# Delegation Brief — Generic Cleanup Checklist (v1)

**From:** Loom (Workflow Architect)
**To:** Ledger (primary) → handoff downstream to Slate → Riv (later)
**Date:** 2026-05-12
**Priority:** HIGHEST — addresses J2's #1 firm bottleneck (offshore bookkeeper rework rate)

---

## Background

J2 has 2 offshore bookkeepers (Philippines) doing cleanup work for ~38 of the firm's ~40 clients. Complex accrual-basis monthly close (Fuzebox, Loop, similar) stays with Jimmie. Rework rate on bookkeeper output is high — Jimmie spends significant time correcting it.

**Root cause** (discovered in Loom's intake 2026-05-12): the bookkeepers operate **without a standardized checklist**. There's a great per-client cleanup SOP from prior work (Padro Services), an aging firm SOP Handbook, and a complex-close generator (Fuzebox) — but no generic cleanup template the bookkeepers can apply to any new client. Each engagement is improvised.

This project ships a generic cleanup checklist plus the supporting workflow architecture, **gated at v1 for owner review before scaling**.

---

## Critical context (read before building)

- **ADHD design** — see [[user-jimmie-adhd]] in memory. Anything Jimmie touches must live inside the Sheet he opens daily. Any new entry must take ≤ 15 seconds.
- **Ship-speed preference** — see [[feedback-jimmie-ship-speed-preference]]. Bias toward full architecture with explicit gates over phased rollout. Flag supervision cost upfront.
- **Current PM tool: Client Hub** (NOT Financial Cents — the SOP Handbook is outdated).
- **Time tracking** — Insightful runs on the offshore team. Step-level time data can be cross-referenced later.
- **You inherit prior team work.** Sage, Maya, Finn, Lux, Lane were a prior generation that birthed you. Their artifacts (Padro SOP especially) are authoritative starting material.

---

## Input artifacts (consume these)

1. **`C:\PKA\Owner Inbox\Padro Cleanup\Padro_QBO_Cleanup_SOP.md`** — Real completed cleanup SOP for Padro Services. **MASTER PATTERN.** Inherit: Pre-Flight + 9 numbered Steps + Open Items + time estimates + CPA handoff package.
2. **`C:\PKA\Team Inbox\build_close_checklist.py`** — Fuzebox close-checklist generator (Sage-era output). Visual format pattern: color-coded Excel, status column (Done / FLAG / REVIEW), Key Issues tab. **Format inheritance only — content is for complex close, not cleanup.**
3. **Mem.ai note: "J2 Bookkeeping SOP Handbook not current"** (id `a4672b28-fea3-4f3f-83af-eaa5dd2a9261`). **Marked outdated.** Use ONLY for high-level framework (Monthly + Biweekly cycles, cash vs accrual paths, quality control concepts). **Discard tool references** — replace Financial Cents with Client Hub mentally as you read.

---

## Your task — Layer 1 (the part you build)

Produce a **generic, client-agnostic cleanup checklist template** in Markdown.

### Required structure (inherited from Padro SOP)

- **Pre-Flight section** — backup, lock prior periods, set cash/accrual basis, pull baseline reports
- **Numbered Steps in standard order:**
  1. Reconcile Cash & Credit Card (foundational — don't proceed until clean)
  2. Clear Unapplied Cash Payment Income
  3. Clear Payments to Deposit / Undeposited Funds
  4. Post Adjusting Journal Entries (orphan accounts, payroll reclass, tax offsets, loans, etc.)
  5. Merge Duplicate Accounts
  6. Reclass Account Types (move out of "Other Expense" / "Other Income" buckets)
  7. Verify Balance Sheet Items (sales tax payable, payroll liabilities, prior-year roll, retained earnings)
  8. 1099-NEC Verification
  9. Final Review & Lock (Client Data Review, re-run reports, set closing date, deliver via Client Hub)
- **Open Items / Owner Input section** — work that can't be done from inside QBO alone
- **CPA Handoff Package** — standard deliverables list
- **Time estimates per section** (Padro totaled 6–9 hours — use as benchmark)

### Required content per step

- One-paragraph description of WHAT and WHY
- Numbered sub-steps with literal QBO click paths where useful (e.g., "Banking → Reconcile → select account")
- **Common pitfalls / stop-and-fix flags** ("if rec doesn't tie, fix the rec before reclassing — don't hide problems")
- **Acceptance criteria** — "this step is done when X is true"
- **Cash vs accrual branching** wherever it matters

### Standard JE Library (companion section)

Build a reusable JE catalog covering the patterns seen across cleanups:
- Orphan account reclassification
- Contractor expenses → W-2 payroll reclass (verify with W-3)
- Payroll refund handling
- Tax holding offsets (state unemployment, SC DOR, etc.)
- Short-term loan opening balance
- Unapplied Cash sweep (presentation-only vs. live)
- Retained Earnings roll-forward

For each: standard DR/CR, standard memo, "when to use," "common variations."

### Placeholders (Slate will fill these in Layer 2)

Use `{CLIENT_NAME}`, `{PERIOD}`, `{BASIS}`, `{ENTITY_TYPE}`, `{PRIOR_YEAR_CPA}`, etc. — anything that varies per engagement.

---

## Output format

- Single Markdown file: **`C:\PKA\Team Inbox\Cleanup_Checklist_Template_v1.md`**
- Plain Markdown — no fancy formatting Slate would have to undo
- Structure must be parseable enough that Slate can templatize into a Google Sheet without restructuring

---

## Acceptance criteria (Layer 1)

Jimmie reads it and can answer:

> *"Would this be enough for one of my offshore bookkeepers to clean up a new client without me explaining anything?"*

- **Yes** → ship to Slate (Layer 2)
- **No** → revise

Estimated build: **4–6 hours Ledger time**.
Estimated review: **30–60 min Jimmie time**.

---

## Downstream layers (informational — not yours to build)

### Layer 2 — Slate templatizes into Sheets
- Ledger's Markdown → Google Sheets template
- Color coding inherited from `build_close_checklist.py` (J2 green/orange palette)
- Status column with conditional formatting (Done / FLAG / REVIEW)
- Time-per-step input column (for Insightful cross-ref)
- Each new cleanup engagement = duplicate the template

### Layer 3 — Review gate inside the sheet (Slate builds, Loom designs)
- Specific rows tagged "Jimmie Review" — bookkeepers cannot mark these Done themselves
- When Jimmie marks FLAG or REVIEW on a row, that row + reason auto-logs to a hidden "Error Log" tab
- **This replaces the separate error log proposed earlier in the session** — capture happens at review time, no extra step from Jimmie

### Layer 4 — Integration (Riv, later)
- Insightful → Sheet sync: pull time-per-step actuals from offshore tracking
- Client Hub → Sheet sync: pipe client info / status requests into the Open Items section
- Coefficient: pull live QBO trial balance into a verification tab

### Layer 5 — Adoption (Loom)
- Pilot on ONE cleanup engagement with ONE bookkeeper
- 2-week iteration window for v1 corrections
- Roll to second bookkeeper, then all engagements

---

## Validation gate

Layers 1+2 ship together → **Jimmie reviews** → Layers 3-5 unlock.

---

## Jimmie's supervision cost (explicit, per [[feedback-jimmie-ship-speed-preference]])

| Layer | Reviewer time |
|-------|---------------|
| 1 (Ledger content) | ~30-60 min |
| 2 (Slate template) | ~30 min |
| 3 (Review gate) | ~15-30 min |
| 4 (Integration) | ~15 min per service |
| 5 (Adoption) | ~30 min per week during pilot, then ~10 min/week |

**Total upfront supervision: ~2 hours.** In exchange: a complete cleanup operating system, not a series of incremental tweaks.

---

## Reviewer (each layer)

- **Layer 1:** Loom reviews first, then Jimmie
- **Layer 2:** Loom reviews first, then Jimmie
- **Layers 3-5:** Loom reviews; Jimmie sees the validation gate output only

---

## What you do NOT do here

- No per-client checklist — generic template only
- No Sheet building — that's Slate's job
- No Keeper / Karbon evaluation — premature; revisit after data exists
- No retro-fitting old engagements — start fresh from next cleanup

---

## Start signal

Begin reading the input artifacts. Produce v1 Layer 1 in one shot — don't wait for additional inputs.