Ah, mobile attribution. I love this topic because every company goes through the same emotional process when they start to learn about it.
The story starts like this. Company A is ready to scale their acquisition campaigns. They have been relying on app store data (from Apple and Google) to know where their installs are coming from. This is ok but there must be a better way. Someone has definitely solved this problem by now.
They start researching providers (which is a mistake) and realized that this is way more complicated than they thought. Even after they implement a mobile attribution provider, they still have unanswered questions and are likely quite confused. The entire experience is fraught with disappointment.
Let me spoil the rest of this article for you. When it comes to mobile apps, EVERYTHING IS HARDER. Attribution is not the exception to this rule.
However, you can still make progress towards your goals despite these difficulties. Let’s jump right in with some basic terminology.
What is Mobile Attribution Exactly?
Mobile attribution simply means that you want to know what marketing channels or campaigns are driving new installs for your app. In theory, this should be a list that says campaign 1 drove 100 installs, campaign 2 drove 200 installs but you’ll quickly realize that this list is full of caveats.
Mobile attribution is equivalent to web attribution (which is much easier) but the mobile version is harder because of the “black boxes” known as app stores.
In short, when a user wants to install your app, they have to go through an app store before they can actually download your app. The moment a user steps through the app store, all analytics tracking stops and you’re left with a void where you won’t see the user again until after they install and open your app.
This “void” is what providers are meant to solve and most do a relatively good job but it isn’t as clear cut as attribution on the web.
Mobile Attribution & UTM Parameters
You might also see UTMs parameters within the mobile world. This helps some companies wrap their heads around how mobile attribution works, though the usage of UTMs isn’t identical to on the web.
Despite the technical challenges, you need mobile attribution to scale any user acquisition campaigns. Without it, you would be blind to how your campaigns are actually performing. In this article, I’ll help you solve some of these challenges (you can send me a check later).
What Executives and Growth Leaders to Know
My background is as a frontend engineer but I spent most of my time working with marketing and product executive. This means that we’ll talk about mobile attribution from a slightly different lens. At this level, the technical issues aren’t as important and instead we need to focus on the bigger picture.
Let’s start with one of the most important ideas that I share with all of my clients: the 3Ps.
3Ps: People, Process, and Providers
To successfully execute an analytics strategy, you need to focus on the 3Ps in the right order. We start with people which means identifying who is going to own this data within your company. Starting by choosing tools is a mistake even though almost every company does this.
You then want to figure out the right process for setting up and maintaining this data. Setting up tools is relatively straightforward, you choose a couple of engineers who can support this initiative and start coding. Maintaining goes beyond the technical work and also looks at how this data will be used in reporting and how people can access it on a regular basis.
Finally, we have providers or vendors. Despite the overwhelming number of vendors out there, most of them will do the job correctly. I’ll talk about the most common providers in a section below.
As you go through this process, keep in mind that “good enough” is the goal. It’s quite hard to get 100% accuracy in mobile attribution and striving for it out of the gate will simply cost you more in missed opportunities.
As an executive and growth leader, you should ideally focus on choosing the right people and providing your input on process and providers. Let your team figure out the bulk of the details.
Once you got this, we can talk about the 3 levels of mobile attribution and how choosing the right level is crucial for your company.
3 Levels of Mobile Attribution Complexity
Everyone wants everything to be perfect. You might have heard questions such as:
“Why is Facebook reporting higher numbers than our mobile attribution tool?”
“Why are we missing installs during this time period?”
“Why can’t we get a number that we can all trust?”
I have the answers for you but they aren’t what you expect. Before, we need to spend a few minutes talking about attribution models.
Despite their heavy use of math, attribution models are part science and part art. An attribution model is how a specific provider will determine which activity (channel, campaign, ad, etc) gets credit for a conversion (app install, purchase, etc).
This is where you will hear things like Last Touch, First Touch, etc. The important thing that you need to know is that every company has a slightly different attribution model. This means that comparing numbers across two providers is almost impossible. You need to account for a lot of different variables, some of which aren’t even publicly available to you.
Instead, you should choose one attribution model that you like (by understanding how it works) and stick to it. This becomes your source of truth and every other attribution model is simply used to directionally point you in the right direction.
In practice, this means that you might choose Branch.io as your primary attribution model. If you’re running campaigns on Facebook, you would use their attribution model (which is quite different) to directionally point you in the right direction i.e. are people clicking on the ads, are the costs going down over time, etc.
Almost every question around mobile attribution comes down to a lack of trust in the attribution model or trying to compare two different attribution models.
That being said, I also want to talk about the 3 levels of mobile attribution complexity. Let’s not run before we can walk even though some companies want to fly before they can even crawl.
Level 1: Mobile Only and Last Touch
At this level, you’re focused on understanding how different campaigns and channels are driving installs. Everything is seen through the lens of a last-touch attribution model. This does mean that some channels will get less credit than they would in a multi-channel model.
You will typically only need one provider that can do all of your reports. This is a great place to be because the complexity is low and you can experiment quite a bit with your campaigns.
Level 2: Web + Mobile and Last Touch
In level 2, you’re focused on connecting web AND mobile attribution. You might have a cross-platform product or simply a website where some activity can happen.
Instead of looking at things on a separate basis i.e. one tool for web attribution and one tool for mobile attribution, you can start to combine it into one single report. You’re still limited to the last touch.
This is also where you’re starting to dive deeper into more flexible data formats such as data warehouses and SQL based reports.
Level 3: Multi-Channel Attribution
Level 3 is all about multi-channel attribution. You shouldn’t be at this level unless you’re spending $1M/year in user acquisition. You will need to hire data analysts who can combine all of the raw data into a cohesive report that makes sense but you’ll get a clear picture of the roles that different channels play.
Let’s Talk About Providers, Baby
If you jumped straight into this section, you should go read about the 3Ps, you’ll thank me in six months.
However, if you’re ready to choose a provider, then welcome! I was listening to some Salt-N-Pepa as I wrote this section so it might come across as a rap verse.
As I mentioned in the introduction, most providers do a relatively good job of shining some light into the black boxes known as app stores. They all do this by developing their own attribution models which take into account different user identifies such as device IDs, fingerprints, etc.
Once you choose a provider that you like, I would recommend spending some time reading about their attribution model but you should know that you don’t have much control over it. You take what you’re given and you have to do your best so I’ll skip the technical details in this article.
There’s a handful of providers that dominate the space. I have worked with almost all of them and they are more similar to each other than different. In no particular order, they are:
As you choose a provider, think about the following things:
- How solid is their ability to deep-link users into your app?
- Are they a Facebook partner? This is crucial because it gives them access to more information about your campaigns and not everybody is a partner.
- Do you like how they report data i.e. does it make sense to you?
- Do they have any other features that you care about e.g. organic app store performance, tracking referrals, etc?
- Do they support any niche advertising networks that you want or need to use? Most spending goes to Google and Facebook but some companies will need to use very specific advertising networks.
- Do they easily integrate with any of your other tools like Mixpanel and Amplitude?
- Do they have an SDK or library for your mobile app? This goes for all those apps that are built on experimental frameworks like Xamarin or something obscure.
I have to also share some notes about the technical implementation process. You will need to add the SDK for these providers to your app, no question about that.
Your choice of the mobile framework will also affect how easy or hard it is. Almost all the problems that I have seen with mobile implementations come down to my client using a framework that isn’t well supported yet. Xamarin, React Native and whatever else is out there comes to mind. This isn’t show stopper but just know that it will likely cause issues.
Finally, you can typically implement different features depending on what you need. You could start with the attribution portion and then come back and add deep linking capabilities. Implement what you need to get started and add more things in the future.
I find mobile marketing fascinating but it can be quite complex compared to what you can do on the web. I wish I could say that things are getting better but I think tracking issues will continue to be an issue on mobile.
Companies like Apple are leaning more towards privacy and making it harder to track users on iOS devices. I would expect Google to follow suit in some areas.
However, if you get the 3Ps right: People, Process, Providers, you should be in the right place to solve all of these challenges, now and in the future.
If you have any specific questions that I didn’t cover, put them in the comment section below.