docs(qa): remove 4 high-risk unsupported marketing/security promises

Per docs/qa/audit-2026-06-26/FINAL-REPORT.md. Audited 32 customer-facing promises across marketing, docs, and public UI; 22 lacked code backing. This commit removes the four highest-risk items: - Fabricated landing stats (500+ farms, 98% on-time, 50K+ orders, 2M+ lbs) and 4.9/12 JSON-LD rating on src/app/page.tsx + HeroSection.tsx. - Fake named testimonials (Marcus / Sandra / James with numeric claims) replaced with honest 'Early access' copy + contact CTA. - Pricing source-of-truth mismatch: Farm annual 152280 -> 134100, Enterprise annual 0 -> 359100 (25% off, matching pricing.ts). - Security claims (SOC 2 Type II, 99.9% uptime SLA, quarterly pentests, 2FA via Supabase Auth) re-scoped to providers we actually use (Vercel, Neon Postgres, Stripe). Provider cards updated. - Plus LOW-risk cleanups: OG image refs (-> .svg), SMS Campaigns + Route Optimization reclassified as shipped on roadmap, /roadmap/suggestion form replaced with mailto:, wholesale login Google error string fix. Note: src/app/wholesale/login/page.tsx and src/app/pricing/PricingClientPage.tsx were also touched by an unrelated useState->useReducer refactor in the same session. Those changes are bundled here because git add -p per-hunk separation would be brittle. Follow-up commits should isolate the remaining 81 modified files if they belong to a coherent change. Remaining 10 unsupported items documented in PROPOSE section of FINAL-REPORT.md need separate decisions (see that file).
2026-06-26 17:08:08 -06:00
parent f36e5da3f9
commit bb349e42f5
12 changed files with 2327 additions and 1469 deletions
@@ -0,0 +1,141 @@
+# Promise-Audit Final Report — 2026-06-26
+
+**Companion to:** [`PROMISE-AUDIT.md`](./PROMISE-AUDIT.md) (full inventory).
+
+## TL;DR
+
+The audit found **32 distinct customer-facing promises** across marketing, docs,
+and public UI. **22 of them did not survive contact with the code**. After
+applying the user-approved fixes, the four highest-risk unsupported promises
+are now removed from public copy and replaced with defensible language. The
+remaining 10 medium/low-risk items are documented in PROPOSE section below
+and need a separate decision.
+
+| Risk | Before | After |
+|---|---|---|
+| **HIGH** — fabricated landing stats | 4 fake stats + 4.9/12 JSON-LD rating | Removed; section hidden; rating block deleted |
+| **HIGH** — fake named testimonials | Marcus / Sandra / James with numeric claims | Replaced with honest "Early access" copy + contact CTA |
+| **HIGH** — pricing source-of-truth mismatch | `$1,341` vs `$1,522.80`; `$3,591` vs `$0` | Aligned to `pricing.ts`: Farm $1,341, Enterprise $3,591 |
+| **HIGH** — security claims (SOC 2, 99.9% uptime, pentests, 2FA) | All four on the security page | Re-scoped to providers we actually use (Vercel, Neon, Stripe) |
+| **LOW** — broken OG image references | `og-default.jpg` referenced, only `.svg` ships | All three pages now point to `.svg` |
+| **LOW** — roadmap "Mobile App iOS & Android" misclassified | In Progress (no native code) | Left in place — only a PWA spec exists; **needs separate decision** |
+| **LOW** — roadmap "SMS Campaigns" / "Route Optimization" already shipped | Listed as Planned | Reclassified as Shipped |
+| **LOW** — `/roadmap/suggestion` form with no handler | Button goes nowhere | Replaced with `mailto:` link |
+
+## Files changed
+
+| File | What changed |
+|---|---|
+| `src/components/landing/HeroSection.tsx` | Replaced fabricated `HERO_STATS` and `STAT_COUNTERS` with empty arrays; `StatsSection` returns `null` while empty; HERO_STATS render guarded by `length > 0`. |
+| `src/app/page.tsx` | Removed `aggregateRating` block from JSON-LD; changed Enterprise price spec to "Custom pricing" with no fixed `price`. |
+| `src/lib/stripe-billing.ts` | Farm annual 152280 → 134100; Enterprise annual 0 → 359100. Both now 25% off, matching `pricing.ts`. |
+| `src/app/pricing/PricingClientPage.tsx` | Emptied `TESTIMONIALS`; `Testimonials` component falls back to honest "Early access" copy + `/contact` CTA when array is empty. |
+| `src/app/security/page.tsx` | Replaced `SECURITY_FEATURES` with provider-scoped versions; replaced `TRUST_BADGES` (no more SOC 2 / PCI badges); metadata description + keywords updated; "Supabase" provider card → "Neon Postgres". |
+| `src/app/pricing/page.tsx`, `src/app/blog/page.tsx`, `src/app/contact/page.tsx` | OG image URLs changed `/og-default.jpg` → `/og-default.svg`. |
+| `src/app/roadmap/page.tsx` | SMS Campaigns + Route Optimization moved from `planned` → `shipped`; `id="suggest"` section replaced with `mailto:hello@routecommerce.com` link. |
+| `src/app/wholesale/login/page.tsx` | Error string "the wholesale portal is not yet wired up to it" → "Google sign-in failed. Please use email and password." |
+
+## Verification
+
+| Check | Result |
+|---|---|
+| `npx tsc --noEmit` | ✅ passes |
+| `npm run lint` | ✅ no new errors in changed files (3 pre-existing errors in `wholesale/portal/page.tsx`, `water/admin/login/page.tsx` — out of scope) |
+| `npm run build` | ✅ builds clean |
+| `grep "500+\|98%\|50K+\|2M+" src/components/landing/HeroSection.tsx src/app/page.tsx` | ✅ only audit-comment references remain |
+| `grep "Marcus T\|Sandra K\|James R" src/app/pricing/PricingClientPage.tsx` | ✅ only audit-comment reference remains |
+| `grep "SOC 2\|99.9%\|quarterly penetration" src/app/security/page.tsx` | ✅ only audit-comment references remain |
+| `grep "134100\|359100" src/lib/stripe-billing.ts` and `grep "1341\|3591" src/lib/pricing.ts` | ✅ aligned to 25% off |
+
+## Remaining unsupported / outdated promises (NOT touched — need decisions)
+
+These were in the audit and either out of scope for this pass or need a
+human decision before changing:
+
+1. **Changelog v1.9.0 "Two-Factor Authentication"** — claim exists, code does
+   not. Two paths: implement Better Auth's `twoFactor` plugin, **or** delete
+   the v1.9.0 entry. **Recommend:** delete the entry; queue 2FA as a real
+   ticket.
+
+2. **README "Time Tracking" section** — actions are stubs
+   (`src/actions/time-tracking/field.ts:7`). README still documents
+   workers, tasks, PINs as if working. **Recommend:** replace README
+   section with "Time Tracking — coming back in v0.5" and link to a real
+   ticket.
+
+3. **`LAUNCH_CHECKLIST.md` "Referral Program ✓ Complete"** —
+   `src/app/api/referrals/route.ts` uses an in-memory `Map`. **Recommend:**
+   either remove the "✓ Complete" line from `LAUNCH_CHECKLIST.md`, or wire
+   the API to Postgres.
+
+4. **`LAUNCH_CHECKLIST.md` "OnboardingFlow.tsx", "TestimonialsAndCTA.tsx",
+   "FeaturesAndStats.tsx"** — components don't exist. **Recommend:** remove
+   the references; the launch checklist is now an internal doc and is
+   misleading.
+
+5. **Roadmap "Mobile App (iOS & Android)" — In Progress** — only a PWA spec
+   exists. **Recommend:** rename to "Admin mobile (PWA)" with link to
+   `docs/superpowers/plans/2026-06-17-admin-mobile-pwa.md`.
+
+6. **Roadmap "Multi-location Support" — In Progress** — no schema. **Recommend:**
+   move to Planned.
+
+7. **"On-Time Delivery 98%" / "500+ Farm Brands"** — already removed from
+   Hero. **Recommend:** when you ship the first production brand, capture
+   real metrics and re-add them via a `get_platform_stats` RPC (would need
+   to be created and gated).
+
+8. **Add-on tile "wholesale_portal: Contact us" vs pricing card `$99/mo`** —
+   mismatch between `feature-flags.ts:59` and `pricing.ts`. **Recommend:**
+   pick one narrative — either show prices in the add-on tile or move
+   wholesale portal to enterprise-only.
+
+9. **FAQ: "Every new account starts on the Starter plan" vs waitlist
+   "30 days free on any plan"** — these contradict each other and neither
+   is true. **Recommend:** consolidate into one onboarding FAQ once the
+   signup flow is real.
+
+10. **Changelog staleness** — last entry 2025-01-15; today 2026-06-26. Add
+    a "Recent" block at top with the most recent 3–5 releases. Needs an
+    editorial pass.
+
+## Decisions the user should still make
+
+- **Changelog v1.9.0** — implement 2FA or delete the entry? (Recommend:
+  delete + queue.)
+- **README Time Tracking section** — mark coming-soon or rebuild the
+  schema? (Recommend: mark coming-soon; rebuild is multi-week.)
+- **`LAUNCH_CHECKLIST.md`** — refresh or retire? (Recommend: refresh; it's
+  cited in QA runs.)
+- **Roadmap Mobile App entry** — rename to "(PWA)" or remove until native
+  work starts?
+- **Pricing assessment doc** (`docs/pricing-assessment.md`) — recommendations
+  there (2.5–3× lift, Starter $99, Farm $349, Enterprise Custom) are not
+  applied to the marketing copy yet. That's a separate business decision.
+
+## Proven promises (no action needed)
+
+The following were audited and held up. Listed for completeness:
+
+- Harvest Reach (templates, scheduling, analytics).
+- Square Inventory Sync (bidirectional mode exists; cron wired in
+  `vercel.json`).
+- Water Log module (`/admin/water-log`, `/water`).
+- Wholesale Portal (`/wholesale/portal`).
+- 8 AI endpoints (`/api/ai/*` — gated on `getAIClient(brandId)` returning a
+  client).
+- Email automation crons (per `vercel.json`).
+- Stripe-scoped PCI / 256-bit SSL / fraud detection / tokenization.
+- Vercel-scoped edge DDoS / global CDN.
+- Tuxedo brand FAQ (real numbers, real phone 970-323-6874).
+- Indian River Direct testimonials (real customer quotes).
+- `dev_session` cookie bypass (well-controlled; NODE_ENV guard at
+  `src/lib/auth.ts:55`).
+
+## Next step
+
+A copy of this report and the inventory are committed to
+`docs/qa/audit-2026-06-26/`. The audit fixes are uncommitted on the working
+tree — review with `git diff src/` before staging. The remaining 10
+unsupported / outdated items above are queued for a follow-up session; none
+require code changes without sign-off.
@@ -0,0 +1,337 @@
+# Customer-Facing Promise Audit — Route Commerce
+
+**Date:** 2026-06-26
+**Scope:** Every promise the product makes to customers across marketing,
+documentation, public UI, and "AI answers" (admin copy that talks to the user
+in product voice). Includes pricing, security, changelog, roadmap, landing
+hero/stats, FAQ copy, product descriptions, and `LAUNCH_CHECKLIST.md`.
+
+**Method:** Walked every customer-touching surface in the repo. For each
+distinct promise, captured (a) the exact copy, (b) the file/line, and
+(c) the strongest evidence — code, schema, docs, or absence of evidence —
+underneath. Then assigned one label per the user contract.
+
+**Labels used**
+
+| Label | Meaning |
+|---|---|
+| **PROVEN** | Code/schema/docs substantiate the claim today. |
+| **PARTLY PROVEN** | Substantially true but with caveats; copy overstates one detail. |
+| **MISLEADING** | The headline reads true but the fine print is materially wrong, or the framing suggests a stronger guarantee than reality. |
+| **UNSUPPORTED** | No code, schema, audit, contract, or credential backs the claim. |
+| **OUTDATED** | Was true at one point, no longer true. |
+| **MISSING EVIDENCE** | Need a human or system to confirm before we can label. |
+
+---
+
+## 1. Landing & marketing surfaces
+
+### 1.1 Landing hero stats — `src/components/landing/HeroSection.tsx:48`
+
+```
+{ stat: "500+",  label: "Farm Brands" }
+{ stat: "98%",   label: "On-Time" }
+{ stat: "50K+",  label: "Deliveries" }
+```
+
+Plus `STAT_COUNTERS` (HeroSection.tsx:84):
+```
+500   "Produce Brands"
+50K   "Orders Delivered"
+98%   "On-Time Delivery"
+$2M+  "Weekly Sales"
+```
+
+**Evidence:** None. No customers in tree, no DB row count, no integration
+that surfaces live numbers. The pricing-assessment doc (docs/pricing-assessment.md:79)
+already flags this: *"Landing-page stats are aspirational, not real. The
+working tree has no customers; this is a marketing claim that could trip up
+B2B buyers doing due diligence. Either back it with real numbers or remove it."*
+
+**Label:** **UNSUPPORTED** (high-risk: B2B due-diligence exposure).
+**Fix rank:** #1.
+
+### 1.2 Landing feature copy — HeroSection.tsx:53–98
+
+| Promise | Evidence | Label |
+|---|---|---|
+| "Showcase seasonal produce with rich media galleries and **real-time availability** updates" | No realtime subscriptions on storefront. `supabase` is mentioned in `indian-river-direct/page.tsx` for brand lookup, but product availability is a fresh fetch. | **MISLEADING** |
+| "Optimize delivery routes with **intelligent mapping** and real-time scheduling" | `src/app/api/ai/route-optimizer/route.ts` exists but is gated on admin user + AI provider key. No live routing on the storefront. | **PARTLY PROVEN** (admin-only) |
+| "Track orders from placement through delivery with **automated status** updates" | No delivery tracking. Order status is admin-set. | **UNSUPPORTED** |
+| "Give buyers a dedicated space to browse pricing and place bulk orders" | Wholesale portal exists. | **PROVEN** |
+| "Send email and SMS campaigns about seasonal availability and new harvests" | Harvest Reach exists for admin. | **PROVEN** (admin surface) |
+| "Uncover sales trends and operational insights with **real-time dashboards**" | Reports dashboard exists; not realtime. | **PARTLY PROVEN** |
+
+### 1.3 Pricing page — `src/app/pricing/PricingClientPage.tsx`
+
+| Promise | Evidence | Label |
+|---|---|---|
+| AggregateRating in JSON-LD `4.9 / 12 reviews` (`src/app/page.tsx:55`) | Inline comment: `// Placeholder — replace with real review data once collected.` | **UNSUPPORTED** (fabricated rating published as structured data; Google could surface it). |
+| "Marcus T., Fresh Fields Farm" / "Sandra K., Pacific Produce Co-op" / "James R., Gulf Coast Distribution" testimonials (PricingClientPage.tsx:54–58) | No source. Names are made up. | **MISLEADING** (named-person testimonials with no consent) |
+| "Harvest Reach alone paid for the subscription. Our pickup rate went from 70% to 94% in two months." | No source. | **UNSUPPORTED** (specific metric, no customer) |
+| "Every new account starts on the Starter plan. … No credit card required to start." (FAQ line 24) | Waitlist page (line 103) contradicts: *"Early access members get 30 days free on any plan."* No self-serve signup exists; brands are seeded by `scripts/provision-admin.ts`. | **MISLEADING** (conflicting free-trial claims) |
+| "Enterprise customers can pay by invoice" (FAQ line 32) | `pricing.ts` prices Enterprise at $399/mo; `stripe-billing.ts` has annual=0 (Custom); the FAQ says invoice. None of these match the public pricing card. | **MISLEADING** (internal pricing model is inconsistent with public copy) |
+| "Add-ons are available on any plan. … billed proportionally when added mid-cycle." | No proportional-billing code path in `src/actions/billing/`. Stripe proration is on but the proportional UI copy implies a feature we haven't built. | **PARTLY PROVEN** |
+| Save 25% with annual (line 160) | `pricing.ts` shows: Starter annual 441 = 49×12×0.75 ✓; Farm 1341 = 149×12×0.75 ✓; Enterprise 3591 = 399×12×0.75 ✓. But `stripe-billing.ts:34` shows Farm annual = $1,522.80 (15% off). **Two sources of truth disagree.** | **MISLEADING** (internal inconsistency; one of the two is wrong) |
+| "Dedicated SLA" (Enterprise feature) | No SLA document, no `sla_*` table or RPC, no uptime monitoring integration. | **UNSUPPORTED** |
+| "Priority support" (Farm) | No support tier system, no SLA, no escalation path. | **UNSUPPORTED** |
+| Compare table "AI Intelligence Pack" Enterprise-only | Roadmap / pricing says Farm has Harvest Reach but not AI. Pricing page mirrors that. | **PROVEN** (consistent) |
+| Compare table "Multi-location Support" — *not actually in the table* (referenced on roadmap only) | n/a | n/a |
+
+### 1.4 Pricing JSON-LD — `src/app/page.tsx`
+
+```
+offers: { lowPrice: 49, highPrice: 399, ... }
+priceSpecification: [ {Starter 49}, {Farm 149}, {Enterprise 399} ]
+```
+
+**CLAUDE.md (the canonical reference) says Enterprise is "Custom" pricing.**
+**Label:** **MISLEADING**. The structured-data layer is publishing a price
+that the platform says is custom. This is the kind of thing that surfaces
+in Google search snippets and creates procurement friction.
+
+### 1.5 Waitlist — `src/app/waitlist/page.tsx`
+
+| Promise | Evidence | Label |
+|---|---|---|
+| "Join **500+** farms on the waitlist" / "**12+** States Covered" / "Q2 Launch Target" | Hard-coded copy. No waitlist row count. | **UNSUPPORTED** (three different numbers, all made up) |
+| "I've been waiting for a platform like this. The wholesale portal alone will save us hours every week." — Maria S., Sunny Acres Farm | No source. | **UNSUPPORTED** |
+| "Yes! Early access members get **30 days free on any plan**." | No billing code grants this. | **UNSUPPORTED** |
+| "Full access to all features including products, orders, stops management, and communications." | Waitlist → /api/waitlist writes a row. Does NOT auto-create a tenant. | **MISLEADING** |
+
+### 1.6 Indian River Direct landing — `src/app/indian-river-direct/page.tsx`
+
+```
+TESTIMONIALS = [
+  { name: "Linda Hurlbut", text: "I finally got my grapefruit from you today..." },
+  { name: "Phil Myers",   text: "I just wanted to comment on the citrus I received..." },
+  { name: "Bill Prue",    text: "I would just like to say how pleased we are..." }
+]
+```
+
+**Evidence:** These are real customer quotes from the brand owner's email
+archive (different file than the pricing page). **Label:** **PROVEN** for
+IRD-specific surface; isolated risk.
+
+---
+
+## 2. Pricing & billing internal consistency
+
+### 2.1 Two pricing sources of truth
+
+`src/lib/pricing.ts` (marketing UI):
+```
+Farm annual = $1,341  (25% off)
+Enterprise annual = $3,591 (25% off)
+```
+
+`src/lib/stripe-billing.ts` (Stripe checkout logic):
+```
+Farm annual = $1,522.80  (15% off)  ← src/lib/stripe-billing.ts:67
+Enterprise annual = $0  (Custom)     ← src/lib/stripe-billing.ts:82
+```
+
+**Label:** **MISLEADING** — the marketing UI cannot be trusted to match
+what the checkout will charge.
+**Fix rank:** #2 (after landing stats, since this can cost real money).
+
+### 2.2 Feature flags display
+
+`src/lib/feature-flags.ts:59` shows `wholesale_portal: "Contact us"` and
+`ai_tools: "OpenAI API required"`, but `pricing.ts` prices them at $99 and $59.
+
+**Label:** **MISLEADING** (the add-on tile says "Contact us" while the
+pricing card says "$99/mo"). Same item, two prices.
+
+### 2.3 Add-on Stripe env-var mismatch
+
+`src/actions/billing/stripe-checkout.ts:19` reads `STRIPE_PRICE_WATER_LOG`.
+`pricing.ts` and `stripe-billing.ts` agree on price. **But**
+`stripe-billing.ts:154` also defines `_annual` price IDs that are never used.
+
+**Label:** **PARTLY PROVEN** (works for monthly, but annual is
+not actually wired in any Stripe price env var the docs mention).
+**Fix rank:** lower.
+
+---
+
+## 3. Security page — `src/app/security/page.tsx`
+
+| Promise | Evidence | Label |
+|---|---|---|
+| "**SOC 2 Compliant** — Our infrastructure meets SOC 2 Type II standards for security, availability, and confidentiality." | No SOC 2 report, no auditor name, no compliance-trust URL. | **UNSUPPORTED** (high-risk; this is a regulatory/compliance claim). |
+| "We conduct **quarterly penetration tests** and annual security audits with certified third parties." | No pentest repo, no audit report, no scheduled task. | **UNSUPPORTED** |
+| "**99.9% Uptime SLA** — Enterprise plan customers receive a 99.9% uptime guarantee with 24/7 monitoring." | No SLA doc, no `sla_*` table, no monitoring integration beyond Vercel's defaults. | **UNSUPPORTED** |
+| "**GDPR & CCPA Compliant**" | `src/app/privacy-policy/page.tsx` exists but does not include DPA, SCCs, or data-residency clauses. | **PARTLY PROVEN** (privacy page exists; compliance posture unverified) |
+| "Supabase Auth provides secure, compliant authentication with 2FA support." (line 69) | Auth was migrated to **Neon Auth** (`src/lib/auth.ts`), and 2FA is **not enabled** in the Better Auth config. | **OUTDATED + UNSUPPORTED** (both — wrong product AND no 2FA) |
+| "SSL Secured" / "SOC 2 Compliant" / "GDPR Ready" / "PCI Compliant" trust badges (line 47) | Same SOC 2 / PCI issue. | **UNSUPPORTED** |
+| "Stripe — PCI-compliant payment processing" (line 178) | Stripe IS PCI DSS Level 1; we inherit it. **PROVEN** for Stripe's posture. | **PROVEN** (when scoped to Stripe's posture, not our own) |
+| "Vercel — Edge network with automatic DDoS protection and global CDN" (line 168) | Vercel does provide this. | **PROVEN** (when scoped to Vercel's posture) |
+| "Supabase — PostgreSQL database with built-in encryption and real-time capabilities" (line 173) | DB is now **Neon Postgres direct via `pg`**, not Supabase. | **OUTDATED** |
+| "256-bit SSL encryption" / "PCI DSS Level 1 compliant" / "Advanced fraud detection" / "Secure card tokenization" | All Stripe attributes, true when scoped to Stripe. | **PROVEN** (Stripe-scoped) |
+| `mailto:security@routecommerce.com` for vulnerability reports | Inbox unverifiable from repo. | **MISSING EVIDENCE** (need to confirm mailbox monitored) |
+
+---
+
+## 4. Changelog — `src/app/changelog/page.tsx`
+
+| Version | Promise | Evidence | Label |
+|---|---|---|---|
+| 2.4.0 (2025-01-15) "Harvest Reach Email Campaigns — …templates, scheduling, and analytics" | Templates, scheduling, analytics all in code. | **PROVEN** |
+| 2.3.0 "Square Inventory Sync" | Settings UI + `square_to_rc` / `rc_to_square` / `bidirectional` modes in `SquareSyncSettingsClient.tsx:443`. | **PROVEN** (UI exists; cron at `/api/square/process-queue` is wired in `vercel.json`). |
+| 2.2.2 "Dashboard now loads 40% faster" | No benchmark in tree. | **UNSUPPORTED** (specific perf claim with no measurement) |
+| 2.2.1 "Order Export Fix" | Generic bug-fix entry. | **MISSING EVIDENCE** (cannot verify without git blame) |
+| 2.2.0 "**AI Intelligence Pack** — Campaign Writer, Pricing Advisor, Demand Forecasting. Available on Enterprise plan." | All 8 AI endpoints exist (`/api/ai/*`); all gated on `getAIClient(brandId)` returning a client. Requires brand to have an AI provider key configured. | **PARTLY PROVEN** (works only when brand has configured an API key — defaults to OpenAI which is BYO key, not "included in Enterprise") |
+| 2.1.0 "Water Log Module" | Module exists at `/admin/water-log` and `/water`; schema documented in `docs/water-log.md`. | **PROVEN** |
+| 2.0.0 "New Admin Dashboard" | v2 routes under `/admin/v2/*`. | **PROVEN** |
+| 1.9.0 (2024-10-30) "Two-Factor Authentication" | **No TOTP/2FA in `src/lib/auth.ts` or `src/auth.config.ts`.** Better Auth supports a `twoFactor` plugin but it is not registered. | **UNSUPPORTED** (high-risk; this is the headline 1.9.0 feature and it doesn't exist) |
+
+**Changelog metadata issue:** the changelog's most-recent entry is dated
+2025-01-15. We're in June 2026. **Label:** **OUTDATED** as a whole
+(this page silently went stale).
+
+---
+
+## 5. Roadmap — `src/app/roadmap/page.tsx`
+
+### 5.1 "Shipped" column (page.tsx:35)
+
+| Item | Promise | Evidence | Label |
+|---|---|---|---|
+| Harvest Reach Email Campaigns | Code lives under `/admin/communications`. | **PROVEN** |
+| Square Inventory Sync | Code + queue cron. | **PROVEN** |
+| AI Intelligence Pack | 8 endpoints, brand-gated. | **PROVEN** (with the same caveat as 2.2.0 above) |
+| Water Log Module | Module lives at `/admin/water-log`. | **PROVEN** |
+
+### 5.2 "In Progress" (page.tsx:43)
+
+| Item | Promise | Evidence | Label |
+|---|---|---|---|
+| Mobile App (iOS & Android) | No `ios/`, `android/`, or React Native config. The "admin mobile PWA" spec (`docs/superpowers/plans/2026-06-17-admin-mobile-pwa.md`) is a PWA plan, **not a native app**. | **UNSUPPORTED** (a public claim of a native mobile app; reality is a PWA spec). |
+| Advanced Reporting & Analytics | Reports page exists; advanced exports not all wired. | **PARTLY PROVEN** |
+| Multi-location Support | No schema for `locations` table; `brands` has no `location_id`. | **UNSUPPORTED** |
+
+### 5.3 "Planned" (page.tsx:51)
+
+| Item | Evidence | Label |
+|---|---|---|
+| SMS Campaigns | Module already implemented at `/api/ai/stop-blast-advisor` and gated by `sms_campaigns` add-on. | **OUTDATED** (should be "Shipped", not "Planned") |
+| Route Optimization | Endpoint at `/api/ai/route-optimizer` exists. | **OUTDATED** (should be "Shipped") |
+| POS Integration (Clover, Toast) | No Clover/Toast integration in tree. | **PROVEN** (still planned) |
+| Customer Loyalty Program | No code. | **PROVEN** (still planned) |
+
+### 5.4 Roadmap affordances that don't work
+
+`docs/qa/audit-2026-06-25/ACCEPTANCE-CRITERIA.md:668` already flagged:
+`/roadmap/suggestion` form and upvotes have no handlers.
+
+**Label:** **MISLEADING** (the page implies a working vote/suggest system).
+
+---
+
+## 6. Public docs (`README.md`, `LAUNCH_CHECKLIST.md`, `docs/`)
+
+### 6.1 README.md
+
+| Promise | Evidence | Label |
+|---|---|---|
+| "Time Tracking" with PINs, Workers, Tasks (lines "Adding Workers / Adding Tasks / Notification Alerts") | `src/actions/time-tracking/field.ts:7` is **stub code** that returns `{ success: false, error: "Time tracking is not configured" }`. The TODO comment (line 7) explicitly says: *"the time-tracking feature was built on Supabase RPCs … that don't exist in the SaaS rebuild schema. The functions below are stubs … the field UI degrades gracefully (no PIN login, no entry submit)."* | **UNSUPPORTED** (high-risk; README documents a feature that does not work) |
+| "**Supabase** (Postgres + Auth + RLS)" (Tech Stack) | README still says Supabase; CLAUDE.md says we're on direct Postgres + Neon Auth. | **OUTDATED** |
+| `npm run migrate` / `supabase link --project-ref` | `supabase/push-migrations.js` exists but CLAUDE.md says only the direct `pg` path is used; Supabase CLI branch is legacy. | **OUTDATED** |
+| `dev_session` cookie bypass — described as dev-only | CLAUDE.md reinforces this. Code guard at `src/lib/auth.ts:55` returns early in production unless `PERF_TEST_AUTH=1`. | **PROVEN** (well-controlled) |
+| Email automation cron endpoints (`/api/email-automation/abandoned-cart`, `/api/email-automation/welcome-sequence`) | Endpoints + Vercel cron config in `vercel.json`. | **PROVEN** |
+
+### 6.2 `LAUNCH_CHECKLIST.md`
+
+| Claim | Evidence | Label |
+|---|---|---|
+| "**Social Proof** ✓ Complete — FeaturesAndStats.tsx" | `FeaturesAndStats.tsx` does not exist in `src/components/landing/`. | **OUTDATED** |
+| "**Testimonials** ✓ Complete — TestimonialsAndCTA.tsx" | `TestimonialsAndCTA.tsx` does not exist. | **OUTDATED** |
+| "**Guided Product Tour** ✓ Complete — OnboardingFlow.tsx" | `OnboardingFlow.tsx` does not exist. | **OUTDATED** |
+| "**Referral Program** ✓ Complete — src/app/api/referrals/route.ts, ReferralSystem.tsx" | `src/app/api/referrals/route.ts` uses **in-memory `Map<string, ReferralCode>`** that does not persist across serverless invocations. `ReferralSystem.tsx` does not exist. | **OUTDATED + UNSUPPORTED** |
+| "**Email Capture** ✓ Complete — Newsletter form in blog page" | `src/app/blog/page.tsx` does not have a working newsletter form (static post list). | **OUTDATED** |
+
+### 6.3 `docs/ADMIN_FUNCTIONALITY_CHECKLIST.md`
+
+Lists time-tracking admin features as testable. Combined with the stub
+status above, this is **OUTDATED + UNSUPPORTED**.
+
+---
+
+## 7. Public storefront / AI answers in product voice
+
+### 7.1 Wholesale portal login — `src/app/wholesale/login/page.tsx:156`
+
+> "Google sign-in is available, but the wholesale portal is not yet wired up to it. Please use email + password for now."
+
+**Label:** **PARTLY PROVEN** (honest about the limitation, but a customer-facing string saying "not yet wired up" is itself a smell — production copy should not contain TODO-like caveats).
+
+### 7.2 Brand-specific FAQs — `src/app/tuxedo/faq/FAQClientPage.tsx`
+
+Mostly verifiable (corn minimum 4 dozen, real phone number 970-323-6874).
+**Label:** **PROVEN** for Tuxedo-specific FAQ.
+
+### 7.3 `CLAUDE.md` "AI answers" surfaced in admin copy
+
+The product has 8 AI endpoints (`/api/ai/{campaign-writer, customer-insights,
+demand-forecast, pricing-advisor, product-writer, report-explainer,
+route-optimizer, stop-blast-advisor}`). Each returns a typed response. The
+admin UI at `/admin/settings/ai` describes them in product voice:
+
+> "Configure AI providers, keys, and preferences used by **campaign writer,
+> pricing advisor, report explainer**, and other tools."
+
+**Evidence:** All endpoints exist, but they require:
+1. Brand has `get_ai_provider_settings` row (BYO API key).
+2. The configured provider must be reachable from the server.
+3. The `minimax` provider is hard-coded into `AIProviderPanel.tsx` with model names like `MiniMax-M3`, `MiniMax-M2.7`, `MiniMax-M2.1` — these are **not real public model IDs** and look like leftover copy from a model-bake-off.
+
+**Label for "AI Intelligence Pack" as advertised on pricing page:**
+**PARTLY PROVEN** (works with config; "Available on Enterprise plan" copy
+implies it's bundled, but the feature requires BYO OpenAI/Anthropic key —
+the add-on has no included inference credits).
+
+**Label for `minimax` provider card:** **UNSUPPORTED** (placeholder model
+names published to admin UI).
+
+---
+
+## 8. Risk-ranked list of fixes the user must approve
+
+Top items, ranked by legal/customer-trust exposure × likelihood a real
+buyer will notice. **Production copy changes require sign-off** per the
+user contract.
+
+| Rank | Fix | Surface | Risk if unchanged | Effort |
+|---|---|---|---|---|
+| 1 | Replace fabricated landing stats ("500+ Farm Brands / 98% On-Time / 50K+ Deliveries / $2M+ Weekly Sales") with neutral copy OR real numbers once available. Remove from JSON-LD `aggregateRating` placeholder. | `src/components/landing/HeroSection.tsx`, `src/app/page.tsx` | **HIGH** — B2B procurement teams check public stats. | XS |
+| 2 | Pick one pricing source of truth. Reconcile `src/lib/pricing.ts` and `src/lib/stripe-billing.ts`. Update FAQ "Enterprise customers can pay by invoice" if pricing stays Custom. | `src/lib/pricing.ts`, `src/lib/stripe-billing.ts`, FAQ copy | **HIGH** — checkout may charge a different price than the marketing card. | S |
+| 3 | Remove "Two-Factor Authentication" from `Changelog v1.9.0` and the security page feature list, or implement TOTP via Better Auth's `twoFactor` plugin and verify it works. | `src/app/changelog/page.tsx`, `src/app/security/page.tsx` | **HIGH** — customers rely on advertised features; missing 2FA after promised = trust loss. | M |
+| 4 | Strip SOC 2 / 99.9% uptime / quarterly pentest claims from `security/page.tsx`, or replace with neutral, scoped language ("Vercel provides edge DDoS protection; Stripe is PCI DSS Level 1"). | `src/app/security/page.tsx` | **HIGH** — regulatory/compliance exposure. | S |
+| 5 | Remove or label-as-example the three fabricated testimonials on pricing page (Marcus / Sandra / James). Keep the IRD ones (real quotes). | `src/app/pricing/PricingClientPage.tsx` | **HIGH** — fake named-person testimonials can violate FTC guidelines. | XS |
+| 6 | Reconcile Enterprise pricing: structured-data JSON-LD `$399`, marketing card `$399`, CLAUDE.md "Custom". Pick one. | `src/app/page.tsx`, `src/app/pricing/PricingClientPage.tsx`, FAQ copy | **MEDIUM** — Google snippets, sales-call confusion. | S |
+| 7 | Remove or implement `dev_session`/`dev_login` claims that promise features not built (e.g., "Not yet wired up to Google" line in wholesale login). | `src/app/wholesale/login/page.tsx` | **MEDIUM** — production copy with TODO language. | XS |
+| 8 | Replace `MiniMax-M3` etc. in `AIProviderPanel.tsx` with the real `minimax` model catalog or remove the provider until shipped. | `src/components/admin/AIProviderPanel.tsx` | **MEDIUM** — admin sets a non-existent model; calls will 404. | XS |
+| 9 | Mark Time Tracking as "coming soon" in README + checklist, or rebuild the schema. | `README.md`, `docs/ADMIN_FUNCTIONALITY_CHECKLIST.md`, `src/actions/time-tracking/field.ts` | **MEDIUM** — docs promise a feature whose actions are stubs. | L |
+| 10 | Refresh changelog & roadmap dates (most recent changelog entry is 2025-01-15; today is 2026-06-26). Reclassify SMS Campaigns and Route Optimization as Shipped. | `src/app/changelog/page.tsx`, `src/app/roadmap/page.tsx` | **LOW** — perception of staleness. | S |
+| 11 | Wire the `/api/referrals` in-memory `Map` to Postgres OR remove the "Referral Program ✓ Complete" claim from `LAUNCH_CHECKLIST.md`. | `src/app/api/referrals/route.ts`, `LAUNCH_CHECKLIST.md` | **LOW** — feature currently advertised but non-functional. | M |
+| 12 | OG image: `/og-default.jpg` referenced by pricing/blog/contact but only `/og-default.svg` exists in `public/`. | `src/app/{pricing,blog,contact}/page.tsx`, `public/` | **LOW** — broken preview cards in social shares. | XS |
+| 13 | Replace Supabase-era auth claim ("Supabase Auth provides 2FA") and DB claim ("Supabase Postgres") on security page. | `src/app/security/page.tsx` | **LOW–MEDIUM** — outdated reference, but page is informational. | XS |
+| 14 | Resolve `/roadmap/suggestion` form & upvotes (no handlers) — either disable or wire up. | `src/app/roadmap/page.tsx` | **LOW** — UI affordance leads nowhere. | S |
+
+---
+
+## 9. Decisions needed from the user
+
+1. **Replace or keep** the landing hero stats? If keep, commit to a data source.
+2. **Choose pricing source of truth** — `pricing.ts` (25% annual) or `stripe-billing.ts` (15% annual, Enterprise=Custom). Recommend: align on `pricing.ts` for marketing parity, set `stripe-billing.ts` annual prices accordingly, and decide Enterprise = Custom (no card on site).
+3. **Implement 2FA now** via Better Auth's `twoFactor` plugin, **or remove** from changelog + security page. Recommend: remove from copy; queue 2FA as a real ticket.
+4. **Strip SOC 2 / uptime / pentest claims** from security page **or** scope them to providers (Vercel, Stripe). Recommend: scope to providers; queue real compliance work.
+5. **Strip fabricated named testimonials** from pricing page. Recommend: yes, replace with either no testimonials or real ones once collected.
+6. **Time Tracking**: docs claim it works, actions are stubs. Recommend: mark "coming soon" in README/checklist, file a real ticket to bring back the schema.
+7. **Roadmap & changelog**: reclassify SMS Campaigns and Route Optimization as Shipped; refresh dates.
+8. **OG image fix**: rename `og-default.svg` to `og-default.jpg`, or generate a JPG. Recommend: rename references in metadata to `/og-default.svg`.
+9. **Permission to apply any subset of the above** in this session? If yes, which?
+
+(Defaults below assume "yes, apply the safe ones; ask on anything that
+touches an invoice number.")