The Institutional UX Rubric

Companion to Codex 6. Codex 6 states the doctrine; this document scores a surface against it. A surface that has not been scored has not been reviewed.

Governing Codices: 6 (experience), 1 (Articles V and VI), 0 Chapter 11 (surfaces) and Chapter 12 (dashboards).

When it applies

The score is written into the Registry entry's proposal record. A surface may not be marked built without a completed review form (TEMPLATE-review.md).

The eight dimensions

Each dimension is scored 0–3 with written evidence. Evidence means a specific observation about this surface — a count, a quotation, a named state — not an assertion that the dimension is satisfied.

1. Primary action clarity

Is there exactly one primary action, and is it obvious within three seconds?

• 3 — one primary action, unmistakable, correctly named as the outcome it produces.
• 2 — one primary action, discoverable without hunting.
• 1 — two candidate actions of equal weight, or a primary action named vaguely ("Continue", "Submit").
• 0 — no discernible next step, or competing primary actions.

Evidence: name the primary action verbatim and every action of comparable visual weight.

2. Progressive disclosure

Is depth reachable, and hidden until reached for?

• 3 — a clear surface layer, a deliberate second layer, and precision available underneath without leaving.
• 2 — depth exists behind a labelled affordance.
• 1 — everything present at once, or depth hidden where nobody will find it.
• 0 — complexity imposed on arrival, or advanced controls that cannot be avoided.

Evidence: list the layers and what each reveals.

3. Cognitive load

Measured against explicit budgets, per view:

• 3 — within every budget, and the view could not be simplified without losing meaning.
• 2 — within every budget.
• 1 — one budget exceeded, with a named remediation.
• 0 — two or more budgets exceeded, or any unexplained figure or duplicated request.

Evidence: the counts.

4. Mastery-first guidance

Does the surface confer capability, or reward activity?

• 3 — the person leaves able to do something they could not do before, and the surface says what that is.
• 2 — guidance is structural: sequence, defaults, and grouping carry the person forward.
• 1 — guidance is mostly instructional copy compensating for structure.
• 0 — any points, streaks, levels, badges of participation, or reward for mere activity.

Evidence: the capability conferred, and the mechanism that confers it.

5. Orientation

Does the person always know where they are, what just happened, and what is next?

• 3 — location, history, and next step are all answerable without scrolling or guessing.
• 2 — location and next step are clear; results of the last action are confirmed.
• 1 — one of the three is missing or ambiguous.
• 0 — destinations are ambiguous, state is unlabelled, or an action completes silently.

Evidence: quote the location indicator, the confirmation, and the next-step affordance.

6. Register and restraint

Codex 6 typography, tone, spacing, and semantic tokens; institutional voice.

• 3 — indistinguishable in register from the institution's best existing surface.
• 2 — consistent with the design system; reuses existing primitives.
• 1 — a bolted-on feel: novel spacing, one-off components, or a marketing sentence.
• 0 — hardcoded colour utilities, a default sans used as identity, startup or AI aesthetics, loud marketing voice, or any prohibited pattern from §7 below.

Evidence: the primitives reused, and confirmation that no colour value is hardcoded.

7. Failure and recovery

• 3 — every error states cause, consequence, and remedy in plain language; every reversible action is reversible; every irreversible one says so before it happens.
• 2 — all error, empty, loading, partial, and unauthorised states are designed and specified.
• 1 — one state unspecified.
• 0 — a raw error surface, a dead end with no remedy, or an irreversible action without prior warning.

Evidence: list the five states and the wording of the primary error.

8. Access and inclusion

• 3 — keyboard-complete with visible focus, correct landmarks and heading order, alt text, contrast verified, reduced motion respected, and usable on a poor connection.
• 2 — all of the above except performance under poor conditions.
• 1 — one accessibility defect with a named remediation.
• 0 — keyboard-inoperable, focus invisible, contrast failing, motion unavoidable, or images carrying meaning without alternatives.

Evidence: how it was checked, not that it was checked.

Scoring rules

1. 1Any dimension at 0 blocks the surface, whatever the total.
2. 2Any dimension at 1 ships only with a named remediation owner and a date recorded in the review form.
3. 3Maximum 24. A surface at 16 or below is redesigned rather than patched.
4. 4Dimensions are never averaged away: the score is reported as eight numbers, then the total.
5. 5Two reviewers scoring the same surface should land within one point per dimension. Persistent divergence is a defect in this rubric and is fixed by amending it (Codex 9), not by splitting the difference.

The final gate

After scoring, answer in writing, in one paragraph:

Did this make the person's life better?

If the honest answer is "it increased engagement", the surface fails regardless of its score and is removed. This gate cannot be passed by a high total.

Prohibited at the interaction level

Restating Codex 1, Article VI as an interface rule. Any of these present is an automatic 0 on dimension 6:

manufactured urgency · countdowns · artificial scarcity · loss-framed nudges · notification pressure · infinite feeds · vanity counts · streak guilt · consent bundled into unrelated actions · obstructed cancellation or export · rankings of people against each other · any mechanic designed to make leaving feel costly.

Files

• TEMPLATE-page.md — written before a surface is designed.
• TEMPLATE-interaction.md — written before a flow or form is built.
• TEMPLATE-review.md — completed before a surface is marked built.