A post-IDFA guide to user acquisition for indies

Until mid-2020, developers knew how to grow their mobile games. A good game was by no means a guarantee of success, but the rules of mobile marketing and user acquisition (UA) were familiar and clear. And there was always time to make more games!

Apple's changes to its identifier for advertisers (IDFA) and particularly the release of iOS 14.5 changed the rules of the mobile marketing game beyond recognition.

With attribution of users taken in-house by Apple, a 24 hour data privacy threshold introduced and no real-time marketing events available, it became incredibly difficult to determine where games' users were coming from, what their value was, and what marketing campaigns were working. In fact, although new marketing data called SKAdNetwork (SKAN) had been introduced, Apple was passing this straight to ad networks, leaving mobile games studios in the dark.

The release of iOS 14.5 changed the rules of the mobile marketing game beyond recognition

The version of iOS released in September 2021, iOS 15, was better news for mobile games developers and publishers. Apple began sharing SKAN data directly with developers as well as ad networks, putting the full range of data in their hands for the very first time.

Making good use of this data is another thing altogether, which is why I still see the majority of mobile game devs operating in what I call 'Zombie Mode' while they wait for someone, somewhere, to help deal with all this data.

The changing formula for a successful game

Hi, I'm someone. I work with dozens of people who spend a large part of their daily lives figuring out how mobile games can use SKAN data. And I won't lie to you, it's an ongoing process. There's no immediate antidote that will snap you out of Zombie Mode or take you back to the good old days before iOS 14.5 bit.

What we do know for sure is that the formula for a hit game changed from Q2 2021 onwards. Before that it was essentially:

Great game (very high retention + very high ARPU) x good user acquisition (low CPI) = hit

Or

Good game (high retention + high ARPU) x great user acquisition (very low CPI) = hit

Small studios are having to prioritise data science that helps them make sense of SKAN

In a nutshell, great games have users who are committed to playing them and are monetised effectively. They also have good marketing campaigns that deliver a low cost per install (CPI). Think Clash of Clans. Good games have this to a lesser extent, but make up for it with better marketing campaigns that give a lower (CPI). Think games inspired by Clash of Clans.

But now, even if devs pay to acquire a user, they'll receive anonymised, hard-to-use data on them if that user has not opted in to tracking. So CPI rises significantly. This means that, in 2022, the formula for a hit game is all of the above PLUS great storage and analysis of large amounts of complex data. The upshot of this is that small studios are having to prioritise data science that helps them make sense of SKAN. Even founders of indie teams with no in-house marketing are having to put a data hat on.

Stopgaps and wishful thinking

For most game devs in Zombie Mode, it's a case of trying to get by. That's why, instead of tackling the daunting prospect of SKAN head-on, many (most) are resorting to a combination of fingerprinting, manual processes, deterministic attribution of users who do opt in to tracking, and a lot of wishful thinking.

So-called probabilistic attribution -- more commonly known as fingerprinting -- is a stopgap attribution method used by many mobile advertisers to avoid truly getting to grips with the privacy-first landscape. Fingerprinting collects software and hardware information about a user to attribute them to an ad campaign. However, it only works for ad networks that use links in their ads, meaning it doesn't work for Snapchat, Google or Facebook. And Apple is taking steps to end the practice, or at least reduce its effectiveness.

Although it's common knowledge that fingerprinting isn't viable in the long-term, I think for many games studios the hope is that if they can get by for now, they may be able to increase their app tracking transparency (ATT) opt-in rates far enough to deterministically attribute the majority of their users and effectively go back to their old ways of working. This isn't a total pipe dream, but it's extremely unlikely.

So for the vast majority of studios in Zombie Mode, it's tempting to delay taking the steps needed to get to grips with SKAN. I'd argue that they're delaying the inevitable and that short term pain is the price for the long-term gain of setting themselves up for success in the long-term.

Does any of this actually apply to me?

The narrative around IDFA and SKAN tends to be that everything is okay now: crisis averted. That's for two reasons. Firstly, massive publishers who are successfully navigating mobile games marketing receive disproportionate attention and are making large profits and enormous acquisitions. This has a secondary impact in that, because game developers aren't hearing much about their peers of a similar size, they're not sure where they fit in the shifting marketing landscape and don't have relatable examples to follow.

For this reason I created a persona called 'Jack'. Jack is the everyman or woman. They lead a UA team of four people in a small mobile games studio and are responsible for making the big calls on advertising strategy. Before iOS 14, mobile measurement partners (MMPs) were Jack's best friends. They gave them install data, standard analytics (DAUs, sessions, revenue, retention), LTV metrics, ROI and ROAS.

Massive publishers who are successfully navigating mobile games marketing receive disproportionate attention

In 2022, with these familiar ways of working gone and only around a quarter of their games' users opting in to ATT, Jack feels a bit lost. Their colleagues feel overloaded with developing games and automating creative and testing -- they don't have the time to automate UA too. So the team gets by with manual processes such as excel sheets downloaded and collated manually, free tools from GameAnalytics and a chaotic Trello board.

Some of these challenges may be offset if you work with a publisher, but I expect at least a couple of Jack's challenges resonate with most of you.

Starting to automate

These challenges become critical issues when you want to scale multiple games at once. Doing this requires automation -- with the bootstrapped manual processes above you'll only be able to take care of scaling one game at a time. This is particularly true for hypercasual games, where targeting is so broad (effectively anyone), that you're generating huge amounts of data.

This is where that short-term pain for long-term gain comes in. I'll dive deeper into the benefits of automation later, but the headline is that once it is in place you won't have to worry about data science, data consultants, data visualisation engineers... And you'll save endless amounts of time that is better served making games and making revenue.

There are judgement calls required to put a strong set of automated processes in place

That's the gain; what about the pain? Well if you want to build your own back end, it's likely to take 6-12 months, and that's after bringing in temporary support from a data expert. The good news is that there are tools out there designed to support small teams to automate such as Growth FullStack, and you can continue to work with many of the tools you already use. Even free tools such as Google Sheets, for example, can connect to an API. You can also combine automation with manual processes, which may be necessary when you need to move fast to release a game and don't have everything fully automated yet!

There are judgement calls required to put a strong set of automated processes in place. I can't cover them all here, but some things to think about are:

• How will you map your processes so others can follow them?
• How are you going to store all this data without it being prohibitively expensive?
• Do you want to keep working with a publisher? Theoretically, automating means you might not need to, but you'll suddenly be working with lots of data you're not used to.
• Who's in charge of reconciling your existing data sets and ensuring data quality?
• And, crucially, how are you going to analyse massive data sets?

Once all of this is sorted (easier said than done, I know), you will only need better storage as your data grows -- hopefully you'll never need to go through the whole exercise ever again!

What can automation do for mobile publishers?

It's not enough to bring all your data sources together and store them neatly. You need to understand how to gain insights from this data so you're not making blind marketing decisions about your games.

At its best, an automated data stack acts as a magic 8 ball that you can ask any question of. For example, for a hypercasual campaign, you want to know what playtime, retention and CPIs are. You need to centralise all the various sources of data, automate the data pipelines and connect business intelligence tools to it. You might want to see data broken down by country or to compare a game's current data with its historic data, or even compare it to your studio's other games of a similar genre. This could show you that the puzzle game you're currently growing has 50% higher playtime than any other puzzle game you've made before. This level of insight at scale just isn't doable manually.

Instead of having to set up the same marketing events over and over again, they can be set automatically. Again, this is easiest in hypercasual where events are standard from one game to another. The metrics are the same -- you just need a higher player LTV than acquisition cost. That's not to say automation isn't the way to go for other game devs too. If you monetise predominantly with IAPs, you can look at which ad network generates the highest volume of users who make $5+ purchases in your game. By spotting that this is largest from, say, TikTok, you can scale up traffic from that source accordingly.

This is really just a snapshot of some of the things game devs and mobile marketers can do with a properly set up data stack. For studios like Jack's, this is the ultimate route to being able to profitably scale multiple games.

Android is changing too

What has been the impact of all of this iOS chaos for Android? Well, last year was the first time that iOS has not outperformed Android in terms of hypercasual ad spend. Just three years ago, iOS received 60% of spend. In 2021, that trend almost reversed.

As well as shifting spend to Google Play, mobile games marketers have also experimented far more with alternative Android ecosystems such as the Samsung Galaxy Store and Huawei AppGallery. With less intense privacy restrictions, lower CPIs and less competition, they're definitely a solid prospect for many games, although the monetisation infrastructure isn't anywhere near as established as iOS and Google Play. Creative testing, too, has shifted even further to Android.

While flourishing Android ecosystems are a good thing for games, they're not a magic bullet. iOS is going absolutely nowhere in terms of its importance for mobile games, while Android is already taking steps that will bring it more in line with iOS this year. October 2021's Android 12 update set the tone by introducing zeroed advertising IDs for opted-out users.

While flourishing Android ecosystems are a good thing for games, they're not a magic bullet

In February 2022 Google upped the ante by announcing the expansion of its Privacy Sandbox to mobile. Previously it was focused on web technologies only. This will eventually (over the course of the next two years or so) have wide-reaching implications for games marketing on Android. But a big change that has gone under the radar is that the Android 12 policy will be extended to all Android operating system versions just a few weeks from now (1st April 2022), to encompass every opted-out Android user.

So for mobile games, I'm sorry to say that there will be adjustments to be made on Android too. Even ad industry giants such as Facebook are already working to accommodate this reality by starting to defer attribution to the Google Play Install Referrer. This makes adapting to SKAN even more pressing, so studios aren't forced to develop new strategies for the two main ecosystems at the same time.

The Cinderella formula

While traditional game development has naturally centred on the game itself, hypercasual developers have hacked the system over the last few years to create a formula that could hold lessons for developers across the board. They typically work on creatives first and carry on to game development only if the creative resonates with potential users.

Building creatives that mimic what a game and its gameplay would look like rather than what they do look like is a huge competitive advantage

With the pressures of privacy-first marketing, this formula bears consideration for developers of all genres. Everyone needs to work on several iterations of a game so they can UA test but, due to limited resources, even medium size studios can only really work on a couple of games at once.

That's why building creatives that mimic what a game and its gameplay would look like rather than what they do look like is a huge competitive advantage. Even small to medium sized teams can work on and test 50 creatives a week; only converting the most successful to the game development stage. They can also iterate on the creatives to figure out whether a different character might appeal more, for example, by comparing the click through rates (CTRs) achieved. And finally, the Cinderella formula means acquiring all-important data at the onset of the game development process.

The Cinderella formula is driven by the reality that for most, game development is a business. I do admit, however, that it could grate at developers who simply want to create the game they'd wish to play themselves, in its purest form!

I hope I haven't come across as all doom and gloom here. It absolutely is difficult to grow mobile games in 2022. The odds are firmly stacked against the little guys. But you're far from alone in still figuring out how to advertise in 2022. Again, I can attest to this as someone who works with people whose entire job is doing just that!

It's true that setting up your company for long-term success at scale will require hard work to automate your marketing and data stacks, regardless of how you're navigating Zombie Mode right now. This doesn't however need to be prohibitively expensive, and you can continue working with existing partners and many of the tools you are already familiar with, including the free ones.

The reward is more time to make great games that are enjoyable for players and profitable for developers and publishers. Ultimately, this is what 99% of us want, we just need to adjust to the new rules of the game.