c16 — Phase 2 Review: Value Articulation per Persona & per Page

Scope: the 14 control-plane-v1 wireframes + FleetSeatNav. Method: read every page's rendered content against one rubric — does the page tell its persona why it matters, what decision it supports, and lead with the "so what" — or does it just render data the user must interpret? Value-articulation scored 1–5 (5 = opens with the persona's stake + an explicit takeaway/recommended action; 3 = useful view, value implicit; 1 = raw data).

1. Headline verdict

The app is structurally excellent and narratively thin. The information architecture, the agent-native vocabulary, and the persona partition are right. But most pages report state rather than articulate value — they show the CISO/CFO/exec the data and trust them to infer the "so what." The pages that were written value-first (Security Posture, Executive Board) score highest; the rest cluster at 2–3.

Mean ≈ 3.1. The single most important pattern: the four pages that score 2 are not under-built — they're the ones that lead with metrics instead of a decision.

2. Three systemic gaps (the highest-leverage fixes)

These cut across every page and matter more than any single-page tweak.

G1 — The persona seats are routed but not framed

Every page reads ?seat= only to highlight the nav — the content and framing never change by seat. A Finance user landing on Fleet Overview sees the same operator framing as a CISO; an Executive drilling into an agent profile sees the same exhaustive forensic layout as a security analyst. The persona partition is real in the URL and the nav, but invisible in the page. This is the biggest "value per persona" gap: we built five doors into one generic building.

G2 — There is no seat landing ("your governance, at a glance")

FleetSeatNav links each seat directly to its first deep page (seatPages[seat][0]) — click Finance and you land on the cost-center table; click Security and you land on the inventory. The Phase-1 design called for each seat to open with "an at-a-glance strip + a grid of its page-cards." That seat home — the one surface that would say "As the CISO, here is your posture, your top 3 risks, and what to do this week" — does not exist. Without it, no seat ever states its own value proposition; the user has to assemble it from the deep pages.

G3 — Pages show state, not "so what" + next action

Most pages are read-only telemetry with no headline takeaway and no recommended action. The "critical few" are buried as row badges inside flat tables instead of surfaced as a lead callout (ungoverned/shadow agents on inventory; the one Critical agent; the top cost/risk driver). Few pages answer "so what should I do?" — there's no remediation queue, no "rotate these 3 keys," no "you're $X over budget."

The model to copy is already in the build. security-posture opens with "How safe the fleet is — the score, what lowers it, and where it is heading" (5/5). Every page should open with that shape: a stake, a number, and a direction.

3. Value articulation per persona

Security / CISO — best-served, lacks a tie-it-together home

Pages: agent-inventory (3), findings (4), security-posture (5), agent-identity (4), data-protection (2), policy-enforcement (4), agent-profile (2).

• Strong: Findings ("the CISO worklist"), Posture, and Identity all name the CISO's job and lead with their stake. This seat tells a coherent story if you know where to look.
• Weak: the agent-profile drill hub scores 2 — it's exhaustive but opens with the agent name and a danger badge, never a risk summary ("Critical because X, Y, Z → recommend: …") or a remediation path. Everything drills into this page, so its flatness undercuts the whole seat.
• Missing: a Security seat home rolling up "fleet score, top risks, open high findings, this week's actions." Today the CISO must visit 4 pages to assemble the board read.

Finance / FinOps — the weakest-articulated seat, despite being "served"

Pages: cost-center (2), showback (2), fleet-overview (operator-framed, 3).

• Phase 1 declared Finance "now served" — and the visibility is genuinely there. But every Finance page answers "what does it cost?" and none answers "should I act, and on what?"
• cost-center has no budget-vs-actual, no forecast, no "$X over / on track," no single trusted number (provider vs. estimated sit side by side with no guidance), and no signpost to showback for chargeback.
• showback buries its own punchline: "Showback ≠ chargeback" and the unverified-identity blocker are a footnote, not a "N identities block accurate chargeback → fix" callout. It also doesn't lead with top exposure ("user X — $Y at Critical risk").
• Net: Finance lands on operator framing and has to do the FinOps thinking itself.

Executive — one excellent page, three operator pages

Pages: executive-board (4), fleet-overview (3), security-posture (5, but security-framed), cost-center (2).

• executive-board is the strongest single surface: hero Governance Index vs. an explicit target, spend-at-risk, governed %, export. It reads board-ready.
• But it lacks a trend narrative ("index +6 pts this month; at this pace we hit target in N weeks") and any benchmark (is 71 good for our size/sector?), and the rest of the seat is operator dashboards, not board views.

Engineering / Platform — mixed; strongest on policy, weakest on performance

Pages: performance (2), policy-enforcement (4), integrations (3), agent-inventory (3).

• policy-enforcement is a highlight: the Audit→Assist→Enforce path + a quantified "what an enforcing policy would have blocked" dry-run (6 ungoverned, 3 root, 2 --yolo, 9.2M unbounded) is a compelling value demo. Missing: blast-radius weighting and "cost of flipping to Enforce."
• performance buries its best insight — that latency ≠ task success — in a footnote callout instead of leading with it; and it surfaces the missing Task Outcomes dimension (Phase-1 net-new #3) only as projected text.
• integrations is honest (live vs. projected) but doesn't rank connectors or offer a "turn on first" recommendation.

Compliance / Privacy — audit-honest, but no conformance-at-a-glance

Pages: compliance (3), data-protection (2), findings (4, security-framed), showback (2).

• compliance has audit-ready why-trails and honest projected-vs-live framing — but a compliance lead can't answer "are we passing SOC 2 / ISO 42001?" at a glance. It needs a conformance snapshot ("SOC 2: 12/14 controls have findings; NIST AI RMF: 3/5 functions covered") above the framework cards.
• data-protection shows movement (paths, volume) honestly but never states the privacy stake ("X% of agents have blanket read access to sensitive paths; Y credentials moved into context") or links to a DLP policy.

4. Per-page missed opportunities (most actionable, condensed)

5. What's working (preserve in hi-fi)

• security-posture's outcome-first headline — the template every page should adopt.
• executive-board's hero index + explicit target + spend-at-risk — board-native accountability.
• policy-enforcement's "would have blocked" dry-run — turns audit data into a value demo.
• Honest projected-vs-live framing everywhere (integrations, compliance, the identity disclaimer) — credible and the right epistemic posture.
• The egress demotion held — no page is organized around egress; the pivot landed.

6. Recommendations, prioritized

Do before / during Phase 3 (highest leverage — fixes G1–G3):

1. Add a seat landing per persona — "Persona · your governance at a glance": 3–4 metrics that matter to this seat + an attention/action strip + the grid of seat page-cards. Fixes G2 and gives every seat a value statement. (Was in the Phase-1 design; not built.)
2. Open every deep page with a headline takeaway + recommended action, and promote the "critical few" out of tables into a lead callout. Fixes G3. Copy the security-posture shape.
3. Make the top pages persona-adaptive — at minimum a per-seat "so what" line; ideally reorder/emphasis by ?seat (agent-profile leads with cost for Finance, risk for Security). Fixes G1.

Strong follow-ons: 4. Finance decision-support: budget-vs-actual + forecast on cost-center; chargeback-blocker urgency + top-exposure lead on showback; a cost-center→showback bridge. 5. Compliance conformance scorecard above the framework cards. 6. Surface Task Outcomes & Quality as a named gap on performance (it's Phase-1 net-new #3). 7. agent-profile risk-summary header + remediation links; policy-enforcement blast-radius + "cost to enforce"; executive-board trend narrative + benchmark.

Polish: cross-page next-step signposting; MoM deltas on fleet-overview KPIs.

7. Implication for Phase 3

The approved hi-fi lead set is executive-board, cost-center, agent-inventory (+profile) — and cost-center (2), agent-profile (2) are exactly two of the lowest value-articulation scorers. Graduating them as-is would carry the "shows state, not value" gap into the polished mockups. Two options:

• (A) Apply the value-framing fixes (§6 #1–#3, scoped to the lead set) as a quick wf2 wireframe revision first, then mock up the improved pages. Cleanest; keeps hi-fi focused on fidelity, not redesign.
• (B) Bake the §6 fixes directly into the m1/m2 mockups (treat hi-fi as the place the value framing matures). Fewer steps; risks conflating "is it pretty" with "does it land."

Recommendation: (A) for the seat landing + agent-profile/cost-center headlines (they're structural and worth proving lo-fi), then proceed to hi-fi.