01 · OVERVIEW

A principal design critic you can run on every screen

Point it at a screenshot, a Figma frame, a live URL or a prototype. It measures the real thing, holds it to one rubric, and hands back a scored, pinned, printable audit.

I wanted a consistent way to raise the bar on the work we make with AI assisting us — grounded in design and psychology principles, with industry references that justify every assessment and every change it asks for.

How the skill works: you give it an image, Figma frame, live URL or prototype; it runs five phases — intake, inspect, evaluate, prioritize, report — and produces a six-page designed report — FIG. 1 — THE SKILL, END TO END

02 · WHAT IT ADDS

Three things it adds to every critique

Industry references, design psychology, and a test for what’s obvious — so our work is held to the principles we care about, not a generic checklist.

A
Backs the fixes with industry references Via the Mobbin MCP, each top fix carries a real screen from a strong product solving the same problem.
B
Grounds the why in design psychology Findings cite principles from Growth.Design’s library of 106 — Anchoring, Hick’s Law, Cognitive Load — wherever the fit is strong.
C
Makes the next step obvious On dense observability screens, every key surface should make clear what to take away and what to do next — shrinking time to correct decision.

03 · THE BAR

Backs every finding with evidence

Each finding runs Issue → Why it matters → Fix, pinned to the screen and backed by a measurement: the actual hex, the measured contrast ratio.

AT RISK0–54

NEEDS WORK55–74

SOLID75–84

STRONG85–100 · RARE

FIG. 2 — THE SCORING SCALE, DRAWN TO SIZE

Audit page: a compare-table toggle measured at 3.97 to 1 contrast, #6A6F74 on #E5E4D8, with the fix to darken it to #595E63 — FIG. 3 — NEW RELIC PRICING: #6A6F74 ON #E5E4D8, MEASURED 3.97:1, WITH THE FIX

04 · THESIS FIT

Holds craft to the metric that matters

For observability, that metric is time to correct decision. Anything else is held to its own job — for the pricing page, a confident edition choice.

THE GOVERNING METRIC · OBSERVABILITY

Time to correct decision.

Credits screens that end in a decision. Read as a direction, never a fake score.

Executive verdict: 74 out of 100, Needs work — UX 75, UI 76, Product 70 — with what's working and the principles at play — FIG. 4 — THE VERDICT: 74 / NEEDS WORK

Method page: because a pricing page is not an observability product, the time-to-correct-decision thesis was not applied — FIG. 5 — METHOD: THE OBSERVABILITY THESIS, DELIBERATELY NOT APPLIED

05 · THE AUDIT

Delivers a report to read, not a fix to accept

A PDF rather than an HTML page, on purpose: something everyone can open, look at and pass around. The critic can misread a page, miss the goal, or flag something the team did deliberately — so people check its understanding before anything changes.

We wanted it to be something someone spends time thinking about, reading and looking at — not something you see on screen and just hit go.

A
Pins every finding to the screen it’s about A cropped plate of the real capture, a numbered pin, and the fix alongside.
B
Names what’s working, too Every verdict opens with the three or four things the design gets right.
C
Made to be passed around A PDF anyone can open, one section per page — easy to share, print and talk through as a team.

Report cover: Design Audit Vol. 02, New Relic Pricing, reading 74 out of 100, Needs work — FIG. 6 — COVER · NEWRELIC.COM/PRICING

Finding 1: the cookie consent modal lands over the hero on mobile, pinned on the capture, with the fix — FIG. 7 — FINDING 1: THE COOKIE WALL, PINNED

Findings 3 and 4: transparent pricing that hides the prices, citing Anchoring and Hick's Law — FIG. 8 — ANCHORING & HICK’S LAW, NAMED WHERE THEY FIT

Appendix: the full pricing page capture with all six findings pinned — FIG. 9 — APPENDIX: EVERY FINDING ON THE FULL CAPTURE

The takeaways: six moves ranked, the top ones each paired with a Mobbin benchmark from Bloom and Dribbble — FIG. 10 — THE TAKEAWAYS: SIX MOVES, RANKED, EACH BACKED BY A MOBBIN BENCHMARK WHERE ONE FITS

06 · OUTCOME

Still in refinement. The goal is for it to be part of every designer’s workflow — keeping the designer as the human in the loop, and in the driver’s seat.

50+designers it’s been shared with so far
Humanin the loop, and in the driver’s seat — the critic advises, the designer decides
AI-firsta key piece of our design transformation