The following example is entirely fictional. It demonstrates the calculation and gate logic, not any supplier’s performance.
| Criterion | Weight | Plan A rating | Plan A points | Plan B rating | Plan B points |
|---|---|---|---|---|---|
| Decision contribution | 25 | 4 | 20.0 | 5 | 25.0 |
| Coverage | 18 | 4 | 14.4 | 5 | 18.0 |
| Accuracy and provenance | 12 | 5 | 12.0 | 5 | 12.0 |
| Contextual validity | 8 | 4 | 6.4 | 4 | 6.4 |
| Workflow | 8 | 4 | 6.4 | 5 | 8.0 |
| Integration | 8 | 4 | 6.4 | 4 | 6.4 |
| Licence fit | 7 | 4 | 5.6 | 0 | 0.0 |
| Security | 4 | 4 | 3.2 | 4 | 3.2 |
| Service | 3 | 4 | 2.4 | 4 | 2.4 |
| Cost | 7 | 4 | 5.6 | 4 | 5.6 |
| Total | 100 | 82.4 | 87.0 | ||
| Gate outcome | Pass | Fail |
Assume Plan B expressly prohibits a required operational use, even though its evaluation was authorised. Plan B is not approvable in that form despite its higher score. Plan A is the eligible option, subject to budget and implementation approval. If Plan B’s agreement is amended, reassess its licence score and gate rather than preserving the earlier verdict automatically.
For close eligible options, run a sensitivity check: vary disputed ratings by one point and use plausible alternative weights agreed by the buying group. If the preferred option changes, present the decision as sensitive and identify the evidence most likely to resolve it. Do not manufacture certainty from decimal places.