Financial due diligence in M&A is a pressure cooker. You are staring down a 60-day exclusivity window, a data room with 15,000 unorganized files, and a target company whose financials have been reconstructed three times in the past year. The typical workflow—download, rename, scan, extract, cross-reference, flag—is manual, repetitive, and prone to error. One missed footnote about a contingent liability or a misclassified lease can cost your firm millions in post-close adjustments. The friction is not just volume; it is velocity. You need to move fast, but you cannot afford to miss anything.
Claude Cowork changes this dynamic. Instead of treating AI as a chat interface where you paste one question at a time, Cowork allows you to upload entire data room folders, set up persistent analysis rules, and run multi-step review workflows in parallel. It acts as a junior analyst who never sleeps, never overlooks a footnote, and can cross-reference 50 files at once. The result is a due diligence process that is both faster and deeper—you surface risks in hours that used to take weeks, and you have the audit trail to prove every assertion.
This post gives you two ready-to-use Cowork prompts. They are not generic ChatGPT queries. They are structured, multi-step instructions that force Claude to read your files, understand your standards, and produce outputs that a CFO can sign off on. Copy them, adapt the bracketed placeholders, and run them inside your Cowork project.
Why Prompts Fail in M&A Work
The common mistake is asking Claude a single question like “Summarize the target’s revenue recognition policy.” That works for a one-off answer, but it does not build a diligence record. You need Claude to read 40 files, apply your firm’s specific risk thresholds, flag inconsistencies across documents, and output a structured memo that maps to your internal checklist. That requires a prompt that teaches Claude your standards, not just your question. The Anatomy of a Prompt format below does exactly that.
First, read these files completely before responding:
[target_revenue_policy.pdf] — Target’s accounting policy manual for revenue recognition
[top_10_customer_contracts.pdf] — Executed contracts with the 10 largest customers by revenue
[management_earnings_presentation.pptx] — Latest quarterly deck showing revenue trends and backlog
[prior_audit_management_letter.docx] — External auditor’s findings from the last two years
Here is a reference for what I want to achieve:
I have attached a sample QoE risk memo from a prior deal that was accepted by the investment committee. It uses a traffic-light risk rating system (Red/Amber/Green), provides a specific dollar estimate of potential misstatement, and cites the exact file and page number for each finding.
Here’s what makes this reference work:
– Every risk finding is linked to a specific document reference and page number
– Revenue risks are categorized by type: recognition timing, variable consideration, contract modifications, and customer concentration
– Each finding includes a low/medium/high probability rating AND a dollar-range impact
– The tone is factual and conservative—no speculation, no unsubstantiated opinions
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: A structured risk memo, 3–5 pages, with a summary table and detailed findings section
Recipient’s reaction: The investment committee should be able to make a go/no-go decision based on revenue risk alone
Does NOT sound like: A sales pitch, an audit opinion, or generic boilerplate
Success means: The memo surfaces at least three specific revenue risks that were not already flagged in the target’s own disclosure, and each risk includes a dollar estimate
My context file contains my firm’s diligence standards, materiality threshold ($500k), and the investment committee’s preferred memo template. 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.
This prompt works because it does three things that most ad-hoc prompts miss. First, it tells Claude exactly which files to read and why each one matters. Second, it gives a concrete reference—an actual memo from a past deal—so Claude can mimic the structure, tone, and level of detail. Third, it defines success in measurable terms: at least three specific risks, each with a dollar estimate. Without those guardrails, Claude will produce a generic summary that helps no one.
Moving from Risk Assessment to Quality of Earnings
Revenue recognition is one slice. The deeper work is the Quality of Earnings (QoE) analysis, where you normalize EBITDA, adjust for non-recurring items, and test the sustainability of cash flows. This is where most analysts burn 80% of their time—rebuilding TBs, reconciling adjustments, and writing the supporting narrative. Claude Cowork can automate the cross-referencing and memo drafting, but only if your prompt is precise about what “normalized” means for your firm.
First, read these files completely before responding:
[trailing_12m_P&L.xlsx] — Target’s detailed profit and loss by month for the last 12 months
[adjusted_EBITDA_trial_balance.xlsx] — The working TB with all proposed adjustments marked by the engagement partner
[data_room_index.csv] — Full list of supporting documents with file paths and dates
[prior_QoE_memo_template.docx] — Our firm’s standard QoE memo format with sections for each adjustment type
Here is a reference for what I want to achieve:
I have attached a completed QoE memo from a 2024 deal that was praised by the investment committee for its clarity. It uses a waterfall chart showing the bridge from reported EBITDA to adjusted EBITDA, with each adjustment explained in a table that includes: adjustment description, dollar amount, supporting document reference, and rationale.
Here’s what makes this reference work:
– Each adjustment is categorized as “normalizing,” “non-recurring,” “pro forma,” or “other”
– The memo includes a sensitivity analysis showing how adjusted EBITDA changes if you vary the two largest assumptions
– All dollar amounts are cross-referenced to the exact cell or file in the working TB
– The narrative is written for a CFO—concise, quantitative, and free of marketing language
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: A QoE memo with a waterfall chart description, adjustment table, and sensitivity section; 4–6 pages
Recipient’s reaction: The committee should be able to approve the adjusted EBITDA figure without requesting additional support
Does NOT sound like: A book report, a list of questions, or an auditor’s management letter
Success means: Every adjustment in the memo matches the partner’s signed-off TB, and the memo can be inserted directly into the IM appendix
My context file contains our firm’s materiality threshold ($250k for QoE adjustments), the partner’s preferred formatting, and the IM appendix style guide. 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.
The second prompt extends the logic of the first but adds a critical layer: reconciliation to a signed-off trial balance. In M&A, the QoE memo is not a draft—it is a deliverable that goes into the information memorandum and ultimately the purchase agreement. The prompt forces Claude to verify that every adjustment in the memo matches the partner’s final numbers. That is the difference between a helpful draft and a production-ready document.
Practical Tip: Build a Cowork Project, Not a One-Off Chat
The real power of Claude Cowork for financial due diligence comes from persistence. Do not run these prompts as isolated conversations. Instead, create a Cowork project named “Deal Name – Financial Diligence.” Upload your firm’s materiality thresholds, memo templates, and style guides as context files. Then run each prompt as a new task inside that project. Claude will remember your standards across all tasks, so the second prompt automatically knows your materiality threshold is $250k for QoE adjustments. This reduces your setup time for each new deal from hours to minutes.
Try this on your next live deal. Start with the revenue risk assessment prompt, review the output, then run the QoE prompt. You will likely find that Claude surfaces risks you missed in your first pass—not because you are careless, but because no human can cross-reference 15,000 files at the same speed. That is the edge Cowork gives you: speed without sacrificing depth.
Published on 8 July 2026 on growwithgpt.com