Yale SOM · Slate · Design review
The design work, seen from outside Slate.
Everything applicant-facing is live on TEST across all five programs, verified to the byte, and admissions reviewers are actively using it. This page is a complete window into that work — the system, the open decisions, and the evidence — so it can be reviewed and directed without a Slate seat or a developer login. It opens in any browser; nothing to install, no account.
For your review
Your review, your call
This portal is the whole thing in one place — no Slate seat, no login, nothing to install. Open it, walk it, and direct it. Below: how to walk the work, where the calls are yours, and a proposed agenda for the review.
How to walk the application — three ways, same access I have
- This portal — every open decision with its before → after, the design system, and the
evidence, rendered live from the real
build.css. Scroll §01 and call each one. - The live TEST application — the real applicant journey across all five programs, already live and in use by Admissions. Walk it end-to-end and flag anything; every page self-identifies its release so a comment anchors to an exact version.
- Together, live — we walk it in the meeting; anything we can fix in the room, we fix in the room (the build system has every knob wired, so iteration is fast).
Proposed agenda
- Feedback → sign-off. Walk the live application; capture the top ~5 items, estimate effort live, and aim to leave with a clear path to sign-off (with Admissions). Many items we can fix in the meeting.
- Migration. Process + who does what. API access is helpful but not required: with it we grab the live HTML and remediate programmatically; without it, there's a content-remediation tool Admissions can run — already in IT's hands and understood. Choose-your-own-path; every path lands at the same remediated result.
- Post-launch. Remediating issues — design-system vs page-level design: who owns which? (Proposed: the design system + build tooling stays with me as steward; page content + decisions with Admissions.)
- Handoff. Steady-state responsibilities. The toolset, fixtures, gates, and remediation tool are built and documented — IT holds the receipts.
- Observations & learnings — if time; otherwise noted for follow-up.
Ownership, for our framing through migration + handoff: the build system, branding platform, fixtures, design system, and remediation tool are mine (created for this work — see the authorship note at the foot of this page); Yale owns the Slate instance, the brand content, and the go-live calls. Both true at once.
01 · The decisions
What's done — and what's yours to call
Each card: what it looks like now → the options → the recommendation → why → effort. Accessibility/contrast fixes are marked Done with the before/after; the brand and taste calls are marked Your call.
Review/Submit links no longer look like errors Done · accessibility
Now: on the last screen before an applicant submits, the section links read as pure red and orange — colors that universally mean error/danger, here meaning "go finish this." They also failed WCAG contrast (red 4.0:1, orange 2.5:1) on the highest-anxiety, most-screenshotted page in the funnel.
After — accessible, calm, intentional
Academic history — incompleteTest scores — needs attention
+ a non-color "Action needed" marker — never color alone.
Done: darkened to an accessible severity pair (#cc0000 5.89:1 ·
#b45309 5.02:1) with a non-color marker. Turns the scariest page in the funnel calm and accessible,
and closes a real legal-exposure point on the exact page a complaint would cite. These links
are editor-authored page content the stylesheet can't reach, so the fix is applied at the content source with a
before/after replay check — the same engine demonstrated here.
Brand the fields applicants type into Your call low effort · low risk · highest reach
Now: buttons are confidently Yale-blue, but the text boxes, dropdowns, and text areas inherit raw OS-default chrome — no branded border, no consistent height, and no Yale focus ring. It's the one place a Yale form still reads as unbranded vendor software.
Before — OS default
After — light brand frame + focus ring (tab in)
- A — light brand frame: one scoped rule (subtle border, consistent height/padding, + the Yale focus-visible ring).
- B — focus ring only.
- C — leave fully native.
Recommend A. Applicants spend nearly all their time looking at fields, not buttons — this is what makes a form page actually look like Yale. It stays strictly inside the agreed architecture: native controls, branded type/spacing only — no application machinery touched.
Your call: A (frame + ring) · B (ring only) · C (native).
One clear primary call-to-action at the front door Your call low effort · low risk
Now: som.yale.edu leads with a bold filled "Apply Now." Our apply-landing and login/register screens lead with quieter outlined buttons — and the only filled-blue button appears after the applicant is already inside a form. The strongest action is hidden deep while the entry actions whisper.
- A — keep all-outlined.
- B — promote exactly one filled-blue primary per entry screen ("Start New Application" / "Create an account"); keep the rest outlined.
- C — adopt the marketing animated-arrow link CTA.
Recommend B. The front door is where the first impression forms after leaving the marketing site; one confident primary action carries that visual confidence inward.
Your call: A (all outlined) · B (one filled primary) · C (animated link).
Plus a set of low-risk quick wins ready on request — H1 cap on form pages,
consistent question rhythm, a darker link underline, 44px mobile nav target, and a calm "success/info" message style
to match the already-styled errors. Full menu: docs/BRANDING_DESIGN_DECISIONS_2026-06-11.md.
02 · The work, seen
The real applicant pages — without a Slate seat
These are faithful offline renders of the actual applicant-facing surfaces, drawn through the same styling cascade Slate serves. This is what reviewers see — visible here to anyone, no login.

Apply landing — the front door (desktop) 
Sign in — branding before login (desktop) 
Application form — side-nav + fields (desktop) 
Review — the pre-submit page, accessible links (desktop) 
Footer — 1:1 with som.yale.edu (desktop) 
Mobile — first-class, not a fallback (375px)
Curated set — the full gallery of captured pages and viewports lives in the offline lab. Each render is screenshot-baselined, so a styling change moves the picture and a diff catches it.
03 · The system
A design system, not a CSS patch
Where Communications controls the surface, this is a real system: a golden-ratio type scale on a
locked 16px root with a measured reading rail; a disciplined Yale-blue palette with zero gold/green leakage on the
web (verified); real Yale brand fonts actually shipped; header/footer at 1:1 parity with som.yale.edu; and
accessibility that's genuinely strong where we own it. The reference below renders live from build.css —
every value is read from the CSS at view time, never transcribed.
04 · Why these choices
The work has a politic — and it's the mission's
These engineering decisions aren't arbitrary taste; they're commitments, and they trace straight to educating leaders for business & society. The application portal is pedagogical in its first three seconds: it teaches an applicant whether the institution expected them to already know the codes. We design to widen that threshold, not guard it.
- Accessibility is justice for the unseen. Access is the applicant's right and our obligation — enforced in CI against WCAG AA (AAA-contrast where we author), at mobile and desktop. We bring receipts: render-verified, test-green, never "green on assertion."
- The burden is ours, never the applicant's. We never blame a slow device or connection — we carry the cost of performance so an applicant on a metered signal or the device they already have never pays for our bloat. Bandwidth is an equity axis: the door opens as wide on 3G as on campus Wi-Fi.
- Leaner is greener. Cheaper selectors and lighter CSS are an equity and ecological commitment — kinder to a slow device and to the grid.
- Use the institution's own resources to open the door wider. The brand layer is a disciplined fork of Slate's own template, built to be reusable so it can lift other schools too — never plundering.
- In the institution, not of it — and honest about the difference. We use the institution's own CSS to widen the door for the applicant standing at it today, and we name where the institution falls short of its own mission rather than let good UX stand in for justice. We refuse the gesture that buys redemption without changing the structure. The full, cited account — including the structural ledger of Yale in New Haven — is in the philosophy.
Full rationale, fully cited: docs/DESIGN_PHILOSOPHY.md
(the lineage — Galtung, Gilmore, Robinson, Ngũgĩ, Žižek, Moten & Harney, Wilderson, Hartman, Freire, hooks).
docs/DESIGN_PHILOSOPHY.md.
05 · Settled
Done and not reopening — state these as strengths
- Type system — golden-ratio scale, 16px root, 64rem reading rail. A real design system.
- Yale-blue palette, zero gold/green on the web (verified 0 occurrences).
- Real Yale brand fonts shipped — YaleNew + Neue Haas Unica as woff2, not Arial.
- Header & footer at ~1:1 parity with som.yale.edu 2025.
- The rails — one density knob, fluid-fill-then-cap.
- One core responsive breakpoint (737) — one clean flip for the application.
- Accessibility where we control it — links 13.4:1, ink 18.7:1, links underlined (not color-alone), comprehensive keyboard focus.
- Mobile is first-class, not a fallback — polish most Slate shops never touch.
06 · Status & stewardship
Where it stands, and the ask
Current state: applicant-facing branding is live on TEST across all five programs, verified to the byte; admissions reviewers are actively using it; production is gated by deliberate choice.
The proposal: a defined review window each cycle — feedback inside the window ships; feedback after it is the next cycle's scope, not a last-minute blocker. The pieces already exist (every page self-identifies its release so a comment anchors to an exact version; the feedback hub collects it). It needs backing as the process — and Communications is the right place for that.
On the radar (not an ask): the post-submission status portals run on a separate Slate rendering system and still look like the old era — the natural next phase, bigger than a restyle.