Writing a business plan with Claude, step by step

A business plan is above all a clear demonstration: a problem, a solution, a market, a model and numbers that hold up. Claude doesn't replace your judgement — but it dramatically speeds up the structuring and the writing. Here is the method.

The working principle is the same at every step: you bring the substance (your project, your data, your choices), Claude brings the structure, the rewriting and the critique. Each section below follows the same format: what to do, a ready-to-copy prompt, and the mistake that most often costs points with a demanding reader.

1. Set the context once and for all

Start by giving Claude all the useful context: your activity, your target, your offer, your stage (idea, launch, growth). The more precise the brief, the better everything else will be.

Prompt: "You are my business analyst. Here is my project: [description]. Target: [customer]. Offer: [product/price]. Ask me 5 questions to clarify what's missing before drafting a business plan."

Common mistake: starting with "write me a business plan" and no context. Claude then produces a generic document that a bank or an investor spots in ten seconds. The five minutes of initial briefing determine the quality of everything else.

2. The executive summary (written last, placed first)

Write it at the end, but put it at the top. Ask for a one-page version that makes the reader want to keep going: problem, solution, traction, ask.

Prompt: "From the sections below, write a one-page executive summary, punchy and factual, for a time-pressed investor."

Common mistake: writing it first, before the sections exist. Claude cannot summarise what hasn't been written yet: it invents generalities instead. Draft the body of the document first, the summary last.

3. Problem, solution, value proposition

Get the problem stated in concrete, quantified terms, then connect your solution point by point. Ask Claude to challenge your value proposition.

Prompt: "Rephrase my value proposition into one clear sentence, then list 3 objections a sceptical customer might raise and how to answer them."

Common mistake: accepting the first rephrasing, which is always flattering. Explicitly ask for the critique ("play devil's advocate"): that is where Claude adds the most value on this section.

4. Market and competition

Claude helps you structure the analysis (segments, size, trends, competitors) from the elements you provide. Important: verify market figures yourself — never publish unsourced data.

Prompt: "Structure a market analysis from these elements: [your data]. Identify the segments, the trends and 3 competitors, and flag what I still need to source."

Common mistake: letting Claude "estimate" the size of your market. A market figure must come from a verifiable source (industry study, public statistics, customer data) — never from generation. Use Claude to structure, not to source.

5. Business model and go-to-market

Have it detail your revenue streams, your pricing and your acquisition plan. Ask for several scenarios to test robustness. If you're still hesitating over which tool to use for this kind of iterative drafting, our Claude vs ChatGPT comparison explains why structured writing is Claude's home turf.

Prompt: "Here is my business model: [revenue streams, pricing, acquisition channels]. Build three scenarios (cautious, realistic, ambitious) with, for each one, the key assumptions and the main risk."

Common mistake: presenting only an optimistic scenario. A professional reader is looking precisely for evidence that you have considered the case where things don't go as planned.

6. The financial forecast

Claude helps you set the logic (assumptions, income-statement structure, break-even points), but the final calculations must be done and checked in a spreadsheet. Use it to write the commentary that makes your numbers readable.

Prompt: "Here are my assumptions: [numbers]. List the cost items not to forget, and write the commentary explaining how break-even is reached."

Common mistake: copying generated numbers without redoing them in a spreadsheet. Claude sets the logic and writes the commentary; the final calculations are done and checked in Excel. To have your data exports analysed, see our guide Claude and your files: Excel, PDF and more.

7. Proofreading and formatting

Finish with a consistency pass: uniform tone, repetitions, promises left unfulfilled between sections.

Prompt: "Review the whole thing: flag inconsistencies between sections, repetitions and unsupported claims."

Common mistake: having it review section by section. The real inconsistencies live between the sections — a forecast number that contradicts the executive summary, a go-to-market promise missing from the action plan. Give it the whole document.

Tailoring the business plan to your reader

The same project isn't presented the same way depending on who reads it. A bank, an investor and a future business partner don't look for the same things in the same pages — and a document calibrated for one can worry the other. Once the core is written, ask Claude to adapt the document — without ever touching the numbers:

For a bank: the reader is looking for safety. Put forward the cautious scenario, repayment capacity, stable cash flows and guarantees. A sober tone and conservative assumptions reassure more than a growth story.
For an investor: the reader is looking for potential. Emphasise the addressable market, traction, the team and the scalability of the model. The ambitious scenario belongs here, provided its assumptions are made explicit.
For a future partner: the reader is looking for clarity on the joint project. Detail the split of roles, governance, the three-year vision and what each person brings. It is less a financial demonstration than a contract of trust.

Prompt: "Here is my business plan: [document]. Adapt the executive summary and the general angle for [a bank / an investor / a future business partner]: put forward what matters to this reader, without changing any figure."

The golden rule: Claude writes and structures, you validate the facts and figures. A credible business plan is built with four hands.

To sharpen your wording at every step, the prompting techniques we detail elsewhere (role, context, format, iteration) apply word for word to this exercise — see our Claude prompting techniques guide. And keep the reflex to iterate: a solid business plan is built over several passes, not in a single request.

Claude can make mistakes on factual or numerical data. Always verify market figures and the forecast before sharing.