At a glance: A privacy-preserving mechanism designed to measure the effectiveness of ad campaigns without relying on user identifiers. Managed on-device, with last-touch attribution. No user data is exchanged between ad networks, which preserves user privacy.
Introduction
Definitions
- Ad tech - Ad tech in the context of the Android Sandbox Attribution API includes mobile measurement partners (MMPs), advertisers, and ad networks, all working together to measure the performance of ad campaigns while maintaining user privacy.
- Sources - ad clicks or views.
- Triggers - conversions such as installs or post-install events.
- post-install attribution exclusivity window - The post-install attribution exclusivity window in the Sandbox Attribution API is a period during which only the installing ad network can attribute post-install events, ensuring that other networks cannot claim attribution for those events within this time frame.
About Android privacy sandbox attribution API
How the Attribution API works
Setup for Ad techs to use the attribution API:
- Complete the enrollment process.
- Register attribution sources such as ad clicks or views.
- Register triggers including user conversions on the advertiser app.
The Attribution Reporting API matches triggers to attribution sources and attribution conversions and sends one or more triggers off-device through event-level and aggregatable reports to the ad techs.
The source contains all the data and information about the ad network, including information about the click or view and campaign details. Multiple sources can be stored on the device.
The trigger holds measurement data and data on the install and post-install conversion. Multiple triggers can happen from different ad networks or MMPs.
An ad tech must register both attribution sources and triggers to receive postbacks from the API.
Note: that the same ad tech must register both the attribution sources and the triggers to receive postback from the API.
AppsFlyer Sandbox attribution solution
Principles and flows
To overcome the functional and structural limitations of the Sandbox attribution infrastructure, the AppsFlyer Sandbox attribution solution takes into consideration the characteristics in the table that follows
| Characteristic | Description |
|---|---|
| Data integration |
|
| Data enrichment |
|
| Data delivery | Enriched Sandbox attribution data is available using the following: |
| Seamless integration |
|
| Agility | Advertisers can change measurement configurations as needed in the dashboard |
Additional Privacy Sandbox attribution information
Reporting options
Sandbox attribution reports don't contain user-level identifiers.
| Report name | Type | Enriched by AppsFlyer | Delivery method | Premium feature |
|---|---|---|---|---|
| Sandbox overview dashboard | Aggregated | Yes | Dashboard | No |
Sandbox attribution solution implementation checklist
| No | Action | Description | Responsibility |
|---|---|---|---|
| 1 | Prepare the app | Migrate the app to AppsFlyer V6.13.1 SDK or above. Note: SDK V6.16.0 or above is needed for revenue.
|
Developer |
| 2 | Partner integration |
Ad network partners must integrate with AppsFlyer using Sandbox attribution integration. See here a list of Sandbox attribution integrated partners. Decide if you want to allow partners to redirect the triggers to the ad network partner. See more details here. |
Marketer |
Troubleshooting guide
| Symptom | Description |
|---|---|
| The dashboard doesn't contain any installs |
|
Timing from app install until dashboard and reports update
Sandbox aggregatable reports received on a given day are processed at the end of the next day—UTC starting at 00:01. The data is available 8 hours later. In other words, it takes around 32-56 hours to process the data, with an average delay of 44 hours.
For example, a report received on Monday is processed and ready on Wednesday. The dashboard will be updated by Wednesday 08:00 UTC.
Traits and limitations
| Traits | Remarks |
|---|---|
| Agencies | Agencies can indicate that they brought the user by implementing Sandbox attribution agency recognition |
| Agency transparency | N/A |
| Timezone | UTC |
| Currency | N/A |
| Organic data | N/A |
| Non-organic data | Based on Sandbox attribution reports |
| Data freshness | Daily |
| Historical data | Depends on the time the ad network is integrated with AppsFlyer. Earliest date is August 19th, 2024 |
| User access | Account users must have the necessary permissions |
Redirecting triggers to ad networks
Within the Privacy Sandbox, each individual ad network has the capability to view the last-touch attribution flow within its network. This means that an ad network can attribute conversions (such as app installs) to the last ad interaction that occurred within their network.
However, this capability is different when comparing it to the broader attribution provided by AppsFlyer within Sandbox:
AppsFlyer can encompass multiple ad networks and provide a holistic view across different networks and declare which network was the last touch before the conversion.
Example Scenario: A user sees and clicks on an ad from Unity Ads on Day 1, then clicks on an ad from TikTok on Day 2, and finally installs the app on Day 3.
Ad network's last-touch attribution: Unity Ads and TikTok can each see if their ad was the last touch before the conversion within their own network. Both Tiktok and Unity will receive winning reports for their own ads under Sandbox attribution.
AppsFlyer last touch attribution: AppsFlyer can view the entire user journey across both Unity Ads and TikTok, identifying that the user interacted with both ads before converting. AppsFlyer can attribute the conversion to TikTok as it was the last touch.
If you want to allow the ad network to view the last touch attribution flow within its network you will need to provide written consent to your AppsFlyer rep so we can then redirect those triggers to the ad network. Please reach out to your AppsFlyer rep for more details.
In the future, we will add such a capability within the AppsFlyer platform to avoid the need for manual written consent.
Comparing Sandbox attribution results to AppsFlyer attribution results
It is not possible to properly compare the Sandbox results to AppsFlyer attribution results for the following reasons:
- There are fewer ad networks currently integrated with Sandbox attribution compared to AppsFlyer. Therefore, in the case of multi-touch and cross-network attribution flows, the results can differ. For example,
Day 1: User clicks on an ad run by Unity
Day 2: User clicks on an ad run by TikTok
Day 3: User installs App
In the regular AppsFlyer flow, TikTok - being the last touch point, will win the attribution. However, with Sandbox, assuming TikTok is still not integrated with Sandbox, Unity Ads will win the attribution. - Sandbox attribution results can also contain "noise". Which is the addition of random data to user-related information to protect individual privacy. This technique is used to obscure specific details about user actions, making it more difficult to measure or identify individual users while still allowing for the aggregation and analysis of overall trends and patterns.