first party attribution-minWhat happens when you need or want to keep your attribution data (install sources & user data) in-house?

Historically, attribution has always been a struggle. Being able to understand where your users are from and what the most engaged cohorts are is a key factor in improving your overall marketing efforts.

attribution providersSome attribution tools that can help

There are solutions, and they’re great. Adjust, Appsflyer, Branch or Kochava are providing not only great attribution dashboards and accurate representations of the user’s distribution, but also the power to run advanced optimization strategies based on funnel conversion rather than simple acquisition numbers.

But what happens when you can’t –– or don’t want to –– use these tools; when you want to keep all identification, attribution and engagement data about your users within your grasp  and in your own database?

I’ll go through the reasons why you may need to track everything as a first party; then I’ll explain why you may be limited in doing so when optimizing your campaigns. Finally, I’ll share some opportunities for each channel to keep optimizing your ad’s for performance.

1) Legal compliance with the app stores

A couple of months ago, Apple published an update on its guidelines:

The latest apple guidelines on third-party tools & kids’ apps

In short: if you have a kid’s app or any app that tracks minor users’ behavior, you’re not allowed to send any identification data (device ID, IDFA, email, etc.) to third-party tools.

Update 23/06: During WWDC, Apple expressed its intent in bringing more privacy rules and more clarity on how each data point is being used for all users

Even if the way these new privacy rules will impact app store guidelines is not specified yet, we can agree that it will lean towards users’ control, therefore reducing what you can do with third-party tools.

2) COPPA and GDPR compliance

Similar to Apple’s guidelines, COPPA imposes “certain requirements on operators of websites or online services directed to children under 13 years of age”. 

Collecting personal information online from a child under 13 years of age is illegal, creating further complexity when it comes to attribution and engagement analysis. 

There’s no need to present the GDPR anymore, which is also impacting the way you’re allowed to manipulate and transfer such personal data.

3) No dedicated budget to attribution tools

As I said, attribution tools are great; but they are not always aligned with every company’s budgeting plans especially when you’re only running ads on specific networks (Facebook/Google and Apple Search Ads), making it hard to justify such budgets as they’re based on monthly install numbers.

4) Willingness to keep everything in-house

This is actually a strong reason. Some companies just want to avoid sending any type of data.

Reasons as to why: 

  • Internal guidelines on data processing and third-party tools usage 
  • Company’s philosophy/mindset on offering their users better privacy
  • Data analysis strategy: the company being more comfortable conducting their own analysis & optimization strategies and leveraging their own set of dashboards and tools

5) How long can we actually keep working with the IDFA/AAID?

The IDFA(for iOS)/AAID (for Android) is the advertising ID that is used to track users across apps and networks. It’s attached to a specific device and doesn’t change.

Some experts are predicting that it may disappear in the near future, and its usage has been heavily criticized in the industry.

Attribution tools will have to figure out another way to keep their attributions as accurate as possible, but it may take some time before they can adapt. 

UPDATE 23/06: IDFA is taking the next exit

In the latest Apple WWDC, they announced new privacy rules regarding users’ data tracking. One of these announcements is the following:

It’s now required to ask for permission to deliver personalized ads. In other terms, it means you need to ask users to allow IDFA tracking. 

It’s not hard to guess the percentage of users who will end up denying this prompt –– from 30 to 50%, probably –– giving attribution providers a hard time and making your capacity for optimizing advertising campaigns more complex as well. 

Attribution: App Installs vs Conversions

The main reason why we use attribution tools is that we want to better understand our advertising campaigns. There are two ways to do this:

  • Install optimization: trying to optimize campaigns based on how many installs each keyword/creative is bringing and the avg. CPI (average cost per install)
  • Conversion optimization: improving budget allocation and identifying best keyword/creative/adgroup/etc. for each type of conversion, ranging from sign up to purchase, allowing a better understanding of your (ROAS) return on ad spend

Where install optimization will be limited in improving the number of new users, conversion optimization allows you to understand the true ROAS of each of your campaigns, helping you acquire more valuable users.

That’s where attribution tools come into play: they allow you to track conversion events and match them with attribution data.

But, as we said, you need to integrate their SDK (software development kit) and send them data, as attribution tools rely on the IDFA/AAID (and other identification data).

Go First Party, One Channel at a Time

I’d like to share some thoughts on what companies can do if they can’t –– or don’t want to –– use attribution providers to ensure they’re still able to optimize their marketing efforts towards conversion & engaged users  ( at least at the channel’s level).

1) Don’t build everything in-house

Building an in-house attribution system will require a lot of technical resources for its development and maintenance.

If there are just a couple of well-known attribution tools, it’s because they are resource heavy and extremely hard to build. Don’t expect to craft your own multi-touch attribution system within months –– or even years. 

2) Using channels’ dashboards

This may be feasible for some channels that will not ask you to send them identification data. 

The bigger ones will end up complying with these rules; some already are (if you’re running Apple Search Ads campaigns, you’ll be OK).

But here is the catch: channels/networks often don’t offer the same level of reporting that you could expect from an attribution provider, and there is no conversion tracking (or at least not without including the identification data we talked about).

Apple Search Ads Use Case

If you’re looking into Apple Search Ads, you can’t track conversion or your return on ad spend. This means that you’ll only be able to track app installs and optimize your campaigns based on that.

This isn’t ideal if you’ve allocated a decent budget to the channel, as you’ll probably have a hard time reaching –– or even calculating –– a correct ROAS. 

There are tools like Search Ads HQ that allow you to send over conversion events, so you can optimize your campaigns towards these events, instead of only looking at app installs.

Unfortunately, Search Ads HQ relies on either attribution tools to send them events or server-side connections.

That’s why 2nd Potion, in collaboration with Phiture, has crafted Elixir –– one of the first first-party attribution enablers.

The way it works: we’ll collect Apple Search Ads attribution data, as well as conversion events triggered by users, and send it over to a specific destination (Search Ads HQ or even a company’s internal warehouse).

Elixir doesn’t store any data as everything is sent from the user’s device, straight to the destination, without including the IDFA. 

Elixir offers a way to get conversion and attribution data in the same place while ensuring the respect of users’ data privacy. Even if it doesn’t replace it, it can be used in addition to your analytics solution for campaign optimization.

A quick overview of the way Elixir works with Apple Search Ads

In this setup, you can decide to send data from your app, directly to your database using Elixir, without any third party receiving/storing any of it.

That means you own and control all your attribution data (in compliance with Apple). It also means that the pricing won’t scale based on MAU (monthly active users), but only based on the storage size going from 10k to 1m MAU won’t cost you more than the server charge.

This doesn’t fix all the issues, but it gives a simple answer to a complex problem: enabling a campaign’s optimization based on conversion for Apple Search Ads.

Google And Facebook

Google provides its own attribution tool, which is accurate for google networks but less likely to be accurate for others

BUT, Google uses IDFA/AIDA to identify users, which means you can’t track installs if you’re not using the IDFA or not ready to integrate the Firebase SDK.

Regarding Facebook, you’ll need to communicate your data in order to track installs. At present, there aren’t any known ways to implement Facebook install attributions without integrating their SDK or a third-party SDK

Both networks rely on you complying with the game’s rules: integrating their SDKs and/or sending your users’ IDFA/AAID. 

UPDATE 23/06: Ready to comply?

As Apple is pushing towards more control on IDFA, we can surely expect Google and Facebook to reply and offer new ways to track installs and conversion events without IDFA/AAID or identification data. 

Maybe it will be through Apple’s SKAdNetwork (a framework built to allow privacy in ad networks tracking).


There are two ways to run campaigns on TikTok/Snapchat: either through their ad manager or with influencers campaigns (I’ll tackle this one later). 

At the moment, both Snapchat and TikTok rely on IDFA and third-party tracking tools to optimize conversion events. 

This means that if you want to get accurate data on true app installs metrics and/or conversions, you’ll need to work with the third-party and share your users’ IDFA.

Only time will tell if these platforms decide to move towards a tracking methodology with better privacy.

Other networks/Influencer’s tracking (Snapchat/TikTok/Instagram/Youtube)

Without going into details, it’s fairly easy to rely on iOS or Android methods when handling universal links to set up a custom domain on your server and get accurate Influencer campaign metrics.

What may be harder is to track these users down the funnel by looking at their engagement and monetization behavior.

With Elixir, we’re leveraging the same methodology we use for Apple Search Ads in attributing installs and conversions, enabling true performance marketing with your Influencer marketing.

And because some ad networks or other networks allow universal links in their ads, you can also track these networks with the same strategy.

If Facebook or Google changed their guidelines in universal link usage in their ads, it would mean that we could already track, with accuracy, direct installs coming from their ad networks.

So.. What’s Next?

Apple’s latest conference has shown us that we can’t rely on the IDFA/AAID and traditional third-party tracking approaches to keep optimizing our advertising campaigns.

Going first party is an option, and many initiatives like Elixir will continue to emerge and offer ways to take the initiative. 

We may be far from the accuracy of attribution partners (at first) and many channels may take some time to adapt, but we can already see how it can work with networks like Apple Search Ads and Influencer campaigns, and we can see that it’s just a matter of months before Google, Facebook, Snapchat and TikTok will follow.

If you’re facing a similar challenge, we’d be happy to chat 😎

Thanks to Alex Billaud, Moritz Daan and Nicholas Fraser