CS-01 Alibaba Teaser Audit — Issue 2: Agentic Signal at Scan Speed

Serial audit of junochen.com/cs-01-alibaba-teaser. Second installment; the first (July 10) resolved the page title to mandate-forward and flagged density as the primary open tension.

Verified current state, July 19, 2026

Live HTML fetched today. Last modified July 19, 2026, 13:48 UTC.

Resolved since July 10:

• Page title: Alibaba.com — Redesigning B2B Trust at $50B+ GMV — Juno Chen. Mandate-forward. Done.
• Meta description leads with role, outcomes, research signal. Done.
• og:title matches the mandate frame. Done.

Still open:

• og:description still omits the −47% stat that the standard meta description carries. Nine days.
• Component atlas: 16 rows, 41 named components, ~874 words. Still on the teaser.
• Section numbering still jumps III → V.
• H1 still reads The signals before anyone touched a screen.

The organizing question

A buyer hiring a design leader for human-agent workflow coherence scans this teaser for 30 seconds. Does Alibaba register as relevant?

No. The case closes with an agentic sourcing architecture that is genuinely differentiated portfolio material. The teaser gives no indication it exists. A buyer scanning for AI-relevant design leadership sees enterprise platform redesign, categorizes it as legacy work, leaves. Never requests the full case. Never encounters the 378 words that would have reframed the entire conversation.

Six-point audit

1. Page title

Resolved. Alibaba.com — Redesigning B2B Trust at $50B+ GMV — Juno Chen works across Slack unfurl, browser tab, search results.

The title positions this as enterprise platform work. A buyer scanning LinkedIn shares for AI-relevant leadership will not click. I'm not recommending a title change. Trust-at-scale is the primary story and it's accurate. The agentic dimension needs to surface inside the page, not hijack the title. Confidence: high.

2. First screen

Hero delivers mandate, scale, outcome before scroll. Proof rows present. The July 10 recommendation to reorder with Outcome first remains open. Secondary.

Agentic dimension on first screen: absent. Hero paragraph covers procurement buyers, the consumer-to-B2B gap, the trust redesign. No mention of AI, agents, the sourcing loop. Nothing.

Add one line to the hero paragraph. After the current closing sentence about redesigning through $50B+ in GMV:

That sentence does three jobs: signals AI relevance at scan speed, previews gated content without giving it away, frames the sprint work as foundation for the agentic architecture. Write it as a bridge, not a pivot. Confidence: high.

3. Length and density

Unchanged from July 10. Page runs 7–8 viewport heights on a standard laptop. The component atlas accounts for more than half the body text. Roughly 874 of ~1,526 words. The teaser has one job: justify requesting the full case in 30 seconds. A 16-row component rack with 41 named cards does not do that job.

Move the component atlas to the full case. Leave a single thumbnail-scale row on the teaser. Three to five component cards. Enough to signal system thinking. Not enough to become a catalog. I wrote this nine days ago. It remains the highest-impact structural change available.

Density costs the agentic signal directly: even if you add the bridge line to the hero, a buyer who scrolls past three sprint sections and hits a 16-row component wall will never reach the closing section. That closing section is exactly where you'd place a fuller agentic preview. The atlas is blocking the path to the most differentiated content in the case. Confidence: very high.

4. Craft signal

The three sprint sections remain strong teaser material. Each pairs a diagnostic sentence with before/after screenshots. Visual quality is high. Legacy-to-redesign contrast is immediate. A buyer sees design judgment, not just execution.

The component atlas also demonstrates craft. Wrong altitude for a teaser. Forty-one components say "I built a system." Three components say "I think in systems." The teaser needs the second message. Confidence: high.

No new recommendations. Sprint sections work. Trim the atlas.

5. Call to action

Two Request the full case study CTAs exist: one in the hero, one closing. Both link to the full case. Gated structure is clear.

The CTA text gives a buyer who cares about human-agent workflow zero reason to click. They see "Request the full case study" and have no idea that behind the gate sits a complete NL-to-order sourcing loop with agent traversal and autonomous transaction completion.

Revise the closing CTA only. Hero CTA serves the general buyer. The closing CTA, positioned after sprint summaries and the agentic preview, should tell the AI-focused buyer what's behind the door:

Or shorter:

Small change. The buyer who has scrolled to the bottom and is deciding whether to click now has a reason that matches their hiring problem. Confidence: high.

6. Agentic signal

I went through the teaser HTML line by line:

• "AI Mode" appears as a tab label in the component atlas. UI element, not narrative. Around viewport 4, buried past ~414 words of scrolling. You'd have to be reading component card names to notice it.
• "Agent" appears once as a supplier-type filter option alongside Manufacturer and Trading Co. Data label, not design story.
• "Chat Now" appears once as a button on a supplier card. Another UI label buried in the atlas.
• The words agentic, natural language, autonomous, automation do not appear anywhere in the teaser. Anywhere.

The full case tells a different story. A section titled The same three surfaces. A complete sourcing loop. Natural-language chat becoming a structured brief. An agent traversing Search and PDP in real time, evaluating trust signals against buyer parameters, completing the transaction autonomously. A human decision gate with three options: place order, request sample, compare more suppliers. Surfaces designed with parsed parameters, supplier cards, budget math, progression labels. This is architecture, not theory.

378 words of the most differentiated content in the entire portfolio. Invisible from the teaser.

Add a brief agentic preview section between the sprint summaries and the closing. Four lines maximum:

This bridges sprint work to the agentic layer without hypothetical framing. Tells the scan-speed buyer: this case is relevant to you. Confidence: very high.

Prioritized action list

Fix first:

1. Add the agentic preview section between sprint summaries and closing. Four lines. Highest-value addition available. Makes the case visible to the buyer segment that is actively hiring right now.
2. Move the component atlas to the full case. Leave 3–5 thumbnail components on the teaser. Top recommendation July 10. Top structural fix today. Every day it stays is a day the teaser runs twice the length it should.
3. Add the hero bridge line referencing the agentic sourcing loop. One sentence after the current mandate paragraph.

Fix second:

4. Revise the closing CTA to reference agentic content behind the gate.
5. Restore −47% buyer-reported security concerns to og:description. Nine days open. Copy-paste from the standard meta description.
6. Fix section numbering (III → V jump).

Leave alone:

• The H1. Still not mandate-forward. But the hero paragraph and proof rows compensate, and the agentic additions will do more for scan-speed relevance than an H1 swap would.
• The sprint sections. They work.
• The hero CTA placement. Secondary to everything above.

The Alibaba case already contains the most compelling agentic design narrative I've seen in a senior portfolio. It is locked behind a gate with no sign on the door. Put a sign on the door.