Providing Ad Network Performance Data
Supporting Products
- Probabilistic Attribution
- Organic Lift
- Intelligent Budget
Introduction
AlgoLift's Intelligent Budget and Organic Lift products require a daily feed of network performance data attained from your marketing analytics provider / cost aggregator. This data when paired with user-level predicted LTV allows us to model return on ad spend (ROAS) for each channel, as well as determine the relationship between paid UA activity and organic installs and revenue. We also get this data through network reporting APIs, but providing data in the format below from a cost aggregator allows us access to data for all channels, including those that we do not yet have API integrations with.
Data History Requirements
The further back you can provide historical data, the more effective our algorithms will be. At minimum, we need 3 months of historical network data for Intelligent Budget allocations and 6 months of historical network data for Organic Lift projections.
Pulling Ad Network Performance Data from Singular
We can pull ad network performance data from Singular's API, to do this we need your Singular API key. More details on where that is in Singular's documentation
Once you have the API key please use a service such as https://onetimesecret.com/ and email to support@algolift.atlassian.net
Network Performance Data
When providing network performance data for export to AlgoLift, please format it according to the following schema:
| Name | Data Type | Examples | Required | Description |
|---|---|---|---|---|
| account_id | varchar(255) | act_100000 | Yes | Ad Account ID |
| source | varchar(255) | apple search ads | Yes | Source / Channel/ Ad Network Name |
| campaign_name | varchar(255) | FB_LAL10p | Yes | Campaign name |
| campaign_id | varchar(255) | 12345623562455474 | Yes | Campaign ID |
| ad_group_name | varchar(255) | ADSET_FB_LAL10p_US | Yes | Ad group name |
| ad_group_id | varchar(255) | 23434553445354565 | Yes | Ad group id |
| keyword_name | varchar(255) | games | Yes | Keyword name (when applicable) |
| keyword_id | varchar(255) | 9382720 | Yes | Keyword ID (when applicable) |
| country | varchar(2) | US | Yes | Country in ISO-3166 alpha-2 country code format |
| platform | varchar(255) | iOS | Yes | Platform (e.g iOS) for the app |
| ad_name | varchar(255) | CRV_FB_LAL10p_US_red_scarf | No | Creative Name |
| ad_id | varchar(255) | 77382929838093948 | No | Creative ID |
| ad_url | varchar(255) | s3:/algo/assets/animated.mp4 | No | The URL to the creative assets (Image/Video) |
| subpub_name | varchar(255) | CAKE_MANSION_7CD124328C | No | The name of the publisher (app) in which the ad has been displayed. |
| subpub_bundle_id | varchar(255) | VUNGLE_And_US_LAUNCH_DIF1231 | No | The public ID of the publisher (app) in which the ad has been displayed. |
| subpub_id | varchar(255) | 5H7J4AQBE2EECF7AF09B5512 | No | The hashed ID of the publisher (app) in which the ad has been displayed |
| impressions | int | 100000 | Yes | Impressions |
| clicks | int | 10000 | Yes | Clicks |
| installs | int | 1000 | Yes | App Installs |
| cost | double | 5050.99 | Yes | Cost (spend) |
Sample Data
Please click here for a copy of sample data for your reference.
Writing Data to AlgoLift's S3
As with user level data, the network performance data should be written to AlgoLift's S3.
The path should use the following format:
s3://algolift-<client>/<app>/ingestion/daily_data/network_metrics/date=<date>/