Curated REDS feeds

Curated Raw Event Data Stream (REDS) are privacy-focused data solutions for third-party data measurement and analytics partners, supporting them with measurement, optimization, and analysis. They are available as Core-, Audience-, and Platform-Curated REDS, as well as Aggregated Curated Feeds.

IMPORTANT: We regularly upgrade our curated REDS. When new schema version is released, you must migrate within the timeline communicated to you.

What you need to know

Here's what you need to know about curated REDS:

The curated REDS framework is designed to empower partners to act on behalf of The Trade Desk clients while minimizing the risk of unnecessary data exposure.
To accommodate evolving data structures, use schema versioning and ensure backward compatibility.
There are two integration options: direct integration and LiveRamp integration. If you're a vendor who relies on RampIDs or LiveRamp to resolve TDIDs and DAIDs, integrating with LiveRamp is the simpler option. However, you can use only Core-curated REDS.

The following sections describe best practices for optimization, feed options with key features and use cases, integration options, and schema versioning to accommodate evolving data structures. For the full workflow, see Getting started with curated REDS feeds.

Best practices

To optimize your curated REDS, follow these recommendations:

Recommendation	Notes
Follow the delivery schedule	Process files daily or according to the data provider’s delivery schedule. Use timestamp-based loading or hashed filenames for secure and incremental data ingestion. To confirm data is received and processed, monitor hour and day prefixes and verify files against expected delivery timelines. To identify missing or repeated datasets, keep clear records of ingested files. This ensures your data is reliable and easy to check.
Be ready for changes	When parsing files, use file headers instead of column positions. To accommodate evolving data structures, use schema versioning and ensure backward compatibility.
Build for fault tolerance	Develop processes to handle corrupt files or erroneous rows gracefully. Ensure the pipeline continues to process valid data while flagging or isolating issues for review. If you encounter characters like `'` or `"`, escape them on your end and avoid treating them as delimiters during data processing.
Make reprocessing safe	Design ingestion logic to re-execute files safely without duplicating data. Track each ingested file, and enable reprocessing historical data, if needed. To streamline data reprocessing and querying, store data in hourly or daily partitions.
Parse data as delivered.	The Trade Desk delivers field values as they appear in the platform and does not modify data for individual consumer parsing needs. The exception is characters that violate the TSV format (tab, line feed, carriage return), which The Trade Desk handles.
Ingest only the fields and tables relevant to your use case.	As datasets grow, you're responsible for managing your data pipelines accordingly.

Feed options

Here are the available curated feed options, along with their key features and use cases.

Feed Option	Data Type	Use Cases	Segment Names	Data Scope/Fit
Core-curated REDS	Log-level data	Measurement, optimization, lift	Not included	All impressions
Audience-curated REDS	Log-level data	Measurement, optimization, lift	Included for your own audience	Only impressions associated with your own audience
Platform-curated REDS	Log-level data	Analysis, ID resolution	Not included	All impressions
Aggregated curated feeds	Aggregated	Measurement, optimization, lift	Not included	All impressions

The following sections explain each feed option in detail, including its purpose and use cases.

Core-curated REDS

To prioritize receiving all campaign impressions regardless of which audience segment generated them, use the Core-Curated REDS feed. It allows access to log-level data on behalf of our clients for measurement, optimization, and lift use cases. For details, see the schema.

Audience-curated REDS

If you need segment-level attribution and want access only to impressions tied specifically to your own audience data, use the Audience-Curated REDS feed. This limits log-level data to only impressions influenced by your audience targeting and ensures privacy while providing visibility into the performance of each segment.

This option ensures you get the impression measurement and attribution data you need based on identifiers like UID2s or TDIDs, while maintaining strict boundaries around data ownership and exposure. To protect user privacy and prevent segment leakage (for example, preventing one vendor from reverse-engineering another's audience), The Trade Desk does not publicly disclose segment names. However, you can use this feed option to access only impressions that were decisioned using your own segments, including the segment names you provided. This approach enables accurate matching and measurement without exposing data from other providers. For details, see the schema.

If you need all impressions from a campaign, regardless of which audience drove them, then use the Core-curated REDS feed.

To help you choose a Core- or Audience-Curated Feed, see the following decision workflow:

Platform-curated REDS

To ingest log-level data for our clients who want to analyze and use their own impression data, use the Platform-Curated REDS feed. For details, see the schema.

NOTE: All Curated REDS data practices apply. For more details, see the retention policy in Direct integration with curated REDS.

Aggregated curated feeds

To leverage aggregated daily data on behalf of our clients, use the Aggregated Curated Feeds option. For details, see the schema.

Integration options

There are two integration options for accessing Curated REDS:

Direct: Use this option if you work with UID2, TDID, or DAID, and can match them to other identifiers in a graph. It provides REDS data directly through our AWS S3 bucket and offers full control over data ingestion and processing.
LiveRamp: Use this option if you are a vendor who relies on RampIDs or LiveRamp to resolve TDIDs and DAIDs. It simplifies data sharing and identity mapping by using the existing LiveRamp infrastructure. However, you can use only Core-curated REDS. For details on this feed type, see its schema.

Schema versions

To accommodate evolving data structures, use schema versioning and ensure backward compatibility. The following table lists the curated REDS and data types associated with each schema version. It includes links to the schemas detailing what fields are included for each data type.

Feed	Clicks	Conversions	Impressions	Video Events	Aggregated
Core-Curated	1-1-0	1-1-0	4-0-0	2-0-0	None
Audience-Curated	1-1-0	1-1-0	4-0-0	2-0-0	None
Platform-Curated	1-1-2	1-1-2	4-0-2	2-0-2	None
Aggregated Curated Feeds	None	None	None	None	2-0-0