Use the interviews to generate approval gates, not a collection of favourable quotations. The following is the desk’s proposed decision checklist; it is not a published industry standard or legal clearance.
| Decision gate | Evidence required | Recommended response if missing |
|---|---|---|
| Workflow is bounded | Approved task, exclusions and permissions | Prepare first |
| Outcomes are verifiable | Reconciliation to operational records | Do not claim resolution or ROI |
| Economics are comparable | Fully loaded cost and equivalent baseline | Do not approve a savings case |
| Handoff is operational | Tested failure branches and named owner | Do not release dependent workflows |
| System writes are controlled | Duplicate, timeout and incorrect-action tests | Restrict write access |
| Calling and data use are approved | Deployment-specific legal and security review | Do not launch affected processing |
| Quality is owned | Monitoring, rollback and incident responsibility | Do not move beyond a limited pilot |
| Exit is workable | Tested exports and transition obligations | Resolve before long-term commitment |
For a first interview round, prioritise Questions 1, 2, 3, 7, 8 and 13, plus the relevant sector and legal questions. Ask for artefacts before the interview so the conversation can examine evidence rather than rehearse the sales presentation.
How should expert quotations be published?
- Obtain an approved quotation, verified role and interview date.
- Disclose supplier, advisory, investment and referral relationships relevant to the answer.
- Separate observed deployment results from opinions and forecasts.
- Publish denominators, measurement periods and exclusions alongside performance claims.
- Retain disagreement: an operator, academic, supplier and lawyer may legitimately address different parts of the decision.
- Never turn a narrow legal or technical comment into a blanket endorsement of a supplier.
The desk’s proposed Your-Calls Vendor Trial Builder, True Cost per Completed Call, Dental Booking Integrity Test Pack and Human Handoff Failure Audit are suitable future destinations for the evidence generated here. They should not be presented as available downloads until built and tested.