plus, of course, there’s a solution from Facebook: IPA.

Interoperable Private Attribution from Facebook … err … Meta

But IPA is not the kind of India Pale Ale you can drink in an English bar. Instead, it’s Interoperable Private Attribution, and it is yet another solution to the eternal marketing dilemma: answering that simplest of questions with far-too-complex answers … what’s working?

And, of course, answering that in a privacy-safe way.

If you’ve checked out Privacy Sandbox for Android, you’ll see a few similarities, Singular CTO Eran Friedman says. The Facebook/Meta IPA proposal is based on three general concepts:

1. Match keys
2. Event generation with sources and triggers
3. Aggregate attribution measurement

Match keys coordinate between publisher data and advertiser data, and crucially have to be built into the computing environment: the browser on the web, or the mobile operating system on smartphones and tables.

“If a user does an action and clicks an ad in one app, and then triggers a conversion in the advertiser app, then there will be a matching key to kind of connect the dots. Then the other piece is the events … two types in the IPA … the source events which happen in the publisher [app]: things like an impression, a click things that the user does in the publisher, and then they have trigger events which happen in the advertiser app.”

Eran Friedman

Those are similar, Friedman says, to conversion events in SKAN, and connect to sources via the match keys.

Double the privacy protection?

However, that’s where Facebook adds yet another layer of privacy protection. Welcome to the concept of trusted servers … or at least semi-trusted.

“They defined a concept of ‘trusted servers’ essentially, which are kind of unbiased, third party services that receive these encrypted postbacks with match keys and trigger events. And these are the ones that are able to decrypt the information and then provide very granular data to both ad networks and advertisers would count how many conversions came from the sources and basically power attribution based on these encrypted postbacks.”

Eran Friedman

Where Privacy Sandbox has a single aggregation service for a trusted third-party, Facebook’s system is designed for two “semi-trusted” services, and both are essential. No single third-party can decrypt the postbacks on its own, making it less likely that any single entity could break privacy in the system.

Advertisers themselves and ad networks could get full data with the help of the semi-trusted services, but the semi-trusted services themselves would have only partial visibility of the granular data.

There is a problem, however.

Because Meta/Facebook is who it is and not Apple or Google, it doesn’t own a mobile operating system or a web browser. While IPA has been built in partnership with Mozilla, which owns the Firefox browser, that only accounts for maybe 3.5% of global browser market share. The obvious question is: why would Apple (iOS, Safari) and Google (Android, Chrome) build support for Facebook’s attribution methodology, Interoperable Private Attribution?

Short answer: they probably won’t.

United Nations of marketing measurement

Which is essentially the reason we need a United Nations of marketing measurement: an entity to bring all the methodologies and technologies from all the platforms and stakeholders together.

Given that the odds of this happening are roughly similar to the UN brokering global peace or solving world hunger tomorrow, the implication is clear.

MMPs like Singular are the trusted third parties of marketing measurement and essentially have to provide an abstraction layer over all the multitudinous methodologies from all the battling players in the market. That abstraction layer then gives marketers a single source of truth without them having to know all the individual intricacies in each platform’s chosen attribution solution.

“We always saw ourselves as the ones who have the role to navigate advertisers and the industry through. Our goal is always to see what we have to work with and build the tools and the reporting capabilities, the management capabilities, so that whenever someone wants to use a tool among the vast toolset that they have, it’s seamless for them to try something out, see if it’s insightful for them, see if it’s relevant for them to use.”

Eran Friedman

But the massive diversity — though it offers challenges —- isn’t all bad.

It’s also driving innovation, Friedman says, since Google could see what Apple did, and Facebook can look at both, and all can come up with better versions in the future.

Subscribe to Growth Masterminds

Subscribe to Growth Masterminds on your favorite podcasting platform:

• Apple Podcasts
• Spotify
• Google Podcasts
• Or others …

What does an MMP look like in the year 2030?

Well to start, no-one knows what “MMP” stands for anymore.

We’re in a time of massive change in mobile, in marketing, and specifically in the niche of marketing that is specific to mobile user acquisition. It’s an odd niche: there was no website user acquisition in the early days of Geocities and Friendster and Homestar Runner. But it’s an important and lucrative niche because mobile was the first truly personal computing platform. Built on the three-foot device (never more than three feet from your body) mobile offers unique access and intimacy to customers (users), making a mobile app install more valuable than a newsletter subscription, a website visit, or any of the other precursor actions to monetizing attention, service, or product in other customer acquisition modes. 

MMPs were built to optimize growth in that niche, but the world is changing. What does the mobile measurement platform of the future look like?

MMPs beyond mobile

This is no shock. 

Any MMP worth its salt is, if not omnichannel, at least multi-channel, with extensive technology to measure and optimize web-based journeys, old-school TV, connected/smart/OTT/streaming TV, email, and multiple other digital and non-digital channels. Sometimes that verges on simplistic — put a Singular link on a billboard, and boom! you’ve got a checkmark in the out-of-door category — and sometimes that’s sophisticated: correlating connected TV ad campaign impacts with app installs and/or marketing conversions via intricate data science.

But it goes beyond the obvious.

With the hundreds of billions of dollars being invested today by Apple, Meta, Google, Microsoft, and thousands of other companies globally to invent the next major computing platform, getting a “user” or (better yet) a customer won’t mean someone installing an app on a slab of glass and metal that slides into their back pocket. 

At least, that’s not all it’ll mean.

Because “MMPs” aren’t first and foremost mobile measurement platforms, they’re first and foremost measurement platforms. The “mobile” part of the name is a modifier, and it could easily be replaced by smartglasses or wearables or even some futuristic cloud-based edge-capable personal AI that inhabits all of our devices and, like Tony Stark’s super-awesome but completely unrealistic Jarvis, has all the required data, access, answers, and insights we need at any given moment.

A world beyond mobile

Not shockingly, the world is not standing still.

For my Forbes columns and TechFirst podcast, I’m tracking at least 11 megatrends from smart matter to infowars to greentech and automation. Several are most relevant for apps, publishers, brands, and marketers, including:

• Artificial intelligence
• Virtuality
• Decentralization
• E-dentity

I’ll go into more detail on these in a subsequent post, but suffice it to say that technological, social, and political changes are impacting the landscape in huge ways. Privacy will continue to grow. We’ll move farther towards a Ready Player One-style metaverse, continuing on the path we’ve walked since ARPANET more than 50 years ago and the first dim stirrings of HTTP in 1990. The yin and yang of centralization-decentralization will continue its pendulum swing, now trending (at minimum in terms of hype) in the direction of decentralization. And AI will infuse literally everything, becoming a core foundational piece of apps, software, and services.

With that as context, here’s a few things that you can expect in an MMP of the future.

One note:

Don’t take the 2030 date too seriously. Many of these exist today in some form. Many more are coming much sooner, but will be much more fully developed. And if I miss something you think should be on the list … let me know!

M is for multiple, many, and multitudinous

It’s obvious today but it will be increasingly necessary: marketing measurement platforms increasingly need to ingest data from a huge number of data sources. That doesn’t just mean many platforms and thousands of ad networks like Google, Facebook, ironSource, and Applovin. It means dozens if not hundreds of fundamentally different kinds of sources, often using vastly different methodologies, and combining it all to create data-driven insight.

Today, that looks like this:

1. IDFA (in very limited supply)
   Some apps still generate a significant percentage of opt-ins, which can still be useful depending on the other party in the adtech ecosystem getting opt-in as well. But I’d be shocked if IDFAs were still available in eight years.
2. GAID
   At least until 2024. After that … see #4 …
3. SKAdNetwork, or SKAN
   Apple will probably be on SKAN v5 or v6, but the broad strokes will likely be the same: no granular data, privacy-first, minimal marketing data.
4. Privacy Sandbox on Android (and web!)
   Privacy Sandbox will be mature and full-featured and very usable, while still keeping granular device-level data behind a privacy shield.
5. Cost & campaign data from ad partners
   Wherever brands are investing in properties and buying ads, cost and campaign data will be available, providing both network-reported inputs (views/impressions) and network-reported results (clicks/actions).
6. In-app data
   Think of “app” broadly here … while it might be in some kind of mobile device like a smartphone, it could be an app on a pair of smartglasses or it could be embedded in another kind of device. The data will still be first-party data via logging, a CRM/DevOps/LiveOps type of tool, and marketing measurement or attribution SDKs or APIs, but each platform will have slightly different rules on what data brands can access or export, even if it’s on a first-party basis.
7. Store data
   App Store, Google Play, alternative app stores, and app stores on emerging platforms like VR and AR: all will generate somewhat analogous datasets around installs, ads, CTR, and A/B tests.
8. Alternative economy data
   Of course most MMPs ingest data about in-app purchases and ad monetization inside apps, but increasingly we’re seeing the rise of complex economies using platform or brand stores of value or cryptocurrencies. Increasingly we’ll also see these start to bridge apps and platforms, and they’ll be a more and more important measure of the health and growth of an ecosystem that publishers want to measure. Think blockchain-query-requiring insights like activity, value, recency, frequency, engagement, and so on.

Tomorrow, who knows what that will all include. Some parts are obvious, but some we won’t even really be thinking about today. Some options:

1. World data
   Weather? Sure. Climate trends? Macro-economic trends? Micro or geo-specific trends? Sure. Hype and buzz from all different areas of communities and verticals that drive people’s time and purchase behavior, both specific to a brand, specific to a vertical, and general but applicable to your business? Absolutely. Some of this happens today, particularly in massive and slow old-school incrementality measurement methodology, but it’s all going to get much more sophisticated.
2. Cloud gaming
   Well, it’s happening today, and is likely to continue to grow …
3. Emerging platforms
   Car operating systems, home operating systems, edge device data …
4. Additional sources
   We don’t know about them yet … but guaranteed there will be more than matter.
   1. Apps on Starlink? 
   2. Mesh protocol services? 
   3. Darknet products to cross splinternets?

Over the next six years, these sources will have to be broadened to multiple app stores (even on iOS) and multiple platforms both handheld, wearable, desk-bound, audible, home-focused, car-focused, and more … plus all the ad networks and marketing tools that grow up around each of them. 

And, as I’ll chat more below, they’ll all add up to increasingly useful incrementality measurement in an age of aggregation not granularity, data but not trackability. Call it science, call it art, call it magic, incrementality is getting better and better and will be a source of not-quite-real-time insight for both strategic and tactical use.

Connected … built in to all your tools

MMPs like Singular already tie into major publishers’ data architecture via API or ETL, but BI is ridiculously bespoke today. Every publisher has something just a little different (or a lot different) which takes time, energy, and focus to build and maintain, and privileges the large and wealthy.

Expect this to get easier and quicker in the future, where the data you need for custom purposes is instantly connected/integrated/used in any tools you wish. Including, if you wish, your ad partners’ platforms.

Always-on continual incrementality testing

As mentioned above, this will become standard. Sure, this is in some sense already doable today, but it’s still fairly clunky and of debatable value: more strategic than tactical. It will become much more integrated into default, automatic marketing platforms.

Aligning the marketing campaigns you want to run against the measurement outputs you want to have — including always-on continual incrementality testing — should be automated right from campaign inception to close, with near-real-time reporting on how incremental each channel and effort actually is. And, of course, marketers should be getting suggestions (and perhaps also automated changes, see below) on adjusting spend and channel mix to minimize duplication or overlap, and maximize results at a given level of ad spend.

Modeled single source of truth

We’re operating in an increasingly uncertain world of performance marketing. 

With the death of granularity, privacy thresholds, censored data, and missing campaign information already on iOS and coming soon to a theater near you on Android, few things are certain. (Although, let’s be honest, in the IDFA/GAID last-click era, certainty came with the cost of some amount of correctness: the world and a customer journey is much more complex than one click and one conversion.)

Multi-touch attribution as a function of near-total tracking is dead. Modeled attribution that intelligently mixes deterministic but aggregated platform data with first-party usage data with marketing inputs data will give marketers a model of reality that they can trust.

Within certain parameters. Plus or minus a certain percentage.

Some things are certain: when you act, there are impacts. Modeling those in an increasingly complex world of marketing data will provide a single source of — if not truth — or at least reasonably trustworthy truthiness.

And that will serve as a firm-enough foundation for future action.

Intelligent

There’s a significant amount of artificial intelligence built into a modern MMP, and it’s continuing to grow fast. Particularly in areas where you need modeled measurement due to censored data, or predictive analytics.

But the MMP of the future requires much less effort and knowledge to set up, and much less challenge to get what it knows out. Natural language queries? Sure, we’re seeing them already and they will be commonplace. Conversations with your measurement partner? Absolutely. Anticipating your needs and providing you the data you want when you want it, in the format you like it? Sure. Ad fraud warning signals? You bet.

Automated alerts are here already, mostly for known knowns and known unknowns. What about the most dangerous class of events, unknown unknowns? Brand danger due to a completely unrelated news incident in which someone wearing a t-shirt with your logo did something unspeakable in what is now a viral VR video? 

You get the picture.

But that’s just the beginning. Today we’re already using machine learning to drive modeled insights. Tomorrow the inputs will be much more complex and the models much more refined.

Automated

Marketers can automate a lot today, including spend. There’s a lot more to come, including goals, recommended campaigns, suggested spend, plus automated shifts as the platform detects opportunities or weakness. Also, think AI-built creative changing automatically with AI-driven intelligence as your measurement platform senses failure. Tailored, of course, not just by platform or channel but across the board as winners and losers become clear.

Much of this is doable today in part. 

It’ll all get significantly easier.

Almost extinct

OK, sure, this is a bit tongue-in-cheek. But let’s be honest: ad networks like to own their own measurement (why could that be?) and suites continue to grow. The independent measurement platform that is solely focused on best-in-class marketing insights is in some sense an endangered species.

However, there are still some independent players with a hard focus on marketing intelligence for their customers. To continue to grow — and exist — they increasingly need to play well with others, so privacy-safe data can continue to flow.

Marketing measurement’s role: make it all make sense

Nothing changes.

For an MMP of the future, nothing changes in the ultimate mission. Which is, of course, to make it all make sense. What is the correlation between everything I’m doing to everything I’m getting … and which parts of what I’m doing are driving the best parts of what I’m getting.

When marketers know that, they have power. Power, quite literally, to change the future in ways that boost their brands’ growth.

That’s the MMP — if the acronym still exists — for 2030.

Games don’t just have ads anymore. We also have entire games as ads. Even worlds as ads, as Lego builds out a kid-safe space with Epic in the metaverse. And increasingly lately we’re getting non-intrusive ads inside games. As in: inside the gameplay. On the walls. In the halls. Embedded right inside the actual architecture and artwork of a game, not grafted in via a rewarded ad loop or pasted in via an interstitial.

In-game ads, clearly, are the new OOH (out of home) advertising. 

In a very metaversy way.

In a sense, this is no shock. Everything is an ad network today, right? I mean, if Doordash, Zoom, CVS, Walgreens, and Instacart are ad networks, if Amazon owns a truly massive ad network, almost everything where people gather can be an ad network. 

Even ads can have ads, right?

https://twitter.com/AdamBlacker25/status/1516859273032175622

And it makes perfect sense: movies and TV still occupy more of people’s time, around 2,000 hours a year, but gaming is closing in. While the average American spends only a few hundred hours a year playing games, binge gamers spend close to 500. And young people are increasingly tuning into games and out of TV.

Which means ad dollars have to go somewhere.

And games is one big and still rapidly-growing space with billions of gamers globally, many of whom don’t really watch TV much anymore, if they watch it at all.

“The way people are consuming media these days, you don’t reach 100% of your audience on TV anymore,” Steve Hartmann, a VP at Experian, recently told WSJ.

So where do in-game ads go?

If you’re not interrupting attention and highlighting your ad full-screen and requiring player interaction to continue, where do your ads go?

On the walls.

In the stadiums.

On the buildings.

On a jersey.

Wherever ads go IRL.

In short, wherever it makes sense given the construction of your game. So it could be a skin in Fortnite for players to choose. It could be the name of a stadium in your sports game. It could be real-looking (but virtual, of course) ads on the side of a race track.

And they can be designed right into the look and feel of your environment, like this in-game ad for a fake product, Nuka Cola, in Fallout:

That could, just as easily, be a real product from a real company that is hoping to inspire real sales. And guess what: that real company could be you — the game publisher — cross-promoting another of your titles.

Ads in games: who’s leading the industry?

Microsoft is reportedly working on ads for Xbox games. That’s smart, because not only is Microsoft massive in gaming, having just acquired Activision Blizzard, it also has a significant ad network. That’s not just Microsoft Ads, which reaches almost a billion people, it’s also Activision Blizzard Media which — coincidently, perhaps — already offers an in-game ad product.

According to another report, Sony is working on something similar for Playstation. That could be used to support free-to-play games on the console, or reduce the cost of pay-to-play games.

(Grain of salt: both of these are unconfirmed by either Microsoft or Sony so far.)

The giants might be testing the waters, but there are multiple startups creating non-intrusive monetization solutions for games:

• Gadsme
• Admix (full disclosure, I’m an investor)
• Simulmedia
• Anzu
• Odeeo (focuses on audio in-game ads)

These are a few that come to mind. (Ping me with additional players if you see that I’m missing one.)

The value of in-game ads

Hey, money is good. It pays for developers (and marketers) and it keeps servers humming. But so do other forms of advertising. What’s good and interesting about in-game ads?

First off, they don’t interfere with gameplay.

Most ads are intrusive. Even if like rewarded ads they happen at the discretion of the gamer, they take gamers out of the world they’re in, out of the game they’re playing, and into a different reality. That may be a necessary evil, but even unnecessary evils are still … sort of evil. So letting players play seems like a good idea.

It’s also kind of free money.

I mean, you might have rewarded videos or interstitials. If so, there’s a natural limit on how many you can show and how often you can expect a player to engage with one of them. But a brand on a storefront is just there … it’s part of your playscape anyways. While it may not work for a game set in a primeval forest on a far-distant planet, it probably works for your game set in a fictional New York City.

And in some sense it’s kinda cool: real ads in synthetic spaces inside games. There’s a realism to that that can work really well.

Oh and guess what … you’re probably not feeding your competitors. Existing ads in games on mobile or console might tend to be for the kinds of things players in your game might like, such as games. In-game ads can be for brands in clothing or cars or laptops, making them non-competitive with your own games.

Challenges of in-game ads

Just because in-game ads offer some unique opportunities doesn’t mean they’re problem-free. While most can be very easily integrated via no-code SDKs, you’re going to have to make them fit somehow in your game.

So you’ll need some pre-thought and product work, at minimum.

In addition, because it’s in-game and intended to be non-intrusive, there’s no click or tap or linking out to an App Store or Google Play or console maker or publisher site. That means deterministic user-level performance data isn’t going to happen, and you’re going to have to rely on data science and mixed models and big data for next-generation marketing measurement.

It’s possible.

It’s doable.

But yes, it’s different.

That said, everything is different now as iOS uses SKAN and Android moves to privacy sandbox. And marketers are just having to adapt to new realities.

Talk to Singular

Looking for help to measure the unmeasurable? Book some time with a Singular expert and we’ll walk you through the Singular solution for marketing measurement in post-IDFA, post-GAID times.