Finance — CFO Redesign Plan

What changes per page

• Drop dark-card-on-stone-50 gradients. Use .aspire-card / .cfo-card patterns (this page is the live example).
• Drop emoji-in-string. Replace 🚨 / 💳 / 📊 / 🔁 with Material Symbols (warning / credit_card / monitoring / autorenew).
• Use --terra for warnings (not red-500), --hunter for CTAs and "okay" state, --text + --text-sec for hierarchy.
• Tabular numbers via --font-num (Unbounded) for big balances; DM Sans for body; DM Serif Display for h2/h3.
• Mobile baseline 390px — no horizontal scroll on phones, tables get .audit-scroll wrappers, grids collapse to 1-col below 640px.
• Studio Dark must pass WCAG AA — checked via /info/accessibility matrix. Same standard the rest of CC just adopted.

The insight shape

{
  id: "venmo-visa-interest-bleed",
  title: "Venmo Visa is bleeding ~$200/mo in interest",
  impact_dollars: 2400,            // annualized
  confidence: "high",              // high · medium · low
  urgency: "now",                  // now · 30d · 90d · ongoing
  lens: ["debt", "cfo"],           // which lenses agree
  recommended_action: "Avalanche this first — $7,174 @ 35.24% APR is the highest-cost revolving balance.",
  needs_confirmation: false,
  related_accounts: ["venmo-visa"],
  data_sources: ["snapshot.json#venmo-visa", "statements/venmo-2026-04"],
  generated_at: "2026-05-19",
  status: "open"                   // open · snoozed · dismissed · resolved
}

Priority score = impact_dollars × urgency_weight × confidence_weight × (1 − ignored_before_penalty). Top 3 on Home, top 10 on a "View all" overflow. Snoozed/dismissed insights age out gracefully.

The badge vocabulary

Current

Last statement within 35 days. High confidence.

As-of

Last statement 35–90 days. Show the date inline.

Stale

>90 days since last statement. Suggest upload.

Estimated

Computed (Zillow, KBB, point estimate). Not a parsed fact.

Missing

Account exists but no recent statement parsed. Show the gap.

On Home: a single "Data current through {date}" line in the header. On detail pages: per-account badges in the table.

Card-variant identification (blocks Rewards Command Center)

• USAA Visa 2957 — Rewards Visa Signature, Preferred Cash Rewards, or Cashback Rewards Plus (military-base 5%)? Earn structure differs sharply.
• BoA card 3291 — Customized Cash Rewards ($0 AF) or Premium Rewards ($95 AF)? Both common. Determines whether the $10,920 balance is on a category-bonus card or a travel card, and whether the May 26 Preferred Rewards cliff applies.
• Venmo Visa last4 — confirm the last 4 so we can match transactions.
• Cap1 Quicksilver 5101 — yours or Jaime's?
• Lowe's Advantage last4 + current balance.

Lifestyle & goals (blocks Goals + insight prioritization)

• What's the monthly "enjoy life" budget you don't want squeezed?
• Categories that are off-limits for "cut suggestions" — i.e. things you and Jaime value? (Travel? Kids? Cleaning service?)
• Top 3 financial goals for next 6 months — pay off X card? Build $Y emergency? Save for Z trip?
• Strict alerts or gentle nudges? (My read: gentle nudges, but want to confirm.)

Data-completeness gaps (blocks specific calcs)

• Student loan current APR & status (Nelnet login needed).
• Bonnie loan start year (for cumulative interest paid).
• Pathways HELOC variable rate basis (Prime + X%).
• Vehicle KBB values to replace the $100K estimate.
• Anniversary dates per card (when the AF actually posts) — most you can pull from the card account online; some I have from statements.

Risk · Scope creep into a Mint replacement

The brief is huge and could pull this toward "rebuild Monarch/Mint." Mitigation: every phase is independently shippable. We don't promise live bank connections, real-time data, or budget envelopes (you already have envelopes at USAA). The dashboard is a decision surface on top of statement data, not a banking app.

Risk · Card-variant ambiguity blocks Rewards module

USAA 2957 and BoA 3291 have very different earn structures by variant. The methodology doc handles this by listing both with "TBD" — the Rewards module will render the same way until you confirm. Until then, "Which card to use" can still answer for the 5 cards we have confirmed.

Risk · Insight engine over-confidence

If we rank insights too aggressively, we'll feel like a financial coach lecturing. Mitigation: every insight gets a calm voice and a "snooze for 30 days" affordance. No streaks, no scores, no shaming language ever.

Assumption · You and Jaime share one dashboard

Per-person filtering exists in data (we know who owns what), but UI defaults to household view. If you want separate Topher / Jaime views as a first-class toggle, that's Phase 1 scope, not Phase 9. Tell me now.

Assumption · Linear OPS-54 stays the system of record for the paydown decision

The Debt Payoff page surfaces scenarios; the decision lives in OPS-54. I won't make the call for you — but I'll make the math undeniable.