Use five evidence checks per criterion, worth one raw point each. The next section defines all 40 checks.
For each check, record:
- P — Pass: the agreed requirement is evidenced; one point.
- F — Fail: evidence shows it is not met; zero points.
- U — Unknown: evidence is missing, stale or insufficient; zero points.
A supplier statement is not an observed test result. A signed future service commitment can support a contractual check, but cannot replace an operational check. If a check has several requirements, all must pass. Replace genuinely inapplicable checks with equivalent buyer-approved requirements before evaluation; do not simply remove them and inflate the score.
Weighted points for a criterion = weight × passed checks ÷ 5.
Total score = sum of weighted points across all eight criteria.
Also calculate:
Evidence completeness = sum of weight × resolved checks ÷ 5, where resolved checks are P or F.
This second number distinguishes a well-investigated weak proposal from an attractive but largely unverified one. Neither a high completeness figure nor a high score overrides a failed gate.
What decision should follow the score?
These are proposed approval rules, not industry standards:
| Result | Proposed decision |
|---|---|
| Any gate on Hold or Fail | Not eligible for production approval. A diagnostic score may be retained internally, clearly labelled provisional. |
| 80–100, evidence completeness at least 90%, every criterion at least 3/5, and mandatory checks passed | Eligible for approval within the assessed scope, subject to budget and accountable sign-off |
| 65–79.9, or a criterion below 3/5 | Remediate, narrow the scope or renegotiate; then rescore |
| Below 65 | Do not select for the current brief without substantial redesign |
Mandatory operational checks: A2, B1 and B2 below must pass regardless of the total. Every published game also needs a passed rights assessment. A high average must not conceal a broken essential title or an unacceptable required environment.