Quarterly earnings call preparation is one of the most compressed, high-stakes workflows in finance. Within roughly two weeks of quarter close, a lean IR and FP&A team must reconcile reported results against guidance, draft a narrative that holds up under analyst scrutiny, anticipate thirty or more likely questions, and rehearse answers with the CEO and CFO. The inputs are scattered across board decks, 10-Q drafts, consensus models, prior call transcripts, and sell-side notes. The output is a document where a single inconsistent figure or an off-tone phrase can move the stock.
The friction is rarely analytical. It is assembly and iteration: pulling the same numbers into five different documents, rewriting the same talking points for the fourth time, and discovering during the rehearsal that no one prepared for the one question the buy side actually cares about. Version control breaks down in email threads, and the team runs out of time for the part that matters most, which is pressure-testing the message.
Claude Cowork changes the economics of this work. Because it reads full documents rather than fragments, you can load the quarter’s entire evidence base (transcripts, decks, filings, consensus data) and have Claude draft, cross-check, and stress-test the earnings materials against it. The team stops assembling and starts editing, and rehearsal becomes a genuine test of the narrative rather than a read-through of a first draft.
Two workflows worth building first
The highest-leverage use cases are the earnings script and Q&A brief, and the mock analyst session. Both benefit from the same discipline: give Claude complete context files, define what success looks like in terms of the audience’s reaction, and require a plan before execution. The prompt templates below follow that structure and are ready to adapt.
First, read these files completely before responding:
[results_summary.md] — final GAAP and non-GAAP results, segment revenue, margins, cash flow, guidance table
[prior_call_transcript.md] — full transcript of last quarter’s earnings call
[consensus_snapshot.md] — sell-side consensus revenue, EPS, and key segment estimates with revision history
[ir_style_guide.md] — house style, approved terminology, disclosure constraints, tone rules
Here is a reference for what I want to achieve:
[Upload last quarter’s final prepared remarks and Q&A brief as markdown]
Here’s what makes this reference work:
Opens with three headline numbers before any context; every guidance change is explained in one sentence with a named driver; avoids superlatives; Q&A answers lead with the direct answer, then one supporting data point, then a forward-looking close; no answer exceeds 90 seconds when read aloud.
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: Prepared remarks of 1,200–1,500 words plus a Q&A brief of 20 questions with model answers, each under 120 words
Recipient’s reaction: The CEO reads it once and approves; the buy side hears a coherent story with no surprises; the CFO feels prepared for hostile questions on [margin trajectory]
Does NOT sound like: A press release, a data dump, or last quarter’s script with numbers swapped
Success means: Zero factual inconsistencies with the results summary after cross-check, and every Q&A answer traceable to a source file
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.
Note the cross-check requirement in the success criteria. Claude Cowork will flag any figure in the draft that does not reconcile to the results summary, which catches the most common and most embarrassing error class before anyone outside the team sees the document. Treat the Q&A brief as a living file: as new sell-side notes arrive, add them to the context and ask for a delta review rather than a full rewrite.
Rehearse against a hostile analyst, not a friendly one
The second workflow is the mock session. Most teams rehearse with people who already know the story, which produces comfortable answers to comfortable questions. Claude can instead be briefed to behave like the toughest analyst covering the stock, working only from public information and your own prior disclosures, and then graded on whether the answers hold up. This is where preparation quality is actually determined.
First, read these files completely before responding:
[draft_remarks.md] — the prepared remarks and Q&A brief for this quarter
[prior_call_transcript.md] — last quarter’s transcript, including analyst follow-ups
[analyst_coverage_notes.md] — recent sell-side notes, ratings, and stated concerns
[results_summary.md] — final reported figures and guidance
Here is a reference for what I want to achieve:
[Upload a transcript excerpt showing the three toughest exchanges from a peer company’s recent call]
Here’s what makes this reference work:
Questions target the gap between guidance and results, not headline numbers; follow-ups probe the second layer when the first answer is evasive; tone is polite but persistent; the strongest questions combine two data points into one challenge.
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: 15 questions ranked by difficulty, delivered one at a time, with a critique of each answer after it is given
Recipient’s reaction: The CFO identifies at least three answers that need rewriting before the real call
Does NOT sound like: A friendly internal review or a list of generic questions
Success means: Every question is traceable to a source file, and the session surfaces at least one inconsistency in the draft remarks
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 the mock session at least twice: once early, when the draft is rough and rewriting is cheap, and once the day before the call, when the goal is fluency under pressure. After each session, ask Claude to produce a short memo listing every answer that drew a follow-up and why, then route that memo to whoever owns the relevant section of the brief.
A practical tip: keep a single context folder per quarter containing the results summary, transcript, consensus snapshot, and style guide, and reuse it across every prompt. The quality of Claude’s output tracks the completeness of that folder more than the cleverness of any individual prompt. And preserve the final approved remarks and Q&A brief as reference files for next quarter; over four cycles you will build a house style that is genuinely yours, and drafting time will fall with each call.
Published on 20 October 2026 on growwithgpt.com