How mobile app tracking is changing (or, making the insane sane)
Bad news: complexity is the future of marketing measurement.
Good news: you can manufacture simplicity to achieve insight. (And growth!)
Challenge: that only works with the right partners.
Mobile marketers have always had a huge amount of data on their marketing campaigns. For marketers with scale, it’s generally been from different partners, in multiple places, and in incompatible formats. Plus, the fraudsters were always lurking, so optimizing for success was always challenging.
But the good news from a mobile app tracking perspective: you got the data. And, you got it pretty much one way.
Mobile app tracking for the past decade
For much of the past decade, mobile attribution has been, at least at a certain level, fairly simple: Ad Network A delivered an ad via Publisher B to a device with IDFA C or GAID D. Person E using an iPhone with IDFA C or an Android smartphone with GAID D saw that ad and tapped it. Your mobile app tracking provider noticed the ad impression and the tap (sorry, I can’t call it a CLICK). And that mobile attribution solution also then noticed when a user of the device with IDFA C or GAID D — presumably, but not provably Person E — opened the advertised app for the first time. Meaning, of course, that the ad had proceeded a trip to the App Store or Google Play, an install decision and action, and a first app open.
It’s a long chain, but it worked:
There were plenty of challenges: was the last tap actually the definitive driver of an install? What about earlier ad impressions? What if someone downloads the app but never opens it? Or, if they open it three weeks later, when they see yet another ad for the app they already installed? What about if the tap (fine, click!) was faked, and a previous click was actually the influential one? Plus, how do you distinguish between a new-to-you user or an existing user on a new device (cross-device analytics). And, how do you drill down into granular data for the deepest optimization? How do you connect attribution data with campaign data to generate ROAS?
Plenty of challenges, indeed.
But it also had a certain simplicity, and by and large, it enabled the mobile app winners of today to achieve their current leadership roles. You could always argue last-click attribution, you always needed vigilance against fraud, and a certain percentage of your traffic was invisible to mobile app tracking thanks to Limit Ad Tracking settings.
But you had enough data to optimize creatives. Enough data to optimize partners. Enough data to optimize campaigns. Enough data to synchronize with media sources to identify the most likely monetizable customers. Enough data to grow.
Old-fashioned attribution is fading away
Now a lot of that is changing. And changing fast.
On the web, the third-party cookie is dying. By 2022, sayonara. But that’s actually pretty long-lived compared to the mobile equivalent on iOS, the IDFA. Apple’s Identifier for Advertisers is not precisely dying, but it’s also not likely to be universally useful as soon as Apple turns on its new iOS 14 privacy protections, likely in early 2021.
Now what will you have?
On Android you’ll still have GAID. How long that will be available, who knows. Google is likely to change its mobile identifier for advertisers at some point, especially because it also offers the Google install referrer, in some ways a simpler and more granular form of SKAdNetwork. But you still have GAID for now.
On the iPhone and iPad side you also still have IDFA, if you decide to ask for it. It might be only for 5% of your traffic, but it could be 10-20%. This is not necessarily a disaster, by the way. Statistically speaking, even at the low end of that range, a 5% sampling is much more than you need, at scale, to get a good sense of what’s happening in a population. (This assumes, of course, that you’re targeting a mass audience. Whale hunting will be more challenging.) National polls, after all, achieve statistical significance with much smaller sample sizes. (Let’s not talk about election polling in the U.S. just right now.)
But you also have SKAdNetwork.
When you don’t have IDFA, you have the option to attribute app installs via Apple’s updated mobile app tracking framework. You won’t get everything you might want as a marketer, but you’ll get privacy-safe data on where successful installs happened, and at least some post-install data. There’s actually some real power to this solution, especially if you use Singular SKAN to maximize the predictability of your marketing attribution data. And it’s available for all users, eliminating the Limit Ad Tracking (LAT) blindspot.
So you now have a mix of IDFA and SKAdNetwork on iOS. At some point, if Google does something similar, you might have to mix GAID and Google install referrer on Android as well.
Going deeper with data
And it doesn’t stop there.
If you’re not happy with deterministic but aggregated data, you might decide you need to do some level of incrementality testing and regression analysis to determine what’s really working and where your future ad investments should be placed.
To do that, you need to aggregate and standardize and normalize all the top-funnel platform data which we haven’t even talked about yet: the data you see in your dashboards for Facebook ads, Google ads, MoPub, Snap, TikTok, Pinterest, Vungle, ironSource, and all the other ad partners you’re working with.
(Singular does this out of the box, by the way, and customers are doing it TODAY.)
Taking a step back, and you see how while in the past you had bright shiny well-lit sources of granular data that were transparent and rich, in the future you’re still getting huge amounts of data, but it’s growing a bit dimmer, a little lower resolution.
The past was privacy-poor and data-rich.
The future is privacy-rich and seemingly, data-poor.
All that new data does, however, matter. While there’s change, and granularity is definitely an endangered species in mobile app tracking right now, smart mobile marketers can still get the attribution they need to optimize for growth.
Mobile attribution for marketing optimization in 2021
Getting all the data is the starting point. Applying proactive fraud models in real-time is still important (yes, even Apple’s data from SKAdNetwork will have multiple attack vectors for fraud). Combining marketing data and attribution data, where possible, is still relevant. But it becomes critical to be able to represent all the different data from varying sources with divergent levels of granularity in ways that mobile marketers can see, understand, and make advertising optimization decisions on.
It starts with upper-funnel cost and campaign data that is still detailed and rich: the campaign analytics that you get from every single ad partner. Marketers have always needed to collect and aggregate this data and present it in a simple, scannable way, and mine it for insights. Now, this is even more important, providing the creative optimization and CTR data that you won’t be getting from SKAdNetwork or an eventual Google equivalent.
Seeing those insights in a central location is critical. Setting up simple dashboards and flexible reporting for tracking and monitoring — and easy slicing and dicing of data when you wish — is huge. And for those that wish you need to be able to easily bring data into your systems without writing code via ETL. That feeds BI, populating your internal modeling, and allowing you to apply machine learning to what you see in first-party data.
It also means being smart about opting for hybrid customer experiences and hybrid customer journeys when it makes sense. SKAdNetwork is only app to app. Web to app is likely to grow in importance since you can measure mobile web search, influencers, and journeys you intentionally route through mobile web for additional insight. First-party data is all about you and your customer, so in-app data and web-to-app journeys are still under your full control. It’s essential to stay privacy safe, of course, and you can do so in web to app journeys when you acquire user information in a first-party way and keep it for first-party use only.
Essentially, your solution for mobile app tracking needs to acquire data from multiple sources, mix and match it together, extract actionable insights while controlling for fraud, enable hybrid journeys when it makes sense, and provide levers for marketing optimization.
It’s always been about data.
Now it’s more complex, but with the right partners, the growth you need is still achievable at a manageable level of complexity. Connecting all the different datasets is the challenge: user-level and aggregate, attribution and conversion.
The result is that you can still run your mobile business. That you can still grow and succeed. At a certain level, it’s definitely getting murkier. But the right tool can still provide the insight you need.