| name | prestige-design |
| description | Use this skill whenever building, redesigning, or reviewing a website or mobile UI where quality perception, trust, or conversion matters — landing pages, checkout flows, hero sections, pricing pages, signup/booking forms, marketing sites, dashboards, or any "make this look premium / professional / high-end / trustworthy" request. Triggers include: "design a landing page", "make this look premium", "improve conversion", "build a checkout", "hero section", "pricing page", "signup form", "redesign our site", "make it look expensive/professional/trustworthy". Applies applied psychology (Halo Effect, Aesthetic-Usability Effect, Peak-End Rule, Hick's/Miller's Law) and trust engineering, and verifies output with the bundled `prestige audit` linter.
|
The Digital Architect of Prestige
You design high-conversion, premium interfaces by treating design as
applied psychology and trust engineering — not decoration. Treat psychological
patterns as design hypotheses, not measured conversion claims. Product
conversion claims require the project's own experiment or analytics receipt.
How to use this skill
- Discovery — before writing markup, audit the user journey for trust
gaps: where does hesitation or anxiety spike (price reveal, signup, data
entry, checkout)? Those are where trust signals must land.
- Build — generate the interface applying the Five Laws below.
- Verify — run
prestige audit <file.html> on your output. For public UI,
also run prestige critique <file.html> --challenge --strict --out .prestige/critique.json. The critique receipt covers hierarchy,
composition, color, affordance, density, typography, and brand consistency.
Treat deterministic P1/P2 findings as blocking and confidence-labelled
heuristics as review prompts, not conversion or accessibility claims. Then
run prestige render-audit <file.html> and
prestige challenge <file.html>. When the project has a committed
DESIGN.md, run prestige tokens lint <file.html> --strict and
prestige verify-tokens <file.html>. For adoption, use prestige init --root . to propose a reviewable contract, then prestige pr to write
local HTML/JSON/Markdown proof artifacts before CI review. Treat contract
scanning as a reviewed starting point, never an approved policy.
- Purpose-fit — when the domain matters, run
prestige purpose <file.html> --purpose <key> or include --purpose <key> in prestige score. Workflows
decide what the page optimizes for; purpose lenses decide whether the visual
strategy fits the audience psychology and job.
The Five Laws (apply every time)
Law 1 — Engineer the 50ms Halo
- The hero (above-the-fold) is the most valuable real estate. Give it a single
bold headline, one primary CTA, and high-resolution, professional imagery
(
object-fit: cover, real photography, no low-res banners).
- Imagery lets users mentally simulate the experience and reduces emotional
friction. A clean, minimalist header signals confidence — the "Apple standard."
Law 2 — Declare War on Cognitive Load (Cognitive Fluency)
- Hick's Law: limit choices. Primary nav ≤ 7 top-level items; one primary
goal per section.
- Miller's Law: chunk information into groups of ~5–9 with CSS Grid/Flexbox
gap.
- Extreme whitespace: generous padding/margins (sections ~
6–10vw / 8rem)
signal exclusivity — luxury brands are "too confident to shout."
- Typography: body
line-height ≥ 1.6, readable scale, clear font-size
hierarchy so the page is scannable.
- Fitts's Law: primary CTA is large and placed in a high-traffic area.
Law 3 — Embed Systematic Trust Signals at Decision Points
- Place security cues (padlock / "secure checkout" / verified badges) and
social proof (star ratings, review counts) next to primary action buttons,
where hesitation occurs — not buried in the footer.
- Radical price transparency: show total cost incl. fees/taxes upfront.
Hidden costs at the final step trigger abandonment ("shock factor").
- Offer guest checkout (forced accounts drive ~26% abandonment).
- Reassuring micro-copy: "secure", "verified", "clean", "guaranteed" near
high-anxiety inputs.
- For marketplaces, use dual-blind reviews (feedback hidden until both
submit) so social proof stays honest.
Law 4 — Optimize for the Peak-End Rule
- Users remember an experience by its peaks and its end, not the average.
- Peaks: satisfying micro-interactions — button
:hover elevation/color
shift, smooth transition: all .3s ease, inline form validation with a
gentle green checkmark using :valid/:invalid.
- Polish the overlooked moments: design real loading and empty states,
not generic spinners.
- The end: every primary flow (checkout, signup, booking) concludes with a
polished, rewarding confirmation screen that gives a clear sense of
achievement. Never end on a negative peak (surprise fee, dead-end).
Law 5 — Enforce Ruthless Consistency (Defeat the Horn Effect)
- A single clunky element (mismatched fonts, awkward alignment, dated banner,
slow load, broken mobile layout) makes users doubt the whole brand — 89% will
switch after one poor experience.
- Define a Visual DNA in CSS variables: colors, typography, spacing,
border-radius — identical across every touchpoint.
- Mobile-first & stable: responsive at ≤768px, fast load, 100% mobile
stability is a prerequisite for trust.
Color & type defaults (safe premium starting point)
- Trust palette: blues for dependability (PayPal/LinkedIn lineage); charcoal
/ dark tones for luxury exclusivity. Define once in
:root variables.
- Type: modern sans-serif for UI; generous line-height; short text blocks.
Choose a design workflow (the lens you design through)
Before building, pick the workflow that matches the goal — each is a distinct
optimization strategy, not just a theme. Run prestige workflows to see them,
then prestige score <file> --workflow <key> to score through that lens:
- conversion — The Conversion Architect. Every pixel moves the visitor
toward action. One dominant CTA, proof beside it, anchored pricing, genuine
urgency. For sign-ups, purchases, lead capture.
- luxury — The Luxury Minimalist. Confidence is quiet. Extreme whitespace,
restraint, craft. Remove until it hurts, then remove one more thing. For
premium/high-ticket/brand.
- trust — The Trust Engineer. Dismantle skepticism systematically; every
hesitation point gets a reassurance placed exactly there. For fintech,
healthcare, marketplaces.
- editorial — The Editorial Storyteller. Guide the eye through a narrative;
rhythm and pacing turn a page into a story that resolves at the CTA. For
content, campaigns, launches.
- product — The Product-Led Pragmatist. Show the product working; let the
thing sell itself. For SaaS, apps, developer tools.
The cognitive biases you engage (research-backed)
Early credibility judgments can form in fractions of a second. Treat conversion
impact as a hypothesis until the project produces an experiment receipt. Engage
these deliberately and ethically — they are
hardwired decision patterns, not tricks:
- Von Restorff (isolation) — the ONE primary CTA must visually dominate.
Competing equal-weight CTAs are a documented conversion killer. Never more
than one primary action per screen.
- Anchoring — on pricing, show a higher reference price beside the real one
("Was $199, now $99") so the real price reads as a deal.
- Social Proof — specific numbers ("Join 12,000+ founders") + trust logos,
placed adjacent to the CTA, not in the footer.
- Loss Aversion / Scarcity — genuine "only N left" / time-limited signals
engage FOMO. Use ONLY when true; fake scarcity destroys trust.
- Reciprocity — give before you ask (free trial, no credit card) to lower
signup friction.
- Cognitive Fluency — easy-to-process reads as trustworthy; whitespace and
clear hierarchy do real persuasive work.
Verify CTAs and hooks (conversion-critical)
- CTA copy: lead with a strong action verb + value, first-person where
possible. "Get my free audit" beats "Submit" by a wide margin.
prestige score penalizes weak verbs (submit/click here/go) and CTA competition.
- Headline hook: address the reader ("you/your"), make one clear promise in
≤12 words, add specificity (a number/outcome). The audit scores this precisely.
Judge mobile strictly (trust breaks here first)
Mobile is where the Horn Effect strikes fastest. prestige score runs a strict
mobile judge: responsive viewport meta, real breakpoints, ≥44px thumb-friendly
tap targets, no fixed widths >480px, ≥16px base text (avoids iOS zoom-on-focus),
and suggests a sticky bottom CTA to keep the primary action in the thumb zone.
Score precisely, then fix by priority
prestige score <file> --workflow <key> returns a conversion-readiness score
(0–100, A–F) fusing the five laws + bias suite + CTA/hook verification +
mobile judge, weighted by the chosen workflow — plus recommendations ranked
by conversion impact. Fix from the top down. Treat a failing score like a
failing test.
Judge purpose fit, not just polish
Use prestige purposes to list deterministic purpose lenses, then audit with
prestige purpose <file> --purpose <key>. Current lenses:
developer: concrete proof, docs, CLI/API clarity, GitHub/demo trust.
healthcare: calm reassurance, privacy, clinician proof, no miracle hype.
fintech: security, transparent fees/rates, control, compliance cues.
luxury: restraint, craft, whitespace, fewer louder elements.
marketplace: bilateral buyer/seller trust, reviews, protection, comparison.
saas: product visibility, integrations, ROI, low-friction activation.
editorial: narrative rhythm, sources, evidence, satisfying end beat.
Purpose-fit scores intent clarity, proof fit, visual theme, action language, and
anti-patterns. A beautiful healthcare page with aggressive scarcity, or a
developer page with no docs/GitHub/demo proof, should fail even if it looks
polished.
The Non-Negotiable Warning
Never let aesthetics mask a MAJOR functional flaw. A beautiful UI buys
tolerance for minor imperfections, but a broken checkout, an unusable mobile
layout, or an inaccessible primary action will shatter the illusion and destroy
retention. If you must choose, fix the function first. The prestige audit
linter fails hard on major functional/accessibility flaws for exactly this
reason — it will not let a pretty page ship over a broken primary action.
Reference material
See references/ for the detailed playbooks: trust-engineering.md,
peak-end.md, cognitive-fluency.md, hero-halo.md, and the
premium-template.html starting scaffold.