Use stage-specific decisions rather than a single “approved supplier” label.
| Stage | Proposed decision rule |
|---|---|
| Below 60 | Do not approve a live pilot. Resolve evidence gaps, narrow the scope or select a simpler alternative. |
| 60–79 | Eligible for a bounded pilot only if every applicable gate passes, criteria 2–5 each score at least 3, and no other criterion is below 2. |
| 80–100 before a pilot | Strong procurement position, not evidence of your operational return. Run the bounded pilot. |
| 80–100 after a pilot | Consider expansion only if all gates pass; criteria 1–6 each score at least 4; every other criterion scores at least 3; and the buyer accepts the measured economics, residual risks and downside exposure. |
| Any score with a failed gate | No launch or expansion in the affected scope until remediation and retesting. |
Why gates matter: a hypothetical option scoring 5 everywhere except 1 for safety would still achieve 87.2/100. It must not launch. Another option scoring 4 in every category achieves 80/100 and may be the better choice—provided its live evidence and all other conditions support expansion.
Complete the decision record:
- Selected workflow and permitted authority:
- Option compared with:
- Weighted score and assessment stage:
- Gate results and unresolved unknowns:
- Measured or modelled economics, clearly distinguished:
- Maximum pilot spend and downside allowance:
- Exceptions owner and fallback route:
- Required changes, owners and deadlines:
- Decision: configure native software / use human or blended cover / pilot / extend pilot / expand / stop.
- Reassessment trigger: material configuration, supplier, integration, pricing or legal change; incident; or renewal.
Bottom line: buy a controlled workflow with demonstrable net benefit. If the proposal cannot show what happens when a caller is misunderstood, a booking fails, a message should be suppressed or the supplier disappears, it is not ready for your customers.
PUT THE RESEARCH TO WORK