QA Test Scripts must have Signoff Approval from both Design and NL, before being conducted
- This ensures that QA has a thorough understanding of the design, before beginning testing
  - Each channel (e.g., Voice/IVR, Chatbot, etc.) must be tested in that channel — do not attempt to “simulate”.
- Shortens necessary testing cycles, produces more reliable test results
Have a formal, defined, standardized, published process for logging, reviewing, and closing design defects
- Include standards of what does and doesn’t constitute a “defect”
  - E.g, “I don’t think it should work that way…” is not a defect; “The design contains an infinite loop…” is a defect.
  - Some areas may need to be “stubbed out” if they link to unfinished design sections
Triage and rank defects
- Classify as “Defect”, “Change Request (which carries an additional charge to the client), “Future Feature”
Assign each defect a Severity and Priority
- Unless something in the design actually causes imminent risk of loss of life or limb, there is no such thing as a “Conversation Design emergency”!
- “Severity” and “Priority” are not interchangeable
  - S1: Prevents correct operation < — > S4: Cosmetic / Aesthetic
  - P1: Must fix in next iteration < — > P4: Defer
  - Assign S & P levels during Design or NL reviews
Exit Criteria for this phase:
- Successful UAT
- Meets all UAT criteria, with no scope creep
- UAT signoff from client