Every month-end or quarter-end, finance teams face the same grinding cycle. Raw data pours in from ERP systems, CRM platforms, and operational spreadsheets. Someone has to reconcile it, structure it, and then—under relentless time pressure—craft a board pack that the CEO and directors will actually read and trust. The friction is real: hours spent formatting pivot tables, rewriting commentary to avoid jargon, and second-guessing whether the variance explanations are clear enough. One misaligned slide or a buried insight can derail an entire board discussion.
ChatGPT changes this dynamic fundamentally. Instead of treating board pack preparation as a linear, manual assembly line, you can use it as an intelligent drafting partner. Feed it your raw financial data, your prior board deck, and your commentary guidelines, and it will produce a structured narrative package—complete with executive summaries, variance analysis, and risk highlights—in minutes, not days. The key is knowing how to prompt it with enough context and structure to get board-ready output, not generic fluff. This post gives you two battle-tested prompt templates that move you from data chaos to a polished deck.
The approach works because ChatGPT excels at pattern recognition and language consistency. It can scan last quarter’s board pack, identify the tone and structure the board expects, and apply that same framework to this quarter’s raw numbers. You retain control over the numbers; the AI handles the narrative construction. The result is a board pack that feels authored by a seasoned CFO, not a junior analyst working on three hours of sleep.
Why Most AI Prompts Fail for Board Packs
The common mistake is asking ChatGPT to “write a board summary” without providing any context. The output is vague, overly optimistic, and full of placeholder language. Directors spot this immediately. To get professional output, you must give the model three things: your data, your prior reference material, and explicit success criteria. The prompts below are designed to do exactly that. They force you to upload files, define the audience, and specify what “good” looks like. Copy them, adapt the bracketed placeholders, and run them with your actual data.
First, read these files completely before responding:
[Q2_2026_actuals_vs_budget.csv] — monthly P&L, balance sheet, and cash flow data with variance columns
[Q1_2026_board_pack_final.pdf] — the previous quarter’s approved board pack with commentary and slide structure
[company_glossary.md] — definitions of internal KPIs and segment labels used in our reporting
Here is a reference for what I want to achieve:
[Upload a sample board pack from a prior quarter that was well-received by the board]
Here’s what makes this reference work:
– Each variance explanation follows a “what happened, why it happened, what we are doing about it” structure
– Commentary stays under 3 sentences per line item
– No technical accounting jargon without a plain-English translation
– Key risks are highlighted in a separate callout box on slide 2
– The tone is direct, factual, and avoids hedging language like “we believe” or “it appears”
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: Executive summary in markdown format, 500-700 words, structured as 5 sections: Revenue, Gross Margin, Operating Expenses, Cash Position, and Key Risks
Recipient’s reaction: The board should immediately grasp the top 3 drivers of variance and feel confident in management’s action plan
Does NOT sound like: A press release, a sales pitch, or an apology letter. No phrases like “despite a challenging environment” unless backed by specific data
Success means: The CEO uses this summary verbatim in the board meeting without rewriting a single paragraph
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.
This first prompt is designed for the opening narrative of your board pack. It forces ChatGPT to read your actual data and the prior approved deck, then replicate the structure and tone that already works. The key is the “reference” section—you upload a past deck that the board liked. The model extracts the patterns (short sentences, risk callouts, action-oriented language) and applies them to your new numbers. The success brief is deliberately specific: the CEO should be able to read it aloud without edits. That standard forces the AI to avoid generic filler and stick to data-driven commentary.
From Commentary to Slide-Ready Deck
Once you have the executive summary, the next step is turning that narrative into actual slide content. Many finance teams waste hours moving text from a Word doc into PowerPoint, adjusting font sizes, and checking alignment. ChatGPT can output slide-by-slide content in a structured format that you can paste directly into your deck template. The prompt below takes the executive summary and expands it into a full slide deck outline, complete with speaker notes for the CFO.
First, read these files completely before responding:
[executive_summary_final.md] — the approved board summary generated from the previous prompt
[board_deck_template.pptx] — our standard slide layout with 4 content areas per slide: header, key data point, commentary, and action item
[speaker_note_style_guide.md] — examples of concise CFO speaking notes used in past successful presentations
Here is a reference for what I want to achieve:
[Upload a 12-slide board deck from Q1 that received positive feedback from the board chair]
Here’s what makes this reference work:
– Each slide has exactly one central message, no more than 7 words in the header
– Speaker notes are 2-4 bullet points, written as complete sentences the CFO can read verbatim
– Data visualizations are referenced but not described in detail (the deck is visual, the notes are narrative)
– The final slide is always “Key Decisions Needed” with 3-5 specific asks
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: Markdown outline with 12 slides. Each slide entry includes: slide number, header, 1-2 bullet points of key data, 2-3 sentences of commentary, 1 action item, and 3-4 speaker note bullets
Recipient’s reaction: The board chair should be able to scan the outline and know exactly which decisions require discussion
Does NOT sound like: A lecture, a data dump, or a compliance report. No slide should have more than 5 bullet points
Success means: The CFO can rehearse from these speaker notes once and deliver the deck confidently without additional prep
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.
This second prompt bridges the gap between narrative and presentation. The critical element is the “speaker note style guide” file—by giving ChatGPT examples of how the CFO actually talks during board meetings, the output sounds natural and conversational, not robotic. The success criterion is practical: the CFO should be able to rehearse once and deliver. That means the notes must be short enough to remember but complete enough to avoid fumbling. The “Key Decisions Needed” slide is non-negotiable; boards hate decks that end without clear asks.
Practical Next Steps
Start small. Take last quarter’s board pack and run it through the first prompt with your current monthly data. Compare the AI-generated executive summary to what your team actually produced. You will likely find that the AI version is tighter and more consistent, but may miss one or two nuances that only a human knows (e.g., a customer relationship that explains a revenue dip). Use that gap to refine your reference files. Add a short document called “context_notes.md” that lists one-off events the board already knows about. Over two or three cycles, you will build a system where ChatGPT drafts 80% of the board pack and your team only edits the remaining 20%.
The real win is not speed—though you will cut preparation time by half. The win is consistency. Every board pack will follow the same structure, use the same tone, and highlight risks in the same place. Directors will learn where to look for key information, building trust in your reporting process. And when a director asks a tough question about a variance, you will have the speaker notes and data trail ready because the AI helped you think through the narrative beforehand. That is the difference between a board pack that gets skimmed and one that drives strategic decisions.
Published on 4 June 2026 on growwithgpt.com