Every quarter, the same fire drill. Your team is buried in spreadsheets, draft scripts, and last-minute revisions. The CEO wants to see the Q&A deck by Thursday. The investor relations team is still waiting for your final revenue breakdown. And somewhere in the chaos, a single financial footnote could trigger a sell-off if it’s phrased ambiguously. The friction isn’t just time—it’s the cognitive load of juggling precision, narrative consistency, and regulatory compliance under a hard deadline.
Traditional earnings prep means endless Slack threads, version control nightmares, and late nights reconciling the CFO’s commentary with the actual 10-Q. But the real pain is that you’re not just editing numbers—you’re crafting a story that must satisfy analysts, auditors, and the SEC simultaneously. One misplaced decimal or a tone that reads too defensive can cost millions in market cap.
Claude Cowork changes this entirely. Instead of treating AI as a passive search engine or a generic text generator, you work with Claude as a collaborative analyst. You upload your financial statements, draft scripts, and historical Q&A transcripts. Claude reads everything, asks clarifying questions, and then helps you build a coherent earnings narrative—scripts, Q&A matrices, risk disclosures—in hours, not days. The key is structured prompting that mirrors how a senior analyst would brief a junior: context, constraints, reference examples, and a clear success brief.
Why Claude Cowork Beats Generic AI for Earnings Prep
Generic chatbots give you plausible-sounding text that often hallucinates financial metrics or misses industry-specific disclosure rules. Claude Cowork, when prompted correctly, operates like a domain expert. It reads your uploaded documents, cross-references your internal style guide, and produces output that aligns with your company’s tone and regulatory obligations. The difference is in the prompt architecture—you’re not asking for “a script.” You’re giving Claude a complete brief with reference files, success criteria, and a request for an execution plan before it writes a single word.
The following two prompts are built for CFOs, controllers, and financial analysts who need earnings materials that pass legal review, satisfy investor expectations, and save at least 10 hours per quarter. Copy them, adapt the bracketed placeholders, and run them in Claude Cowork with your actual files attached.
First, read these files completely before responding:
[Q2_2026_10Q_DRAFT.docx] — Our unaudited quarterly financial statements including revenue, COGS, operating expenses, and cash flow
[Q1_2026_EARNINGS_SCRIPT_FINAL.md] — The script we used last quarter, approved by legal and IR
[ANALYST_Q1_TRANSCRIPT.md] — Full transcript of Q&A from last quarter’s call, including analyst concerns about gross margin trends
[COMPANY_TONE_GUIDE.md] — Our internal style guide for external communications (tone: confident but cautious, avoid superlatives, quantify all claims)
Here is a reference for what I want to achieve:
[Upload Q1_2026_EARNINGS_SCRIPT_FINAL.md as reference]
Here’s what makes this reference work:
– Each financial metric is introduced with a 2-sentence context before the number
– Risk factors are addressed proactively in the CFO commentary, not buried in Q&A
– Every forward-looking statement includes a specific qualifier (e.g., “based on current visibility”)
– The script uses active voice and avoids passive constructions like “it is expected”
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: Full script, approximately 3,500 words, structured as Opening Remarks (CEO, 5-7 mins) and CFO Commentary (10-12 mins)
Recipient’s reaction: Analysts should feel the narrative is transparent, numbers are explained in context, and sensitive items (declining gross margin, rising R&D spend) are addressed head-on
Does NOT sound like: Overly promotional (“record-breaking,” “unprecedented”), defensive (“despite headwinds”), or vague (“we are optimistic”)
Success means: The script passes legal review with zero compliance red flags and our IR team can use it as-is for the call
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.
How to Use This Prompt Effectively
The first prompt is your heavy lifter. When you paste it into Claude Cowork, make sure you’ve actually uploaded the referenced files—the 10-Q draft, the prior script, the analyst transcript, and your tone guide. Claude will read them, then respond with clarifying questions. Expect questions like: “Do you want the CFO commentary to include year-over-year comparisons for all line items, or only material changes?” and “Should I flag any metric that deviates more than 5% from guidance?” Answer briefly, and Claude will produce an execution plan before drafting. Review that plan. If it misses something (e.g., a new debt facility), tell Claude before it writes.
The output is a complete script you can drop into your presentation deck. But the real value is that Claude’s reasoning is transparent—you can ask it to rephrase any section, add a sensitivity table, or cross-reference the script against the 10-Q footnotes. This is coworking, not outsourcing.
First, read these files completely before responding:
[Q2_2026_EARNINGS_SCRIPT_DRAFT.md] — The script generated from our previous prompt (or your own draft)
[ANALYST_Q1_TRANSCRIPT.md] — Full transcript of last quarter’s Q&A session
[INDUSTRY_REPORT_Q2.md] — Latest sector research from our sell-side analysts covering competitive trends and benchmark multiples
[LEGAL_APPROVED_MESSAGING.md] — Our pre-approved talking points for sensitive topics (litigation, guidance changes, executive departures)
Here is a reference for what I want to achieve:
[Upload ANALYST_Q1_TRANSCRIPT.md as reference]
Here’s what makes this reference work:
– Questions are categorized by domain (revenue, margins, cash flow, guidance, M&A, regulatory)
– Each answer has a 3-tier structure: headline (1 sentence), supporting detail (2-3 sentences), and a forward-looking bridge (1 sentence)
– Answers never introduce new material information not already in the script or 10-Q
– The matrix includes a “red flag” column for questions that require legal consultation before answering
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: Q&A matrix in table format (Markdown), approximately 25-30 questions, with columns: Question, Headline Answer, Full Answer, Red Flag (Y/N), Legal Reference
Recipient’s reaction: Executives should feel fully prepared—no question should surprise them, and every answer should feel natural and confident
Does NOT sound like: Evasive (“we don’t comment on that”), overly technical (jargon without explanation), or contradictory to script language
Success means: The IR team uses this matrix as the sole reference document during the call, and we receive zero analyst complaints about inconsistent messaging
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.
Practical Next Steps
Run both prompts sequentially—first the script, then the Q&A matrix. But don’t treat the output as final. Use Claude Cowork’s iterative capability: after the matrix is generated, ask Claude to cross-reference each answer against the script to ensure no contradictions. Then ask it to generate a “red flag summary” for your legal team, highlighting any question that touches on pending litigation, forward guidance, or non-GAAP reconciliations. This turns a 40-hour prep cycle into a 4-hour review cycle.
For teams that want to institutionalize this process, create a “quarterly earnings pack” folder in Claude Cowork with your tone guide, last quarter’s script, and a template for the 10-Q highlights. Each quarter, you only need to upload the new financials and analyst transcripts. The prompts stay the same. The results get faster and more consistent every time.
Published on 5 June 2026 on growwithgpt.com