What Are the Best AI Prompts for Business Strategy? 10 to Use

The best AI prompts for business strategy do not ask the machine to write your strategy. They ask it to test your thinking, expose your assumptions and turn a messy plan into something a team can follow.

I have watched plenty of leadership teams type "write me a five-year strategy" into ChatGPT and get back a confident, tidy document that could belong to any company in the country. That is the problem. Strategy is the set of choices that make you different, and a generic tool will hand you the average unless you give it your facts, your constraints and a job worth doing.

Below are ten prompts I use with directors and planning teams. They work in ChatGPT, Claude, Microsoft Copilot and Gemini. Each one gives the AI a role, the material to work from, the output you want and the rules it has to follow. If you want the CEO-level versions for board prep and stakeholder updates, see the CEOs page in this series.

What are the best AI prompts for business strategy?

The best AI prompts for business strategy are the ones that make the AI argue with you rather than agree with you. Use it to sharpen a SWOT, model scenarios, test pricing, red-team a plan and compress strategy into one page. Never use it as the source of the strategy itself.

Here is the difference in practice.

Write a SWOT analysis for a UK manufacturing business. Make it detailed and professional. Include strengths, weaknesses, opportunities and threats, and add some recommendations at the end so I can share it with my team. Keep it to a page or two please, and make it sound strategic and well thought through.

You are a sceptical strategy adviser to UK SMEs. Here are our last two years of results, top customers by revenue and a note on our team: [paste]. Build a SWOT using only this evidence. For every point, cite the line it came from. Mark anything you are inferring as "assumption". Then tell me the two points a competitor would most like us to ignore.

The first prompt invites a list of clichés. The second ties every claim to evidence, labels guesses and ends with a question that forces a conversation. That pattern runs through everything below. For the general mechanics, my guide on how to write prompts for AI covers the basics.

Which AI prompts help you understand where your business stands?

Start with prompts that make the AI work only from what you give it: your numbers, your customer feedback, your competitors' public material. Where it fills gaps from its own training, you want those gaps labelled.

A SWOT done properly

Act as an independent strategy adviser. Below are our management accounts summary, our three biggest customer complaints this year, and staff survey headlines: [paste]. Produce a SWOT table with no more than four points per box. Every point must quote or reference the source material. Then list the three strengths we are most likely overrating and explain why. British English, plain language, no consultancy jargon.

Why this prompt works

Requiring a source for every point stops the AI padding the boxes with generic filler.

A competitor scan from public material

You are a market analyst. I will paste the homepage copy, pricing pages and recent press releases of four competitors: [paste]. For each one, summarise who they target, what they promise, how they price and what they seem to be avoiding. Present it as a table. Then tell me where there is a gap none of them claims, and how confident you are, from low to high, with a reason.

Why this prompt works

Working from pasted material keeps it grounded, and the confidence rating flags weak conclusions.

Market sizing assumptions, not market sizing answers

I want to estimate the market for [product or service] among [type of UK customer]. Do not give me a market size figure. Instead, build a bottom-up method: list each assumption needed (number of customers, buying frequency, price, share we could win), suggest where I could find a real source for each, and give a sensible range I should test. Present it as a table with columns: assumption, range, how to verify.

Why this prompt works

It asks for the method and the sources to check, not a number the AI may invent.

How can ChatGPT and Claude help with scenario planning and pricing?

AI is good at scenario planning because it does not get attached to the base case. Give it your plan and your drivers, and it will build alternatives faster than a whiteboard session, which leaves your team free to argue about the ones that matter.

Three scenarios with early warning signs

You are helping a UK leadership team plan for uncertainty. Our plan for next year assumes: [list key assumptions, e.g. revenue growth, hiring, input costs]. Build three scenarios: base, tough and stretch. For each, describe what has changed, the effect on revenue, costs and cash in plain terms, and the decisions we would need to make. Finish with five early warning signs we should track monthly to tell which scenario we are in.

Why this prompt works

Early warning signs turn a planning exercise into something you can monitor.

A pricing review

Act as a pricing consultant for UK B2B services. Our current price list, average discount, and what customers say about value are here: [paste]. Identify where we are probably underpricing and where we are exposed. Suggest three pricing changes, from cautious to bold. For each, list the risk, which customers would push back, and how we could test it with a small group first. Do not recommend changes you cannot justify from the material.

Why this prompt works

Asking for a small test first keeps the advice practical and reversible.

Make it better

Add a line about your market position ("we are the premium option in our region" or "we compete on speed") so the suggestions match the business you actually run.

How do you red-team a business strategy with AI?

You red-team a strategy by telling the AI, explicitly, that its job is to find the holes. Most AI tools default to being encouraging. If you do not give them permission to be blunt, they will praise your plan and polish your sentences.

The hostile reviewer

You are a sharp, sceptical investor who has seen many UK SMEs fail. Read our strategy below: [paste]. Do not summarise it or praise it. Give me the five weakest points, ranked by how much damage they could do. For each, say what evidence would change your mind. Then write the single question you would ask our managing director that would be hardest to answer.

Why this prompt works

Banning praise and asking for ranked weaknesses gets past the AI's default politeness.

The assumptions audit

Here is our strategy document: [paste]. List every assumption it makes, stated or unstated, about customers, competitors, our team, costs and timing. Put them in a table with three columns: assumption, how critical it is to the plan (high, medium, low), and how we could test it within 90 days. Highlight any two assumptions that contradict each other.

Why this prompt works

Unstated assumptions are where strategies usually break, and this drags them into the open.

How do you turn a strategy into a one-page plan and OKRs with AI?

Use AI to compress and translate. A long strategy document helps nobody on a Tuesday morning. A one-page plan, clear objectives and a briefing a team leader can actually deliver are where the AI saves real time.

The one-page plan

Turn the strategy below into a one-page plan for our whole team: [paste]. Use these headings: where we are going (one sentence), why (three bullets), our three priorities this year, what we will stop doing, and how we will know it is working. Write for someone who has never seen the full document. No jargon, no slogans, under 300 words.

Why this prompt works

"What we will stop doing" forces choices, which is what most one-page plans lack.

OKRs and the annual plan

You are an operations director helping a UK business set objectives and key results. Our three priorities for the year are: [paste]. For each priority, draft one objective and three measurable key results with a baseline and target. Flag any key result that is really an activity rather than an outcome. Then suggest which team should own each one and what the first quarter milestone should be. Present as a table.

Why this prompt works

Flagging activities dressed up as outcomes is the most useful check on any OKR draft.

The team briefing

Here is our one-page plan: [paste]. Write a 10-minute briefing script a team leader in [department] can deliver to their team. Explain what the plan means for that team specifically, two things they should do differently this quarter, and three questions the team is likely to ask, with honest answers. Friendly, direct tone. British English.

Why this prompt works

It translates strategy into one team's language, which is where most plans stall.

Which AI tool is best for strategy work: ChatGPT, Claude, Copilot or Gemini?

Any of the four will run these prompts well. The better question is which one your organisation has approved and where your documents already live. Here is how I would match the tasks above.

If you are weighing up the two most common workplace tools, I compare them in ChatGPT vs Copilot for business. Copilot users will also find more examples in my list of the best Copilot prompts.

How should leaders use AI for strategy safely?

Use AI for strategy safely by keeping confidential material in approved tools, checking every fact it adds, and making sure a human owns every decision. Strategy documents are often the most sensitive papers in the business, so do not paste confidential plans, financials or personal data into an unapproved or free consumer tool.

This is the routine I suggest for a leadership team running a strategy session with AI:

1. Agree the toolUse a business account your organisation has approved, with clear settings on whether your data is used for training.
2. Prepare the materialGather the facts first: results, customer feedback, competitor pages. Remove anything personal or not needed.
3. Ask for challenge, not answersUse the SWOT, assumptions and red-team prompts before anything that generates content.
4. Verify what it addsAny number, trend or claim the AI introduces gets checked against a real source or deleted.
5. Decide as humansRecord what you decided and why. The AI informed the discussion; it did not make the call.

If your business has no rules on this yet, start with my guide on how to write an AI policy, and read how to protect business data when using AI.

Why are strategy prompts alone not enough for your team?

Prompts alone are not enough because strategy work with AI depends on judgement: knowing what to paste, when to distrust an answer and how to turn output into a decision. A list of prompts gives people a starting point. It does not give a team shared habits.

In my sessions, the prompt people reuse most is the red-team one, but only after they have seen it tear apart a plan they wrote themselves. That is the point of practising on real work. If you are thinking about how AI fits your wider plans, my guide on how to write an AI strategy for business is a good next read, and the AI Advisory and Custom Training Programme builds sessions around your own strategy material. For prompts that help managers carry the plan into day-to-day work, see the managers page, or go back to the full series on business prompts.

Get your leadership team planning with AI properly

I run practical sessions where your team uses these prompts on your own strategy, numbers and plans, with clear rules on data. In person in Oxfordshire and London, or online.

See all programmesTalk To Harry

• Basic AI Training Your whole team, prompting properly in a day.
• Prompts, Context and Skills Deep Dive A half day on Claude and ChatGPT.
• AI Capability Assessment See where your team stands first.

Harry Lang
Harry Lang founded The Oxford AI School in Witney, Oxfordshire. He trains UK businesses and teams to use AI well, writes the Marketing Unfiltered newsletter, and is the author of Learn To Fly With AI.