‏‎Ad Targeting dataset‎‏

Changelog

The changelog lists documentation or dataset changes per release. If there is no heading for a monthly release, that means there were no changes to report.

Page names can change!

Page name changes can affect the contents of the include and exclude columns because targeting for categories related to demographics (employment, education, and so on) is based on user-defined values that point to pages.

We suggest you review the names impacting your research on a regular basis to catch any changes and adjust accordingly. You can see specific changes since the last collection of data in the delta table.

2024 -11-11 Update

After January 7, 2025, access to Ad Targeting in Researcher Platform will require Amazon WorkSpaces Secure Browser. See WorkSpaces Secure Browser in the Researcher Platform documentation for instructions. You will not lose any data in this transition. With WorkSpaces Secure Browser, you do not need a VPN to access Researcher Platform. The Get started page provides the URL so you can try it.

Until January 7, 2025, you can access Ad Targeting through a Virtual Private Network (VPN).

2022-12-01 monthly data update

As of this update, the following issues have been addressed:

  • We found that we were rendering the names for certain ad targeting options incorrectly, adding a prefix of “TIGER_” (which has meaning internally), and making them difficult to read. We have found and fixed the root cause, and the names going forward will be rendered correctly. This affected approximately 466,000 ads.

  • We discovered that when an advertiser used the same name in multiple categories in their targeting criteria, we were only capturing the last one encountered in the include or exclude columns. For example, "Sephora" might have been used as both an employer and an interest. Going forward, we are capturing all instances of the name by including the category in parentheses after the name, making each instance unique and unable to overwrite a previous instance. This affected approximately 508,000 ads.

2022-11-01 monthly data update

As of October 21, 2022, previously UTF-8 encoded text in the Ad Library data table is now decoded and displayed as it was originally written. This affects the following table columns:

  • ad_creative_bodies
  • ad_creative_link_captions
  • ad_creative_link_descriptions
  • ad_creative_link_titles

2022-10-01 monthly data update

As of this update, all data is now alphabetically ordered in the location fields (include_location and exclude_location). The one-time mass reordering this enhancement required is reflected in the delta table as many more changes than usual.

2022-09-01 monthly data update

The following changes have been made to the table schema in order to more clearly and accurately represent included and excluded targeting criteria:

  • include_any_of, include_any_of_interests, include_all_of, include_all_of_interests, and exclude_interests columns have been removed.

  • A new column, include, has been added. This new column contains a list of all of the included groups of targeting criteria, separated by type (interest, education level, demographic categories, behavior categories, and so on). The data type for the include column is list<map<string, string>>.

  • The exclude column now contains all the excluded targeting criteria, including excluded interests. The data type for the exclude column has changed from string to map<string, string>.

  • The include_location and exclude_location column data types have changed from string to map<string, map<string, mixed>>.

See Table Schema for updated column information.

Effect on the delta table

Data changes resulting from the changes to the table schema are not reflected in the delta table for the period between the 2022-08-01 update and the 2022-09-01 update. All other data changes are reflected in the 2022-09-01 delta table.

2022-08-01 monthly data update

The following change has been made to the documentation:

The table schema column descriptions have been updated to correct a misrepresentation of data in the include_any_of, include_any_of_interests, include_all_of, and include_all_of_interests columns. Until we correct this problem in the structure of the dataset, be aware that to get a complete picture of how an ad was targeted:

  • Combine the include_any_of and include_all_of columns

  • Combine the include_any_of_interests and include_all_of_interests columns

See Table schema for updated column descriptions.

2022-07-01 monthly data update

The following changes affect the data:

  • Location data was deleted from some target pages, making it impossible for the dataset to include the specific location. This affected less than 200 ads in the entire dataset. Where location data was unavailable, we replaced it with <Unknown Country>: <unknown place>.

  • The include_connection, include_friend_connection, and exclude_connection fields have been updated for many ads that were using connections, but were not previously marked as using connections. You can see the specific changes in the delta table.

  • In approximately 7% of ads, some or all location data was inadvertently omitted. We fixed the problem and updated the dataset. You can view specific updates in the delta table.

The following new columns have been added to the delta table:

  • status—Type of change being reported.

  • column_name—Name of column affected by the change. Can also be empty.

  • previous_value—Value before the change.

  • current_value—Value after the change.