The critique is the score

A PRD tool has two jobs: help you write the doc, and tell you honestly whether the doc is any good. This release sharpens the second one — and closes the gap between the two.

One score that means something

The critique no longer reports a vague 1–5. It now gives a real 0–100 — the same number you see broken down by dimension, computed deterministically so it's reproducible and defensible. That score is your PRD's score of record: it shows on the critique and across your PRD list, with the readiness band (Excellent · Good · Needs Work · Not Ready) carried in the editor header so the grade reads at a glance without repeating the number.

The old "Quality" check is now Draft completeness — a quick pre-flight on how complete the draft is, not a competing grade. Edit a PRD after critiquing it and the score is flagged stale so you know to re-run. A polished draft and a sound PRD are different things; now the product says which is which.

Hover any dimension to see its weight and what it covers — the rubric is published, not a black box.

PRDs that generate with rigor built in

The critique used to demand things the generator never produced — then flag their absence. No more. New PRDs now come with:

• Structured user stories — a specific persona, a clear want and benefit, a MoSCoW priority, and their own acceptance criteria.
• An Edge Cases section — failure modes with triggers, expected handling, and severity, in the document itself.
• Deeper flows — every flow area carries a precondition and at least two Gherkin scenarios (a happy path and an error path).
• Metrics that connect — each analytics event maps to the success metric it measures, and every metric gets a Metric Monitoring plan: owner, alert threshold, rollback trigger, review cadence.
• Structured non-functional requirements — performance, reliability, security, and now accessibility (WCAG) and localisation too, each with a target and owner.
• Personas, typed dependencies, owned risks, and real launch phasing — rollout %, feature flag, rollback trigger, and a success gate per phase.
• A References section — sourced facts collect into one place.

A readiness check that isn't fooled

The coach's readiness gate now measures the substance of your answers, not how much you typed. A two-word persona or a one-line problem stays amber; a measurable outcome with a real number moves the needle. One thoughtful paragraph beats ten throwaway replies.

Compliance you can act on

When a PRD trips a compliance signal — personal data, eKYC, payments, health — the critique now raises a concrete, anchored finding naming the applicable regulation (UU PDP, OJK eKYC, GDPR, PCI-DSS, HIPAA) and asking for an owner, right in your editor.