Introduction to ROAS campaigns
Learn about Return on Ad Spend (ROAS) campaigns, optimization options, and eligibility with Unity Ads User Acquisition.
Read time 4 minutesLast updated 2 days ago
Note
The current version of D28 In-app Purchase (IAP) Return on Ad Spend (ROAS) campaigns is a beta release. The features and documentation might change between now and the next release.
ROAS goals
With a ROAS campaign, you set a ROAS goal (also known as Target ROAS or tROAS) that determines how the system bids for users. When you set a ROAS goal, Unity uses machine learning to predict each user's revenue potential and bid dynamically. Although Unity can't guarantee that your campaigns achieve your ROAS goal, the system prioritizes your target and aims to achieve the goal with dynamic bidding. If you set a low ROAS goal, the system can place higher cost-per-install (CPI) bids. This increases the number of installs, but can result in fewer high-value users. Conversely, if you set a high ROAS goal, the system places lower CPI bids, which can decrease the number of installs but targets a higher return.Optimization types
When you create a ROAS campaign, you select an optimization type based on how users generate revenue in your app. The following table illustrates what kind of user each optimization type prioritizes:Optimization type | Acquired users |
|---|---|
| In-app Purchase (IAP) | Users likely to make in-app purchases |
| Ad Revenue | Users likely to engage with in-app ads |
| Hybrid | Users likely to generate revenue through both in-app purchases and ad engagement |
Optimization windows
After you select your optimization type, you set an optimization window for your campaign. This window determines the timeframe of revenue prediction for a user. For example, if you select a day-seven (D7) window for your Ad Revenue campaign, the system predicts how much ad revenue a user will generate in the seven days after installing your app. Refer to the following table for which optimization windows are available for each optimization type:Optimization type | Optimization windows |
|---|---|
| In-app Purchase (IAP) |
|
| Ad Revenue |
|
| Hybrid |
|
Post-install event data
To run a ROAS campaign, Unity needs data about how users engage with your app after installation. This Post-install event data allows Unity's machine learning models to predict user revenue and optimize your campaign based on your chosen optimization type. You can share this data with Unity through two methods:- Mobile Measurement Partner (MMP) integration: Third-party providers that track and organize app data for advertisers
- Custom server-to-server integration: Direct integration between your app and Unity's system
MMP integration validation
When you integrate your preferred MMP, the Unity Dashboard validates your integration to prevent critical errors that can disrupt optimization, attribution, and campaign performance. If the validation detects errors, the dashboard displays a warning message detailing the issue and how to resolve it. You can't launch paused or new campaigns until you resolve all required integration issues.Data cohorts and maturity
The group of users that Unity's models learn from is often referred to as a cohort. Unity's ROAS optimization options optimize toward multiple cohort windows (D0, D7, and D28). The Learning phase thresholds for each optimization type require different mature cohort windows. Refer to the following table for examples of the earliest maturation date for each cohort window:Cohort window | Date of data passed | Earliest cohort maturity |
|---|---|---|
| D0 | January 1 | D0 cohort mature on January 3 |
| D7 | January 1 | D7 cohort mature on January 10 |
| D28 | January 1 | D28 cohort mature on January 31 |
Learning phase
When you launch a new ROAS campaign, it enters a Learning phase in which Unity's models collect the relevant post-install event data for your optimization type. This allows you to launch ROAS campaigns immediately without first running an Install campaign to gather data. During the Learning phase, your campaign might experience performance fluctuations as the machine learning models train. Campaign performance typically stabilizes once the Learning phase completes and the campaign has collected enough data. Your campaign might not achieve its ROAS goal during this phase, but performance typically improves once the campaign reaches Live status.Learning phase thresholds
The Learning phase lasts until your campaign meets the necessary learning thresholds for your optimization type. Refer to the following table for the data thresholds required for each Optimization type:Optimization type | Optimization window | Requirement to exit Learning phase |
|---|---|---|
| IAP | D7 | 75 unique, D7 matured purchasers |
| IAP | D28 | 75 unique, D7 matured purchasers |
| Ad Revenue | D0 |
|
| Ad Revenue | D7 |
|
| Hybrid | D7 |
|
Monitoring the Learning phase
To monitor your campaign's progress toward learning thresholds, use the Reporting Dashboard with the following settings:- Apply the Advertiser Game ID filter.
- Set the reporting window to Last 90 days.
- Select metrics based on your optimization type. Refer to the following table for details.
Optimization type | Metrics | Learning completion |
|---|---|---|
| D7 IAP |
|
|
| D28 IAP |
|
|
| D0 Ad Revenue |
|
|
| D7 Ad Revenue |
|
|
| D7 Hybrid |
|
|