Stepthrough
Start free

ChatGPT business case prompt: a one-page draft from your facts

Reports and planningOperationsFinanceWorks in ChatGPT, Claude, Gemini and Copilot

Updated · By Robert Breen

To write a business case with ChatGPT, copy the prompt below, fill in the decision and your real numbers, and paste any notes underneath. It returns a one-page case: the problem, the options, the cost of each, the risks, a recommendation and the figures still missing. Use it when you need approval for a vendor change, a new tool or a hire and want a clean first draft your manager can react to.

The AI should not make up savings. If you do not give it a number, it must write "Number needed." You supply the figures and check the math; it organizes the argument.

Have a professional review it. This prompt helps you explain your own numbers, not give financial advice. Have your accountant or financial advisor review anything you will act on.

The prompt

Copy it into ChatGPT (or Claude, Gemini or Copilot) and replace every [BLANK] with your own details. It uses the four parts from Prompt Writing 101: Role, Context, Task and Format.

Role: You are helping [YOUR ROLE] at [COMPANY] write a one-page business case for a manager who has two minutes to read it.
Context: The decision: [DECISION]. Facts and numbers I can stand behind: [FACTS AND NUMBERS]. Use only these. Do not invent savings, costs, percentages or deadlines. If a figure is missing, write "Number needed" and list it at the end.
Task: Write the case: the problem in two sentences, the options (including doing nothing), cost and benefit of each using my numbers, the main risks, your recommendation with one reason, and what I need to confirm before I present it.
Format: Five short headed sections and one small table comparing the options. Plain words. Under 350 words. End with a list called Numbers needed.

Fill in the blanks

[YOUR ROLE]
Your job title, such as operations manager. It sets the point of view.
[COMPANY]
Your company name and what it does in a few words.
[DECISION]
One plain sentence: what you want approved and by when.
[FACTS AND NUMBERS]
Current costs, quotes, dates and results you can prove. Paste your notes here, with labels.

Example, filled in

A made-up example from the free lesson Run a Weekly Ops Review Where Every Task Has an Owner using ChatGPT. Fairline Supply Co. is the made-up office-supply distributor in the weekly ops review lesson, where a packaging vendor keeps running short of boxes. The details below are an illustration of how the operations manager might ask for approval to add a second vendor.

Role: You are helping the operations manager at Fairline Supply Co., an office-supply distributor, write a one-page business case for a manager who has two minutes to read it.
Context: The decision: approve a second packaging vendor for shipping boxes before the holiday rush. Facts and numbers I can stand behind: our current vendor was short on boxes in three of the last four weeks, 140 returns are waiting to be processed, and a second vendor has given a written quote, with the price per box in my notes. Use only these. Do not invent savings, costs, percentages or deadlines. If a figure is missing, write "Number needed" and list it at the end.
Task: Write the case: the problem in two sentences, the options (including doing nothing), cost and benefit of each using my numbers, the main risks, your recommendation with one reason, and what I need to confirm before I present it.
Format: Five short headed sections and one small table comparing the options. Plain words. Under 350 words. End with a list called Numbers needed.

What a good answer looks like

  • Three options appear: stay with one vendor, add a second vendor, or switch fully, each with only the numbers you gave.
  • "Number needed" shows up wherever you left a cost blank, such as the price per box.
  • The recommendation is one sentence with one reason, not a list of five.
  • Risks name real things from your notes, like three short weeks in four, not general business risks.

How to check it before you use it

An AI draft can sound right and still say something you never told it. Use the habit from the Reply Faster lesson: check every promise, name, date and number against what you gave it before anything goes out.

  • Recompute every total and percentage yourself. AI can write a believable number that is wrong.
  • Delete any benefit you cannot prove with your own records or a written quote.
  • Fill in the Numbers needed list before the meeting, or say plainly that they are still open.
  • Ask your finance contact to check any cost figure that will drive the decision.

Why this prompt works

Managers approve things they can understand fast. A fixed structure, with doing nothing as an honest option, makes your ask easy to judge. Forcing "Number needed" keeps the draft honest about what you still have to find out.

Practice it in the free lesson

Run a Weekly Ops Review Where Every Task Has an Owner using ChatGPT: Turn a weekly operations and vendor review into the decisions, the next steps with owners and dates, the risks to watch and a team recap, without turning a vendor's "should" into a promise. You do every step yourself in a practice copy of ChatGPT, and nothing touches your real accounts.

Start the free lesson

Also useful: Prompt Writing 101: Role, Context, Task, Format

Terms used here

Return on investment (ROI)Executive summaryGroundingHallucination

More reports and planning prompts

Frequently asked questions

How do I write a business case with ChatGPT?
Give it the decision, your real numbers and the options, tell it not to invent figures, and ask for a one-page case with a list of missing numbers. Then verify every figure before you present it.
What goes in a business case?
The problem, the options including doing nothing, the cost and benefit of each, the risks, a recommendation and what is still unknown.
Can ChatGPT calculate the savings?
It can set out the math, but you must check it. Give it the inputs and recompute the totals yourself or in a spreadsheet.

All free prompt templates · AI glossary