01

Translate policy into explicit rules

Business language such as 'recent customers' or 'do not over-survey' is too ambiguous for reliable configuration. Convert each statement into testable conditions, precedence, and an owner.

  • Eligibility source and event
  • Exclusion precedence
  • Contact frequency window
  • Journey and channel priority
  • Consent and suppression treatment
02

Test boundary conditions

Most failures happen at boundaries: missing timestamps, duplicate events, late files, timezone changes, reopened cases, or contacts qualifying for more than one journey.

  • Exactly-on-boundary timestamps
  • Duplicate and corrected records
  • Missing or malformed attributes
  • Multi-journey qualification
  • Late-arriving data
03

Separate send logic from survey logic

Invitation eligibility, message delivery, link behavior, survey branching, and response handling are different control points. Validate them independently before testing the end-to-end journey.

04

Keep release evidence

A compact test pack should show input conditions, expected decisions, actual results, and sign-off. This reduces repeated debate when rules change later.