What should a generated communication preference task pack contain?

Ask for these components when building an evaluation around customer communication preferences.

- A preference record separated by communication purpose. - A channel register with verified contact destinations. - Queued messages linked to their intended recipients. - Policy rules for required service communications. - An alternative route when a channel becomes unavailable. - Expected queue changes for each customer instruction.

A retail customer withdraws permission for promotional email while retaining email receipts. The dataset should make those purposes distinguishable. A single `email_enabled` flag would force the agent to choose between outcomes that the customer's instruction does not require.

Generated records could include a promotion awaiting dispatch, a purchase receipt and a service update. Their purposes determine how the new preference applies. The supplied policy, rather than the evaluator's unstated assumptions, should explain any communication that must continue through another permitted route.

Evaluation teams can inspect whether the task tests instruction interpretation, queue handling and purpose-specific rules. They should also check that the expected answer does not depend on legal knowledge absent from the pack. Any sector-specific obligations used in scoring need an explicit, reviewable representation in the task materials.