Skip to main content

​
Overview

This guide walks advertisers through uploading first-party data to Clean Room data sources for matching with publisher data.
Prerequisites:
  • First-party data prepared with common identifiers (hashed emails, mobile IDs, etc.)
  • Data taxonomy defined

​
Steps

1

Navigate to Data Sources

In the left menu of your demand-side workspace, click Data Sources.
2

Add a new source

Click Add Source.
3

Provide source details

Enter a name and description for your data source:
  • Use a clear, descriptive name (e.g., “Lapsed customers”)
  • Include details about the data set in the description
4

Prepare taxonomy file

Prepare a taxonomy file in CSV format. The taxonomy defines segment IDs and names.
5

Upload taxonomy

Click Upload Taxonomy and select your file.
6

Access GCS bucket

Once the taxonomy is validated, you’ll be granted access to a GCS (Google Cloud Storage) bucket.
7

Upload raw data files

Upload your raw data files to the provided GCS bucket using the following format:
  • File format: Tab-separated file (TSV) with NO headers
  • Columns (in order): id, tag, segments
    • id: The user identifier (e.g., hashed email, mobile ad ID, IP address)
    • tag: The identifier type - use one of: email_sha256, appnexus, or ip_address
    • segments: Comma-separated list of segment IDs as defined in your taxonomy file
  • Example row: abc123def456\temail_sha256\tsegment_001,segment_042,segment_103
Data will begin processing automatically.

​
Permissioning Data (Publisher-Driven Workflow)

If following the publisher-driven workflow, you also need to permission the data source to the publisher’s demand-side workspace:
1

Get publisher's Organization ID

Ask the publisher for their demand-side Organization ID (found in their Settings).
2

Grant permission

Input the publisher’s Organization ID, set any time limits or usage restrictions if needed, and click Grant Permission.
Data preparation best practices:
  • Hash emails using SHA-256 — lowercase hex encoded — before uploading (do NOT upload raw email addresses)
  • Normalize identifiers before hashing: lowercase emails, trim values
  • Validate data quality and completeness before uploading large datasets
  • Test with a small sample file first to verify format is correct
Data processing can take several hours depending on dataset size. Monitor the data source status in the interface.

​
Next Steps

Creating Clean Room Audiences

Build audiences from your data

Back to Clean Room

Return to product overview