Alibaba Full Case Audit — What a Panel Sees Before You Walk In

Fetched cs-01-alibaba.html on July 3, 2026. Everything below reflects the live page.

The gate is still cosmetic. This is still the first thing to fix.

Flagged June 30. Nothing has changed.

The full Alibaba case study returns HTTP 200 and delivers 119KB of HTML before client-side JavaScript checks localStorage for pf_access_v1 and redirects unauthenticated visitors to a login page that checks a hard-coded SHA-256 hash. View-source defeats the gate. Disabling JavaScript defeats the gate. Any crawler defeats the gate. The content is fully readable before the "protection" executes.

Beyond the June 30 finding: the homepage links to the Alibaba teaser, and the teaser links directly to cs-01-alibaba.html twice. Hero CTA and closing CTA. That is a two-click public path from your homepage to the "protected" case. And the full case's own navigation links to adjacent full cases (Equinox+, Supply Chain), so anyone who reaches Alibaba can walk the entire portfolio. No sitemap.xml, no robots.txt. Both return 404. That means no crawl restrictions exist. Search engines are free to index anything they find, including pages reachable via the two-click teaser path. The practical risk is moderated by the absence of a sitemap pointing crawlers there and the lack of inbound links, but the protection is zero, not partial.

Fixes, ranked by effort:

1. Today, fifteen minutes. Remove the direct links from the teaser page to the full case URL. Replace with a line saying the full case is shared after booking. Does not fix the gate. Removes the public path.
2. Correct fix. Server-side authentication. A 401 that never delivers HTML without a valid session. Vercel middleware handles this without a backend.
3. If neither happens this weekend. Confirm the full case URLs are not linked from any other public surface. Right now they are.

Most panels are off through the Fourth. You have until Tuesday morning.

Delta against June 30

Issue	Status
Client-side gating delivers full HTML before redirect	Still open
Teaser title underleveling ("Product Designer" framing)	Resolved — hero now reads Head of Design & Research, North America
Homepage positioning as leadership-forward	Resolved
Full case content audit against panel-read standards	First audit — everything below is new ground

How a panelist reads this case

They scan the first screen, form a hypothesis about altitude, and spend the next ten minutes confirming or disconfirming it. The first sixty seconds determine whether the remaining minutes are generous or skeptical.

Six dimensions. Each one assessed against what the page currently does, whether it meets the standard, and what to change if it doesn't.

1. Problem diagnosis

The standard: The panelist should finish this section believing Juno saw what others missed.

What the page does: The hero states Juno "named the gap" between desktop traffic (25%) and transaction value (80%), and that "no one had named it yet." The diagnostic section organizes the problem into three failures: first impression, search journey, transaction confidence. Each failure is backed by a specific finding (37-point sign-in deficit, 73% of sessions ending without engagement, 47% citing payment security fears).

Verdict: Mostly meets the standard. The claim of naming the gap is strong. The three-failure framework reads as analytical structure imposed on messy data, which is what senior judgment looks like. A panelist finishes this section believing Juno identified something real.

One gap. The transition from "I named this gap" to "here are the three failures" is abrupt. The panelist doesn't see the inferential leap. How did a traffic/revenue mismatch become three specific failure modes? That leap is where senior judgment lives. Right now the case reads: gap existed, here are the problems. It should read: gap existed, I hypothesized these three failure modes, the research confirmed them.

Fix (HTML): One sentence between the gap statement and the three failures. Something close to:

That sentence is what makes the diagnosis read as judgment, not discovery.

2. Research methodology

The standard: The panelist should understand that the research was the persuasion, not just the discovery.

What the page does: Names 32 cross-functional interviews, a Baymard audit, and gaze tracking on $500+ order sessions. States they "converged on the finding" that desktop was designed for consumer shoppers while actual users were procurement buyers.

Verdict: Does not meet the standard. This is the biggest gap in the case.

The research reads as a process checklist. The panelist learns what methods were used and what they found. How the research built the executive mandate is invisible. At Director+ level, the research story is a leadership story. Who was skeptical? What did the gaze tracking show that changed a specific stakeholder's position? How did 32 interviews become an executive mandate rather than a research report sitting in a deck?

A CPO reading this will think: "She did good research." The evaluation you need is different: "She used research to move an organization." That second read is the one that gets hired.

Fix (HTML): Do not rewrite the section. Add two sentences that show the research as a persuasion instrument. Which finding created the executive shift? Was it the gaze tracking showing buyers checking external prices? Was it the 37-point sign-in deficit? Name the finding that moved the room, and name the room.

That is the sentence a CPO remembers.

Fix (verbal): If you cannot update the HTML before sharing, prepare a 30-second version for the panel conversation. "The research was a persuasion instrument. The gaze tracking became the artifact that moved the executive team from incremental to structural. I designed the research to produce that artifact."

3. Sprint rationale

The standard: For each of the three sprints (Homepage, Search, PDP), the panelist should see why this decision, in this order.

What the page does: Explicitly maps each sprint to a diagnostic failure. Sprint 01 = first impression. Sprint 02 = search. Sprint 03 = transaction confidence. Each sprint shows before/after visuals, specific design decisions, and outcome metrics.

Verdict: Nearly meets the standard. The mapping from diagnosis to sprint is clear. The sequencing rationale is absent. Why homepage first? A panelist who has run platform redesigns at scale knows that sequencing is a strategic decision. Did you start with homepage because it had the highest traffic? Lowest-risk proof of concept? Prerequisite for the search changes? Without that reasoning, the sprints read as a timeline. The panelist wants to see a strategy.

Fix (HTML): One sentence at the top of the sprint overview. "We started with first impression because [reason]. Each sprint's outcome funded the mandate for the next." The panelist does not need a paragraph. They need to see that sequencing was a decision.

4. IC + manager proof

The standard: Cross-functional leadership should be specific. Who Juno influenced, what she unblocked, how she operated across functions.

What the page does: Lists the team as Design, Research, PM, and Engineering. Sprint 01 includes one specific moment: PM concern about sign-in drop versus the decision to remove the credential wall. No other sprint names a specific stakeholder interaction, conflict, or operating mechanism.

Verdict: Does not meet the standard. The Sprint 01 PM tension is excellent. It is the single strongest proof of cross-functional leadership in the entire case. But it stands alone. Sprints 02 and 03 have no equivalent. A panelist reading sequentially will notice: one moment of specificity followed by two sprints of generic execution narrative. That pattern suggests either the PM tension was the only real fight, or that Juno defaults away from surfacing organizational complexity. Neither reading helps.

Fix (HTML): Add one specific cross-functional moment to Sprint 02 or Sprint 03. Name the person (by role, not name, if needed), the tension, and how you resolved it. The Sprint 01 PM tension is your model. Match that altitude. You know what happened in those sprints. The case needs it on the page.

Fix (verbal): Prepare two additional cross-functional stories for the panel conversation. The case gives you one. You need three. When a panelist asks about navigating disagreement, give a different story than the one already on the page.

5. Tradeoff reasoning

The standard: At least one design decision should show a real alternative considered and rejected, with reasoning. Panelists calibrate seniority on tradeoff reasoning more than on outcomes.

What the page does: Sprint 01 shows the PM concern about sign-in drop as an implicit tradeoff: remove the credential wall and risk losing sign-ins, or keep it and maintain the consumer-marketplace first impression. The outcome (+4% sign-up rate) resolves the tension. Sprints 02 and 03 show no comparable rejected alternative.

Verdict: Partially meets the standard. The Sprint 01 tradeoff is real and well-framed. But panelists calibrate on tradeoff density. One tradeoff across three sprints reads as a case that went smoothly. Senior panelists are suspicious of smooth, because their experience at $50B GMV tells them nothing goes smoothly.

Fix (HTML): Add one visible tradeoff to Sprint 03. The decision to show tier pricing inline on the PDP is a natural candidate: transparency versus supplier control over pricing negotiation. Did suppliers resist? Was there a version where pricing was revealed only after inquiry? Name the alternative considered and why it was rejected. One sentence changes the read.

6. Agentic sourcing vision

The standard: The closing vision should land as a forward-looking POV grounded in the case's evidence, not a speculative appendix.

What the page does: Rebuilds the same three surfaces (chat, search, PDP) as an agentic sourcing loop. Uses a specific example (500 custom water bottles, ≤20-day lead time, ≤$10/unit, Trade Assurance required). References the same trust signals (TA badges, response rate, verified supplier) established in the case.

Verdict: Meets the standard. One transition risk to manage. The vision is grounded. It uses the case's own evidence as the foundation. The specific example with real parameters makes it concrete. The trust signals carry forward from the redesign.

The risk is the seam. Coming after three sprints with measured outcomes (+7% DAU, +20% transactions, +47% new visitors, +2.2 NPS), the vision section has no metrics. A panelist reading sequentially moves from measured proof to speculative projection. That transition, unmanaged, makes the vision feel appended.

Fix (HTML): One framing line before the vision section:

This bridges measured outcomes to forward vision. Without it, the panelist has to build that bridge themselves, and under time pressure they won't.

Fix (verbal): When walking a panel through this section, lead with the bridge. "Everything we proved in the redesign, that trust signals drive conversion, that B2B-native sort controls change search behavior, becomes the foundation for agentic sourcing. The agent needs the same trust infrastructure the human buyer needed. We built it."

What to fix before Tuesday

1. Remove teaser-to-full-case direct links. Fifteen minutes. Eliminates the two-click public path.
2. Add the research-as-persuasion sentence (Dimension 2). Biggest gap. One to two sentences showing the research moving the organization, not just informing it.
3. Add one tradeoff to Sprint 03 (Dimension 5). One sentence naming a rejected alternative with reasoning.
4. Add one cross-functional moment to Sprint 02 or 03 (Dimension 4). Can combine with the tradeoff fix if the rejected alternative involved a stakeholder.
5. Add the diagnosis-to-framework bridge (Dimension 1). One sentence showing the inferential leap from gap to three failure modes.
6. Add the vision bridge sentence (Dimension 6). One line connecting measured outcomes to forward POV.

After these six changes, the case reads as a design leader who moved an organization and has a point of view about what comes next. Right now it reads as a strong designer who did good work.

The first reading is the only reading. Make it count before Tuesday.