Reference — generated from the toolkit
Source of truth: toolkit/templates/pfd-rubric.md. Edit it there; this page is regenerated on build.
PFD quality rubric — {product / feature name}
- PFD: {link, with version} Scored by: … Date: {YYYY-MM-DD}
Scoring descriptors
| Dimension | 1 — not ready | 3 — workable | 5 — exemplar |
|---|---|---|---|
| Scope clarity | Scope implied by a feature list; out-of-scope missing or "TBD". | In-scope explicit; out-of-scope stated but doesn't name the tempting adjacencies. | Both lists explicit; out-of-scope names the adjacencies an agent would otherwise wander into (e.g. "MTAs in scope; cancellations and renewals out"), personas and products covered are unambiguous. |
| Rule exemplification | Rules stated as prose intentions; few or no worked examples. | Every rule has Given/When/Then ACs, but examples are patchy — some rules lack a negative or boundary case. | Every rule carries a positive, a negative and a boundary worked example with concrete values (dates, amounts, rounding) per the acceptance-criteria standard; ambiguity lint is clean. |
| Glossary compliance | No glossary link; terms used loosely (insurer/carrier interchanged, MTA vs endorsement blurred). | Glossary linked; a handful of orphan terms or stale entries remain. | Every domain term in the PFD resolves to the glossary — zero orphans — and the glossary version is pinned. |
| NFR & compliance completeness | Missing, or aspirational ("fast", "secure", "compliant"). | Applicable rows selected from the NFR catalogue but some lack concrete targets. | Every applicable row selected with a testable target (p95 latency, RPO/RTO, WCAG 2.2 AA, UK GDPR residency, FCA Consumer Duty posture, EU AI Act classification where AI is used). |
| Commercial section quality | Missing or placeholder text. | Segments and positioning drafted; pricing/packaging hypothesis vague or not yet Commercial-signed. | Segments, pricing/packaging hypotheses and positioning stated with evidence traced to the signal register; Commercial sign-off recorded. |
| Open-question hygiene | Questions untriaged, or hiding in prose ("we may need to…"). | Triaged blocking vs non-blocking, but blocking questions remain open or lack owners. | Every question triaged with an owner and a date; zero blocking questions remain. |
Scores
| Dimension | Score (1–5) | Evidence (section / line in the PFD) |
|---|---|---|
| Scope clarity | … | … |
| Rule exemplification | … | … |
| Glossary compliance | … | … |
| NFR & compliance completeness | … | … |
| Commercial section quality | … | … |
| Open-question hygiene | … | … |
| Average | … |
Gate threshold
- No dimension below 3, and average ≥ 4 → the PFD may be put to the Definition of Ready gate.
- Any dimension at 1 → back to Definition; do not request the gate.
- A rubric score is a form check. It does not replace SME sign-off on domain correctness — a beautifully exemplified wrong rule scores 5 here and still fails the gate.
Gold standard
The reference point for "what a 5 looks like" will be StudyBuddy's PFD once written — the first fully worked exemplar. Link it here when it lands.