Access MMM feeds

If you have completed the Marketing Mix Modeling (MMM) provider initial setup, you can now access the MMM feeds with advertiser data to include in your MMM model. After an advertiser selects you as their MMM provider and configures a feed in the platform, The Trade Desk sends an email to the contact alias you provided during initial setup. You can then retrieve the advertiser MMM feed files in your designated provider folder in an Amazon Simple Storage Service (S3) bucket owned by The Trade Desk.

IMPORTANT: To access your assigned folders, you must create an AWS access ID and secret key. For details, see Amazon S3 Access.

Here's what you need to know about the files you receive:

• All files are GZIP-compressed to reduce the overall delivery size.
• All files are tab-delimited and contain raw, unprocessed data that may include duplicate information.
• The time zone for all dates and times in file names and fields is UTC unless otherwise specified.

Here's what to keep in mind when you schedule your ingestion:

• Files are updated daily. If the advertiser has no activity for a given day, no file is generated for that day.
• Files are deleted 90 days after initial delivery.

TIP: Since The Trade Desk delivers new files daily, be sure to set up a process that ingests them once per day.

The following sections provide information about the email notification and examples of the Amazon S3 bucket URI path format that you can use to locate your files.

Feed notification email

After an advertiser creates a feed, The Trade Desk sends an email with the subject line New Marketing Mix Modeling Feed Created to the contact alias you provided during your initial setup. Each notification email contains the identifiers for one feed.

The following table lists the details included in the notification email.

MMM feed file locations in S3

MMM feed files are stored in your designated provider folder in an AWS S3 bucket owned by The Trade Desk. Here's what you need to know about how the files, folders, and logs are organized in the AWS S3 bucket:

• Files are organized into sub-folders based on the partner and advertiser levels, ensuring structured access to the data.
• Sub-folder names follow a consistent URI path format, where placeholder values (indicated by square brackets []) represent unique identifiers such as ModelerName, PartnerId, AdvertiserId, and Date.
• Each log file is stored in the corresponding date folder.

TIP: You can also find the PartnerId and AdvertiserId values in the email notification that was sent when an advertiser configured the feed.

The following is an example of a log file path URI template, followed by a complete example demonstrating the full folder structure in Amazon S3:

s3://thetradedesk-useast-partner-datafeed/exposure/[ModelerName]/[PartnerId]/advertisers/[AdvertiserId]/mmm_aggregated_insights/sv=[SchemaVersion]/date=[Date]

s3://thetradedesk-useast-partner-datafeed/exposure/MMMCompany/abc123/advertisers/xyz123/mmm_aggregated_insights/sv=1-1-1-1/date=2026-04-01

The following sections highlight the difference in base URIs at the partner and advertiser levels. The key distinction between the two S3 paths is data granularity: the partner-level path stores aggregated insights for the entire partner without differentiating between advertisers, while the advertiser-level path includes additional sub-folders for advertiser-specific data, enabling more granular access and organization.

Partner-level URI format

Here's an example of the base URI format at the partner level:

s3://thetradedesk-useast-partner-datafeed/exposure/[ModelerName]/[PartnerId]/mmm_aggregated_insights/../

For example:

s3://thetradedesk-useast-partner-datafeed/exposure/MMMCompany/abc123/mmm_aggregated_insights/../

Advertiser-level URI format

Here's an example of the base URI format at the advertiser level:

s3://thetradedesk-useast-partner-datafeed/exposure/[ModelerName]/[PartnerId]/advertisers/[AdvertiserId]/mmm_aggregated_insights/../

For example:

s3://thetradedesk-useast-partner-datafeed/exposure/MMMCompany/abc123/advertisers/xyz123/mmm_aggregated_insights/../

Folder path variables

The following table lists the variables used in your provider folder paths in S3.

MMM feed file naming convention

Each log file follows a consistent naming structure, ensuring easy identification and organization. The log file names include various parameters such as the advertiser ID, feed ID, date and time details, and a unique hashed value. All files for MMM feeds have the .log.gz extension, indicating they are GZIP-compressed.

Here's a template of a log file name, along with a complete example to help you understand the format:

aggregatedperformance_[AdvertiserId]_v[FeedVersion]_[StartDateTime]_[ProcessedDateTime]_[FeedId]_[UniqueHashedValue].log.gz

For example:

aggregatedperformance_xyz123_v1_2025-01-01T000000_2026-04-01T080100_1234_82ce3baa0ca8afbe57069b785a7ea2af.log.gz

TIP: The StartDateTime value can also be found in the email notification that was sent when an advertiser configured the feed.

The following table lists the variables used in your MMM feed file names.

FAQs

The following are most commonly asked questions about the MMM feed files and how to access them.

How often are the files updated?

Files for MMM feeds are updated daily.

How do I update my AWS keys?

To update your AWS keys, you must revoke your current keys, generate new keys, and then replace the keys in your integration. For details, see Amazon S3 Access

How long are the files stored in the AWS server?

Files are stored in the AWS server for 90 days after initial file delivery and then automatically deleted. The deletion date is based on when the file is delivered or updated, not on the data it contains.

How do I map The Trade Desk entities to my model?

If needed, you can request a mapping file to accurately map The Trade Desk entities such as campaigns, ad groups, and creatives to your model. To request a mapping file, work directly with the advertiser to contact their account team at The Trade Desk.

The following table is an example of a mapping file that associates two campaign names and IDs from the platform with a single brand and two marketing initiatives as user-friendly names.