Back in March, Google shared that Android, the most widespread global operating system, is going to deprecate GAID — its user identifier for advertisers — by 2024.
GAID (Google Advertising ID) is an ID generated by Android for a device, which is used for advertising purposes. This level of abstraction gives the user a certain degree of privacy while maintaining high functionality in the advertising ecosystem.
The three main uses of GAID are:
- Marketing performance measurement (i.e attribution)
- Segmentation – building user lists for marketing purposes
- Ads personalization – presenting contextual ads based on users’ activities and interests.
In this piece, I’ll cover the main use cases for which GAID is currently being used by the ecosystem. I’ll also map the effect of this deprecation and the proposed solutions brought by Google to address these new challenges.
Let’s drill down each category, and analyze the expected impact of the deprecation of GAID — with the assumption that the Android Privacy Sandbox will go live as with the suggested roll-out timeline.
Marketing performance measurement (attribution)
In Android, the two main attribution methods are Google Play Referrer and ID Matching (GAID). When reviewing the proposed changes being made to these types of attribution methods, we noticed that Google’s latest Privacy Sandbox documentation doesn’t discuss the following:
- Deprecation of Google Play Referrer (which is actually a good thing!)
- A consent mechanism similar to Apple’s App Tracking Transparency. This is cause for the assumption that unlike iOS and IDFA, GAID is going to be deprecated as a whole (which isn’t necessarily a bad thing either).
If Android’s Privacy Sandbox will work in parallel to Google Play Referrer, this will still provide great attribution capabilities, and will allow advertisers (via their MMPs) to attribute users in real-time.
In addition, this will allow advertisers to support deep linking (and deferred deep linking) where the end-user can experience a seamless, contextual transition between the ad and the app.
Google Play Referrer is a great privacy supporting solution that doesn’t require the use or sharing of a user ID across multiple parties.
The two main downsides of Google Play Referrer are:
- Direct to install only – if a user clicks on an ad and doesn’t install the app immediately, Google Play Referrer will not attribute the user to that ad.
- Google Play store installs only – out-of-store apps cannot be attributed with Google Play Referrer.
The Privacy Sandbox’s on-device measurement is the Attribution Reporting API, which will provide attribution and conversion measurement across Android apps and the web. The main capabilities of the Attribution API are as follows:
- Each ad network will be able to receive attribution according to their own touchpoints and attribution settings (but not in a cross-network manner).
- Only MMPs will be able to generate attribution across different ad networks and advertiser-selected attribution settings.
- The reporting of the Attribution API can be further divided into:
- User-level reporting – mostly relevant to publishers, limited to 3 bits of reporting, and has delayed data freshness.
- Aggregated level reporting – a very extended set of capabilities with almost unlimited campaign and user properties breakdowns, an LTV of up to 30 days post install, and almost real-time reporting.
AppsFlyer’s vision is to unite all available attribution sources — like Google Referrer and The Privacy Sandbox — to create one holistic snapshot.
Segmenting users is heavily used today to generate lists of user attributes that can be sent to networks and marketing engagement tools for the purposes of engaging these users with tailored marketing content.
Although user-level lists will no longer be possible with Google’s new solutions, we can assume that the industry will adapt to either one or both of the following options:
- Using Google’s proposed FLEDGE solution.
- While GAID is the most ubiquitous identifier for these purposes, advertisers may use first-party identifiers, as long as it respects user’s permission and consent.
The FLEDGE solution suggested by Google offers great capabilities for advertisers to define their audiences and partner with ad networks to show ads, which are then segmented for on-device audiences lists. This solution maintains users’ privacy by working on-device without sharing the end users’ identifier.
The Privacy Sandbox and the deprecation of GAID will have a big impact on the ad networks’ marketing capabilities. It will mainly impact ad networks that have grown accustomed to collecting user activity data across third-party apps.
Unlike own-app context-app ads, Google has proposed the cross-app TOPICS solution.
According to the TOPICS solution, Android will serve each publisher app with different topics that are meant to help interest-based ad serving.
During the first phase, the mechanism will be built by learning from publicly known information that is uploaded to Google Play by advertisers when publishing their app. Some examples include: bundle names, app categories, etc.
Although we can already foresee that this is a weaker method for ad personalization, we can also presume that this solution is going to be greatly broadened as inputs to the TOPICS engine continue to grow.
With the deprecation of GAID and the lack of cross app identifiers, the advertising industry is definitely going to be shaken.
That said, at AppsFlyer we see this as a great opportunity for advertisers — including publishers and ad networks — to leverage the new proposed solutions that Google and us will be supporting.
As industry leaders and successful pioneers of similar privacy changes done in iOS, we are confident that Google’s change will be beneficial for both end-users and advertisers that adapt.