Configure an Amazon Marketing Cloud First-Party Upload
If you're using the Amazon Marketing Cloud (AMC) integration, LiveRamp can work with you to also upload off-Amazon data into AMC. Bringing advertiser-uploaded data into AMC enables you to leverage first-party data in combination with event-level data available through AMC for more granular data cuts and insights.
Some examples of advertiser-uploaded data include:
Off-Amazon conversion events
In-store visits and purchases
Product metadata
CRM audience segments
Advertiser mapping tables
See the sections below for the requirements and prerequisites needed to successfully upload first-party data to AMC. Your LiveRamp representative will work with you directly to complete the process.
Prerequisites
Before getting started with an AMC first-party data upload, the following prerequisites must be completed:
Configure the Amazon Ads API connection to LiveRamp Clean Room with the help of your LiveRamp representative.
Configure an AWS S3 bucket to store first-party files. This should be a new S3 bucket that is not tied to the AMC instance. S3 bucket requirements include:
12-digit data upload AWS account ID used for the data upload (used to grant AMC access to read the first-party upload files).
S3 bucket read access for AMC to read the data files. Enable the following permissions:
s3:GetObject
s3:GetObjectVersion
s3:ListBucket
s3:GetObjectTagging
s3:GetBucketTagging
Add a tag to the bucket
Key : instanceId Value : {your AMC Instance ID}
Steps to Configure an Amazon Marketing Cloud First-Party Upload
Once the prerequisites have been completed, perform the following steps to configure an AMC first-party upload:
Prepare your formatted data files according to requirements and schema. AMC supports two dataset types: fact and dimension. See the table and information below for the formatting requirements for these dataset types.
Dataset Type
Usage
Upload Refresh Type
Requires Timestamp
Requires Partition Scheme (Period)
Supports Hashed Identifier Columns (ID Resolution)
Fact
Time series data
Incremental
Yes
Yes
Yes
Dimension
Static tables
Full
No
No
Yes
CSV and GZIP data files are supported.
Hashed identifiers must be normalized and hashed using SHA-256 before being uploaded to AMC.
Fact (time series data) must be partitioned according to the partition type (period), which could be per minute, hour, day, or week. Multi-part files are accepted.
Provision an S3 bucket using one of the following methods:
Your organization provisions and hosts an S3 bucket for storing first-party data files.
LiveRamp provisions and hosts an S3 bucket for storing first-party data files.
Enable S3 bucket permissions using one of the following methods:
Customer-hosted: Your organization grants S3 bucket permissions to LiveRamp for your AMC instance to read the files. S3 bucket permissions should include:
s3:GetObject
s3:GetObjectVersion
s3:ListBucket
s3:GetObjectTagging
s3:GetBucketTagging
LiveRamp-hosted: Your organization puts files in LiveRamp-hosted and provided S3 bucket.
Load data into AMC.
Note
Data load times may vary based on data volume and ID resolution dependencies.
The following occurs during the data loading process:
A schema is created via an API call. This is a table with the required column names and fact/dimension data.
Note
The dataset ID (AMC table name) can only contain lowercase letters (a-z), numbers (0-9), and underscores ("_").
The file is put in the S3 bucket according to the defined schema.
Note
If multi-part files are used, a manifest file is required. This is a text file that includes the paths for all of the multi-part files. For example, if there are 100 files, 100 paths will be listed in the manifest file.
Questions are authored joining the uploaded first-party data with the AMC data.
Troubleshooting
If queried data returns empty rows, the minimum count of users may not have been met. AMC requires a minimum of 100 distinct users for a row of data to populate.
The most recent upload is what is queryable. If a dataset update is needed, perform a new upload to replace the prior file.
If a table needs to be dropped to fix a data load error, the contents of the table must be deleted before the table is dropped. This can be done via an API call.