Rule-based AMC lookalike audiences via API: seed, reach, refresh

How it works

The lookalike API extends the API for rule-based audiences. It uses the same query language and datasets; you define the seed group with an audience query. AMC only creates the lookalike audience in the target product, no additional rule-based duplicate.

An existing rule-based audience cannot be used as the seed. Copy its SQL query into the request that creates the lookalike audience instead.

Check the seed size

The seed audience must contain at least 500 and at most 500,000 user_id values, otherwise creation fails. You can count the approximate size beforehand with a query:

SELECT
  COUNT(DISTINCT user_id)
FROM
  amazon_attributed_events_by_conversion_time
WHERE
  conversion_event_subtype = 'shoppingCart'

Choose the reach

The required parameter lookalikeAudienceExpectedReach sets the target size: similar to the seed group or broader. The ranges differ by region:

Refresh and runtime

• refreshRateDays determines how often the SQL query is re-run and the seed audience completely overwritten. Valid values are 7 to 21.
• With refreshRateDays = 0 the audience is created in Amazon DSP but deactivated after 30 days; deactivated audiences have no members.
• Synchronisation with the advertiser account takes 2 to 24 hours, including the time to create the seed.
• Typical reasons for FAILED: seed under 500 users, a table not supported for audiences in the query, or missing required parameters.

Frequently asked questions