Use hard gates first, then weighted preferences. A strong average should not compensate for missing essential footage, unresolved player identities or an intended use that cannot be authorised.
Proposed hard gates
Each gate should be marked pass, fail or unresolved:
- Priority competition-season evidence meets the agreed minimum.
- Critical identity errors are resolved.
- Intended processing, storage and sharing are authorised.
- Required users can perform the operational workflow.
- Data handling and security requirements are accepted.
- Implementation can meet the actual recruitment deadline.
An unresolved gate blocks an unconditional purchase recommendation. It is not proof that the supplier cannot pass; specify the evidence needed and the person responsible for obtaining it.
Proposed weighted scorecard
| Criterion | Illustrative weight | Evidence to score |
|---|---|---|
| Incremental decision value | 30% | Supported gaps closed beyond the baseline |
| Coverage reliability | 20% | Usable matches, histories, fields and delivery behaviour |
| Accuracy and provenance | 15% | Verified claims, definitions, identities and source lineage |
| Usability and adoption | 15% | Task completion and total user effort |
| Interoperability and portability | 10% | Joinability, reproducibility and exit tests |
| Fully loaded cost and burden | 10% | Equivalent-scope cash and capacity comparison |
| Total | 100% |
Agree weights before receiving results. Suggested anchors are 0: tested and unacceptable; 1: materially deficient; 2: adequate; 3: strong; 4: exceeds the agreed requirement. Keep “not tested” outside the numeric scale and do not silently redistribute its weight.
Use the score to organise discussion, not to manufacture a universal winner. Have the scouting champion, technical evaluator and budget owner sign off on different parts of the evidence.