Today, Segment Limits allow users to constrain a segment by specifying a record count, selecting an order-by dimension, and applying ascending or descending sort logic. This is useful when users want the “top” or “bottom” records by a given field, but it does not support cases where users want a limited set of records that is more representative of the broader segment population.

I would like the platform to support a randomized selection option within Segment Limits, so that users can limit a segment to a defined number of records while still drawing those records from across the broader universe, rather than only from one edge of a sorted dimension.

Example use case
A segment contains 1,000,000 users. I want to limit the segment to 1,000 users. If I choose user_created_date as the relevant dimension, I do not want only the oldest or newest 1,000 users based on ascending or descending order. Instead, I want the system to return a randomized set of 1,000 users drawn from across the segment population, so that the resulting audience includes a mix of user_created_date values.

Why this is valuable
This would help users create smaller, more balanced samples of a segment for testing, activation, QA, audience analysis, and operational workflows. It would reduce bias introduced by purely ascending or descending selection and make limited segments more representative of the full eligible universe.

Suggested functional behavior
 • Add a new Segment Limits selection option such as:
 • Sort Type: Ascending / Descending / Randomized
 • Allow randomization to occur within the qualified segment population before the limit is applied
 • Preserve the selected record limit count
 • Optionally allow the randomization to be:
 • Dynamic, where each refresh can produce a different random sample
 • Stable/seeded, where the same limited sample can be reproduced until the segment logic changes

Acceptance criteria
 • A user can configure a segment limit using a defined record count and select a randomized selection method
 • The resulting limited segment returns records distributed across the eligible population rather than only from the top or bottom of the chosen sort field
 • The output respects the requested record limit exactly
 • System behavior is clearly defined for whether the random sample refreshes each run or remains fixed

Potential UI approach
Under Segment Limits:
 • Record Limit: 1,000
 • Dimension: user_created_date
 • Order Type:
 • Ascending
 • Descending
 • Randomized