Financial due diligence is where M&A deals are won and lost. A typical target hands over a data room containing thousands of documents: audited financials, management accounts, trial balances, AR aging reports, contracts, payroll registers, tax returns, and a quality-of-earnings (QoE) workbook that may or may not reconcile to anything. The deal team has weeks, sometimes days, to normalize EBITDA, spot revenue recognition issues, understand working capital seasonality, and quantify risks before the SPA is signed. In practice, senior analysts spend most of that window on extraction and reconciliation — copying numbers from PDFs into spreadsheets — instead of judgment.
The friction compounds because diligence is iterative. Every new document request, every management follow-up, every revised QoE model creates another round of cross-checking. Version control breaks down. Findings live in email threads. And the people best equipped to interpret the numbers — the CFO, the controller, the deal lead — are the ones furthest from the raw documents.
Claude Cowork changes the economics of this work. Because it operates directly on your files and folders, it can read the entire data room, reconcile the QoE to the trial balance, flag inconsistencies across documents, and draft the diligence findings memo — then revise it as new information arrives. The analyst stays in control of judgment calls, but the extraction, cross-referencing, and first-draft writing collapse from days into hours. The two prompt templates below show how to structure that work so the output is deal-grade, not generic.
Setting up the engagement before you prompt
Before running either prompt, organize the data room into a clean folder structure and write a short context file that captures your firm’s diligence standards: materiality thresholds, EBITDA adjustment conventions, the definition of adjusted working capital you use, and the house style for findings memos. Claude Cowork reads that file first, which keeps every output consistent across the deal team and across engagements.
First, read these files completely before responding:
[target_financials_folder] — audited financials, management accounts, and trial balances for FY[year-2] through FY[year]
[qoe_workbook.xlsx] — management’s draft quality-of-earnings model
[context.md] — my firm’s diligence standards, materiality thresholds, and EBITDA adjustment conventions
[prior_diligence_memo.md] — a prior engagement memo I consider best-in-class
Here is a reference for what I want to achieve:
[Upload prior_diligence_memo.md as the reference]
Here’s what makes this reference work:
It separates reported EBITDA from adjusted EBITDA line by line, cites the source document for every adjustment, quantifies each item in dollars and as a percentage of EBITDA, ranks issues by deal impact rather than by discovery order, and closes each section with a recommended follow-up question for management.
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: Diligence findings memo, 1,200-1,800 words, with a summary table of proposed adjustments
Recipient’s reaction: The deal lead should be able to decide within ten minutes which adjustments to challenge and which to accept
Does NOT sound like: A generic checklist or a restatement of management’s own QoE narrative
Success means: Every proposed adjustment is traceable to a source document, and no adjustment above the materiality threshold is missed
My context file contains my standards, constraints, audience. Read it fully before starting.
DO NOT start executing yet. Ask clarifying questions first.
Give me your execution plan (5 steps max) before you begin.
Run this prompt at the start of every engagement. The clarifying-questions step matters: it forces the model to confirm which periods are in scope, whether the trial balances are pre- or post-audit, and how you want stub periods handled — all of which prevent rework later.
First, read these files completely before responding:
[qoe_workbook.xlsx] — management’s quality-of-earnings model with reported revenue and working capital by month
[management_accounts_folder] — monthly management accounts for the diligence period
[trial_balances_folder] — monthly trial balances
[ar_aging_folder] — AR aging reports by customer and month
[revenue_contracts_folder] — top 20 customer contracts and any side letters
[context.md] — my firm’s working capital definitions and materiality thresholds
Here is a reference for what I want to achieve:
[Upload a prior reconciliation working paper or bridge schedule you consider best-in-class]
Here’s what makes this reference work:
Every variance is bridged month by month with the source document cited, differences are classified as timing, classification, or genuine discrepancy, and each genuine discrepancy above materiality is escalated with a quantified deal impact and a specific document request.
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: Reconciliation schedule plus a short exceptions log, 800-1,200 words
Recipient’s reaction: The controller should trust the bridges without re-performing them and know exactly which items need management follow-up
Does NOT sound like: A high-level summary that glosses over unexplained residuals
Success means: Every monthly variance above the materiality threshold is either explained or logged as an open item with an owner
My context file contains my standards, constraints, audience. Read it fully before starting.
DO NOT start executing yet. Ask clarifying questions first.
Give me your execution plan (5 steps max) before you begin.
A practical tip: after the first run, keep the exceptions log as a living file in the same folder and re-prompt Claude Cowork as new documents arrive. Each new data room drop becomes an incremental update rather than a fresh analysis, and the findings memo stays current through signing.
Start with the QoE prompt on your next live or retrospective deal and compare the output against your team’s own first draft. The gap you see in time spent — not quality — is the number that makes the case to your partners.
Published on 27 October 2026 on growwithgpt.com