Skip to main content

workflow-ib-datapack

WHAT: Sell-side datapack assembly — 10-section VDR-ready dataset covering historical financials, KPIs, customer data, org charts, capital structure, key contracts, litigation register, IP register, headcount and compensation, and public-comp reference set; with cross-section consistency checks and version control. WHEN: Invoke at sell-side process kickoff to build the canonical dataset that feeds the CIM and VDR; when refreshing financials and KPIs at each quarter-end during a multi-quarter process; when organising buyer DD responses against an existing datapack post-LOI.

Source facts

Repository
fall-development-rob/corp_finance
Last source activity
May 11, 2026 at 14:10
Detected SKILL.md language
English
Stars
8
Forks
2

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

Showing SKILL.md

SKILL.md
Source instructions · Read-only preview
name
workflow-ib-datapack
description
WHAT: Sell-side datapack assembly — 10-section VDR-ready dataset covering historical financials, KPIs, customer data, org charts, capital structure, key contracts, litigation register, IP register, headcount and compensation, and public-comp reference set; with cross-section consistency checks and version control. WHEN: Invoke at sell-side process kickoff to build the canonical dataset that feeds the CIM and VDR; when refreshing financials and KPIs at each quarter-end during a multi-quarter process; when organising buyer DD responses against an existing datapack post-LOI.
# Datapack Builder Workflow You are a senior investment banking associate assembling the sell-side datapack. The datapack is the structured input that feeds the CIM, populates the VDR, and answers the first 80% of buyer DD requests. Every numerical artifact cites its source. Cross-section consistency is mandatory. ## Core Principles - Data-driven: all financials sourced from FMP tools or labelled as management-provided. - Internally consistent: a buyer reading two artifacts from different sections should not find contradictory numbers. - Redaction discipline: clean version (internal use) vs redacted version (buyer distribution); never mix in the same VDR. - Version control: every artifact carries `Version: vX.Y` and `Updated: YYYY-MM-DD` header. ## When to Invoke - **Sell-side kickoff**: 4-6 weeks before launch; build datapack so CIM and management presentation reference a single canonical dataset. - **Post-LOI DD support**: organise buyer DD responses against existing datapack rather than starting fresh. - **Refresh cadence**: re-cut financials and KPIs at each quarter-end during a multi-quarter process; tag the version. ## Section List (10 Sections) ### Section 1 — Historical Financials (3-year) Income statement, balance sheet, cash flow — annual for 3 years, plus latest LTM. - Call `fmp_income_statement`, `fmp_balance_sheet`, `fmp_cash_flow` with period "annual" and limit 3 (public comp datapacks). - Call `build_three_statement` to produce a clean reformatted three-statement view from raw inputs (private targets). - Quality of earnings: clearly labelled adjustments table (one-time, non-recurring, run-rate) with rationale. - Output: `datapack/01-financials/income-statement.md`, `balance-sheet.md`, `cash-flow.md`, `qoe-adjustments.md`. ### Section 2 — KPIs and Operating Metrics - Revenue by segment, geography, product, channel; volume and price decomposition. - Customer KPIs: count, ARPU, churn, NRR, LTV, CAC, payback. - Operational KPIs sector-specific: utilisation, throughput, NPS. - Monthly cadence preferred; 24-36 months of history where available. - Output: `datapack/02-kpis/INDEX.md` and per-KPI-family files. ### Section 3 — Customer and Contract Data - Top-25 customer table: name (or coded ID for redacted version), revenue, tenure, contract end date, renewal status. - Concentration analysis: top-1 / top-5 / top-10 share; flag any customer >10%. - Contract types: subscription, transactional, project-based; weighted average duration. - Output: `datapack/03-customers/top-25.md`, `concentration.md`, `contract-mix.md`. ### Section 4 — Organisational Charts - Company-wide org chart: functions, headcount per box, reporting lines. - Senior team detail: name, title, tenure, prior experience, equity stake. - Locations and headcount by site. - Output: `datapack/04-org/org-chart.md`, `senior-team.md`, `locations.md`. ### Section 5 — Capital Structure - Current debt schedule: tranche, principal, rate, maturity, covenants, security. - Equity holders: ownership table, options/RSUs outstanding, vesting profiles. - Off-balance-sheet items: leases (ASC 842), guarantees, earnouts. - Pro forma capital structure under indicative deal scenarios. - Output: `datapack/05-capital/debt-schedule.md`, `equity-cap-table.md`, `off-bs.md`. ### Section 6 — Key Contracts List - Material customer contracts (top-25 plus any with change-of-control provisions). - Supplier and vendor contracts: top-10 by spend, single-source flags. - Real estate leases: locations, rent, term, renewal options. - Licensing, distribution, JV, and partnership agreements. - Each entry: counterparty, value, term, CoC clause flag, redaction tag (clean / redact-customer / redact-pricing / restricted). - Output: `datapack/06-contracts/contracts-register.md` and per-contract abstracts. ### Section 7 — Litigation Register - Active matters: case name, jurisdiction, counterparty, claim amount, reserved amount, status, expected resolution. - Settled matters in last 36 months. - Threatened or pre-litigation matters disclosed by counsel. - Regulatory investigations and enforcement actions. - Output: `datapack/07-litigation/active.md`, `settled.md`, `regulatory.md`. ### Section 8 — IP Register - Patents (granted and pending), trademarks, copyrights, domain names. - Per-asset record: jurisdiction, registration number, filing date, status, owner, expiry. - Trade secrets and know-how schedule (categorised, not detailed). - Inbound and outbound licences. - Open-source dependencies and licence types (for software businesses). - Output: `datapack/08-ip/patents.md`, `trademarks.md`, `licences.md`, `oss-bom.md` (where applicable). ### Section 9 — Employee Headcount and Compensation - Headcount roll-forward: opening, hires, attrition, closing — by function and location, monthly for 24 months. - Compensation bands by function and seniority. - Equity programme: outstanding options/RSUs, vesting schedule, dilution profile. - Key person dependencies: roles where departure would materially impair operations. - Output: `datapack/09-people/headcount-rollforward.md`, `comp-bands.md`, `equity-program.md`, `key-persons.md`. ### Section 10 — Public-Comp Reference Set - Select 6-10 trading peers via `comps_analysis`. - Per peer: revenue, EBITDA, growth, margins, EV, trading multiples (EV/EBITDA, EV/Revenue, P/E). - Precedent transactions in sector: announce date, target, acquirer, EV, multiple, deal rationale. - Output: `datapack/10-comps/trading-comps.md`, `precedent-transactions.md`. ## MCP Tools | Tool | Purpose | |------|---------| | `fmp_income_statement` | Historical income statement (public targets) | | `fmp_balance_sheet` | Historical balance sheet (public targets) | | `fmp_cash_flow` | Historical cash flow statement (public targets) | | `build_three_statement` | Reformatted three-statement view (private targets) | | `comps_analysis` | Trading peer selection and multiples for Section 10 | ## Output Convention ``` datapack/ INDEX.md <- master index; links every artifact; version + last refresh 01-financials/ 02-kpis/ 03-customers/ 04-org/ 05-capital/ 06-contracts/ contracts-register.md abstracts/ 07-litigation/ 08-ip/ 09-people/ 10-comps/ ``` `datapack/INDEX.md` lists every artifact with: section number, artifact title, source, last refresh date, redaction tag. Previous versions retained under `datapack/_archive/<version>/` — never overwritten. ## Quality Gates - [ ] Every numerical artifact cites its source (FMP call signature, management report ID, or audited filing) - [ ] Every contract entry carries a redaction tag — no untagged commercial terms - [ ] Financial artifacts in section 01 reconcile to the same EBITDA used in section 02 KPIs and section 10 comps - [ ] Headcount in section 09 reconciles to org chart in section 04 - [ ] Capital structure in section 05 reconciles to balance-sheet debt and equity in section 01 - [ ] Every artifact has `Version: vX.Y` and `Updated: YYYY-MM-DD` header - [ ] Clean vs redacted versions separated; never mixed in the same VDR folder ## Related Skills - `workflow-ib-cim` — CIM references the datapack as its canonical data source - `workflow-ib-buyer-list` — Section 10 comps output informs the strategic buyer universe - `workflow-ib-process-letter` — data room referenced in process letter mirrors the datapack structure - `fmp-market-data` — FMP tool reference for Sections 01 and 10 - `corp-finance-tools-core` — `build_three_statement` and `comps_analysis` tool references
View on GitHub