Baseline established: July 3, 2026. First structural audit of the full Equinox+ case at junochen.com/cs-04-equinox.html. The teaser page was fetched alongside it for continuity checks. Every finding below is verified against live HTML as delivered by the server today.
Most hiring committees are offline this weekend. H2 headcount unlocks mid-July. You have ten days to close every gap on this page before it starts doing interviews without you in the room.
The gating is broken. Same pattern. Same severity.
The full case page delivers all content to the browser before the client-side JavaScript checks localStorage for pf_access_v1. The redirect script sits at the top of the body. The content follows it in the same server response, 154,227 bytes of it. A raw HTTP fetch with no browser session receives everything: hero, research, all five architecture decisions, launch outcomes, team scope, every image reference.
View-source bypasses the password. Link unfurl bots bypass the password. Search crawlers bypass the password. Anyone who right-clicks receives the full case regardless of access state.
The content is fully exposed. This is the same client-side gating vulnerability flagged on the essay page, same severity, same fix.
Minimum viable fix: Move case content into a separate HTML fragment loaded via JavaScript only after the access check passes. The server should never deliver protected content in the initial response.
Correct fix: Server-side access control. Unauthenticated requests receive only the gating UI.
One additional gap while you're in the code: neither page contains og:title, og:description, or og:image tags. When a recruiter drops this link into Slack or a hiring-committee email thread, it previews as nothing. The teaser page needs this most urgently because the teaser is the page you send to people who haven't been granted access. Its unfurl description should preview the strategic frame: five premium brands, one platform, competitive constraint against Peloton, 0-to-MVP in three months. That description is doing outbound sales work every time someone shares the link. The full case page needs OG tags too, but the teaser's are higher leverage. Thirty minutes of work that changes how every shared link renders.
What a buyer reads in the first scan
The hero does real work. "Five brands. One platform. No compromise." is a clear architecture statement. The stat row (600K members, 4.8★, 5 brands, 3 months) gives the panel four anchors before they scroll. The brief block establishes role, team size, and scale. Solid scaffolding.
The problem starts one scroll later.
The Peloton constraint appears and then functionally disappears. The Brief names three constraints: no leaderboard-first engagement, no brand flattening, no treating club access as separate from digital. These are strong. They tell the reader exactly what kind of design problem this was. But Decisions 01 through 05 don't reference them. Each decision reads as a standalone architecture choice with its own internal logic. The constraint that generated the decision is implicit.
A VP Design scanning this case will read the Peloton constraint, nod, scroll to the decisions, and find no connective tissue. In conversation they'll ask how the competitive constraint shaped your IA. You'll answer well because you lived it. But the page is doing the first interview without you present, and right now it drops the strategic thread at the exact moment the decisions should be pulling on it.
Add a single line at the top of each decision section naming which constraint it resolves. Decision 02 (instructor-first navigation) answers "no leaderboard-first engagement." Decision 04 (token architecture) answers "no brand flattening." Decision 05 (club loop) answers "digital never replaces the club." The register:
"This decision answers the first constraint: no leaderboard-first engagement model. If the instructor is the product, the leaderboard can't be."
One sentence per decision. The reader should never have to infer the connection. Make the through-line load-bearing.
The speed claim is actually well-handled. The page names what shipped at launch and what was deliberately cut. The "one call under pressure" cell describing the six-weeks-out prioritization of SoulCycle, Equinox, and Pure Yoga based on instructor-followership density is the strongest single paragraph on the page. It shows a design leader making a scope call with data backing and real consequences.
One refinement: the cut list currently reads as a flat inventory. Reorder it by what hurt most to cut versus what was easy to defer. Precision Run and HeadStrong were brand worlds with real stakeholders who wanted their content represented at launch. Cutting them is a different kind of decision than deferring social sharing. The panel wants to see that you felt the cost.
What they'll probe in conversation
The data-to-IA derivation is close but not airtight. Decision 02 presents the retention signal: members who followed an instructor within 30 days retained at nearly double the rate. Then it says you "spent a week in the data before the architecture became followership." The retention stat is the right anchor. "Spent a week in the data" is a process claim without visible process.
What was the alternative IA you considered and rejected? Content-category browsing? Brand-first navigation? A panel member will ask. The page should pre-answer.
Name the rejected IA model. Something like:
"The default IA would have been content-category browsing, the model every competitor used. The retention data made that indefensible."
One sentence. This turns a design intuition into a visible analytical move with a traceable evidence chain, one the panel can follow step by step. The rejected alternative is what makes the chosen path read as rigorous rather than reconstructed.
The multi-brand architecture tradeoff names benefits but not costs. Decision 04 describes shared behavior where consistency matters, variable expression where brand meaning matters. Token architecture. Five independent brand expressions, one component base, one deployment. Every phrase here is a benefit.
Token architecture at five-brand scale has real costs. Design system velocity slows because every component change must be validated across five visual contexts. Brand Creative Directors have autonomy over expression but not over structure, which creates negotiation overhead on every structural change. QA multiplies. These are the tradeoffs a Director+ candidate should name, because naming them proves you understood the full system, the costs, the negotiations, the validation overhead, and built it anyway.
Add a "what this cost" line after the architecture description. Two sentences. Name the QA multiplication, the cross-brand validation overhead, or the Creative Director negotiation process. Whichever was most real. When a tradeoff only shows upside, panels read it as a sales pitch. Visible costs on both sides read as engineering judgment.
IC contribution and leadership scope are both present but tangled. The operating-scope strip works: you hired 2 of 4 designers mid-project, owned IA and design system and token architecture, influenced roadmap and MVP scope. The "one call under pressure" cell shows leadership judgment. But the five decision sections read as team output described in the first person. Decision 02 says "Juno spent a week in the data." Decision 03 says "the detail Juno wanted most." These are IC signals embedded in what otherwise reads as leadership narrative.
Pick two decisions where your hands were on the work. Make the IC contribution explicit. Pick one where your role was directing the team. Make the delegation explicit. A sentence like:
"I built the first token map myself before handing the system to [designer] to scale across the remaining brands."
That separates the IC moment from the leadership moment in a single line. The panel needs to see both altitudes clearly, each carrying its own weight.
Where the page leaves them without answers
The 4.8★ App Store rating is claimed as a general outcome, not tied to a specific design decision. The launch section says "top review themes mapped to the five decisions" and names instructor discovery, live class quality, and cross-brand navigation. "Mapped to" is doing load-bearing work here without showing the mapping.
This matters more than it might seem. At Director+ level, every candidate on a shortlist has shipped something with strong metrics. Skeptical panels have learned to discount outcome numbers that sit next to work without a visible causal thread, because they've seen too many candidates attach org-level results to individual contributions. 4.8★ becomes evidence only when the causal thread is visible.
Replace "top review themes mapped to the five decisions" with two specific pairings. "Instructor discovery, the most-cited positive theme, traces directly to Decision 02. Cross-brand navigation traces to Decision 04." Straight lines the reader doesn't have to assemble.
Priority sequence
Holiday weekend. Ten days before H2 searches start moving. Two work sessions.
This weekend (code and metadata):
- Fix the gating. Today. The content is exposed.
- Add OG tags to both pages. Teaser gets the strategic-frame description. Full case gets a standard case identifier.
- Fix the teaser watermark that reads "Case Study 03" when the URL, sidebar, and full case all say 04. Fix the footer navigation mismatch (teaser links to Allē as next, full case links to Alibaba). And address the reused image asset (
instructor-feature.jpgserving three different alt-text labels). Small things that create trust cracks before your content gets a chance.
Before July 14 (content):
- Add constraint-to-decision connective tissue. One sentence per decision section linking back to the Peloton constraints. Highest-leverage content change on the page.
- Name the rejected IA alternative in Decision 02. One sentence that makes the analytical move visible.
- Add costs to the multi-brand tradeoff in Decision 04. Two sentences.
- Separate IC and leadership altitude in at least two decision sections.
- Tie the 4.8★ to specific decisions with direct pairings, not "mapped to."
The research is real, the architecture decisions are genuinely interesting, and the scope strip does more work than most portfolio pages attempt. What's missing is the connective tissue that turns five good decisions into a visible decision framework. A panel reading this case should finish knowing what you built, how you thought, what you rejected, and what it cost. Right now they reliably get the first. The other three require the fixes above.
-
Multi-brand architecture as differentiator: Headway's two open design director roles, Provider Experience and Core Insurance, test different speed profiles within the same company, and the Equinox+ token-architecture pattern maps directly to the multi-surface, permission-aware design system both roles need.
-
Ramp's agentic design language: Ramp's Director of Product Design posting describes designing memory, shipping PRs, and building tools alongside agents, which means the Equinox+ "one record, multiple contexts" architecture translates if you frame it as the structural precursor to agent-aware product surfaces.
-
Trust essay as connective proof: The live Trust essay names consistency as what the design delivers versus accuracy as what the model achieves, and the Equinox+ token architecture is a pre-AI version of the same problem: maintaining brand trust across five variable expressions from a shared system.
-
Capital One's operating-model test: Capital One's AI in Experience Design director role reports to the CDO and owns AI tooling, process evolution, and talent upskilling, so the Equinox+ scope strip showing mid-project hiring, cross-brand Creative Director negotiation, and roadmap influence is the right leadership-altitude evidence for that room.