How-To Guide

How to Use AI for Scenario Planning in a Small Business

A practical two-hour workflow for stress-testing your next 12 months with AI — four scenarios, the numbers behind them, and the early-warning signs that tell you which one you are in.

B Biztrategy Published 4 November 2026 · 8 min read
Group of professionals engaging in a collaborative meeting with technology.

Most small business owners have one plan for next year, and it quietly assumes next year will look like this year with a bit more growth. Then a big client leaves, a supplier raises prices by 15%, or a new AI tool halves what customers will pay for one of your services — and the plan is useless by March.

Scenario planning is the fix. Large companies have done it for decades with strategy teams and expensive workshops. You can now do a credible version in an afternoon with an AI assistant, a spreadsheet, and honest numbers. This guide shows you how, step by step, with the prompts to use and the traps to avoid.

Why scenario planning beats a single forecast

A forecast asks “what will happen?” Scenario planning asks “what could happen, and what would we do?” That shift matters because the question small businesses actually struggle with is rarely prediction. It is preparation. When owners tell us “I'm worried I'm falling behind” or “what if I make the wrong choice?”, the anxiety comes from having no plan B, not from having the wrong number in cell F12.

A good scenario set gives you three things a forecast cannot:

  • Pre-made decisions. You decide in a calm afternoon what you would cut, hire or price differently, instead of deciding under pressure.
  • Early-warning signals. Each scenario comes with indicators that tell you which future is arriving, months before it shows up in your bank balance.
  • Robust moves. Some actions make sense in every scenario. Those are the ones to start now.

If you already have a monthly forecast, keep it. Scenario planning sits on top of it. Our guide to using AI for cash flow forecasting covers the base layer; this post is about what happens when the assumptions behind that forecast break.

What AI is good at here (and what it isn't)

AI assistants such as Claude, ChatGPT and Gemini are useful scenario partners because they are fast, tireless, and have no emotional attachment to your current plan. Specifically, they are good at:

  • Generating a wide list of drivers and uncertainties you had not considered.
  • Playing devil's advocate against your favourite assumption.
  • Turning a narrative (“two clients churn and rates fall 10%”) into a structured set of spreadsheet inputs.
  • Drafting the scenario write-ups so your team can actually read them.

They are poor at three things, and you should plan around them. First, they do not know your real numbers unless you give them. Second, they will happily invent market statistics that sound plausible — never let a figure the AI produced into your model without checking its source. Third, they are bad at arithmetic across long chains of assumptions. Let the AI reason about the scenarios; let a spreadsheet do the maths.

The two-hour scenario planning workflow

Block two hours, open your AI assistant of choice in a business or team plan (so your financial data is not used for training), and have last year's profit and loss summary to hand. Work through five steps.

Step 1: Give the AI a one-page business snapshot (15 minutes)

Write a short, factual brief: what you sell, to whom, revenue by service line, your three biggest customers as a percentage of revenue, headcount, monthly fixed costs, and cash in the bank expressed as months of runway. Rounded numbers are fine. Then paste it in with this prompt:

“You are a strategy adviser to a small business. Here is a one-page snapshot of the business. Before doing anything else, list the five assumptions this business most depends on that could realistically be wrong in the next 12 months. For each, say why it matters and how quickly it could change. Do not invent statistics.”

Step 2: Pick your two critical uncertainties (20 minutes)

The classic method is a two-by-two grid: choose the two uncertainties that are both high impact and genuinely unpredictable, then combine their extremes into four worlds. For most SMBs we work with, one axis is about demand (client budgets, order volumes) and the other is about how AI changes the value of what they sell.

“From the assumptions above, recommend the two that are highest impact and least predictable. Avoid anything I control directly. Then build a 2x2 grid with a short, memorable name for each quadrant and a three-sentence description of what that world looks like for my business.”

Push back on the first answer. Ask “which of these quadrants is least likely, and why?” and “what uncertainty did you leave out that would worry a sceptical bank manager?” The second round is usually better than the first.

Step 3: Turn each world into numbers (40 minutes)

This is where most AI scenario exercises fall apart: they stay as stories. Ask the AI to translate each quadrant into a short list of inputs you can drop into a spreadsheet:

“For each of the four scenarios, give me a table of changes versus my current plan: revenue by service line (% change), average price (% change), new clients per month, staff costs, and one-off costs. Keep each change to a whole percentage and explain each in one line. Mark any number that is a guess.”

Copy the table into a simple model with one column per scenario and let the spreadsheet calculate monthly profit and cash runway. Check every line yourself. If a change looks too gentle or too dramatic, change it — you know your business better than the model does.

Step 4: Decide what you would do in each world (30 minutes)

“For each scenario, list the three actions a careful owner would take in the first 90 days, the one decision they should avoid, and the cash level at which they must act. Then identify any actions that appear in all four scenarios.”

The actions that appear everywhere are your no-regret moves. Start them this month.

Step 5: Write the one-page summary (15 minutes)

Ask the AI to compress everything into a single page: the four scenarios, the numbers, the trigger points (more on those below), and the no-regret moves. Share it with your team or business partner. A plan only one person has read is not a plan.

A worked example: a 12-person marketing agency

Here is how this looks in practice, using an illustrative agency with €1.1 million in annual revenue, 40% of it from its two largest retainers, and four months of runway.

The AI flagged five fragile assumptions. The owner chose two axes: client marketing budgets (holding steady versus cut by 20% or more) and AI pressure on pricing (clients still paying for content production versus expecting AI-level prices for it). The four worlds:

  • “Business as usual” — budgets hold, pricing holds. Revenue +8%.
  • “Squeeze” — budgets hold, but content rates fall 25%. Revenue −12% unless the service mix changes.
  • “Belt-tightening” — budgets fall, pricing holds. One retainer is lost; revenue −18%.
  • “Perfect storm” — budgets fall and rates collapse. Revenue −30%, runway drops below two months by month nine.

Two no-regret moves appeared in all four: reduce dependence on the two biggest retainers to under 30% of revenue, and package strategy and measurement work as a separate, higher-margin service that is harder to replace with AI. The owner also set a hard rule: if runway drops below three months, freelance spend is paused before any hiring decision is revisited.

None of this required a consultant. It required two hours, honest inputs, and the discipline to put the numbers in a spreadsheet rather than trusting the chat window. If the “squeeze” quadrant looks uncomfortably familiar, our guide to AI-proofing your business model goes deeper on repositioning services that AI is commoditising.

Turning scenarios into trigger points

Scenarios only earn their keep if you notice which one you are in. For each scenario, agree two or three leading indicators you already track or can track cheaply, and a threshold that triggers action. For example:

  • Pipeline value below 1.5 times next quarter's target for two consecutive months.
  • Average deal size down more than 10% against the trailing six months.
  • Proposal win rate below 25% for a full quarter.
  • Days sales outstanding rising above 45 days.

Put these on one row of your monthly management report. Once a quarter, paste the latest figures back into the same AI conversation and ask: “Based on these indicators, which scenario are we drifting towards, and which of the pre-agreed actions should we trigger?” That fifteen-minute check is where the real value of the exercise lives.

Mistakes that make AI scenarios useless

  1. Letting the AI choose without context. Generic prompts produce generic futures (“a recession”, “rapid AI adoption”). Your snapshot is what makes the output specific.
  2. Accepting invented figures. If the AI says “industry budgets typically fall 18% in a downturn”, ask for the source or delete it. See our guide to preventing AI hallucinations for the checking habits that apply here too.
  3. Only modelling bad news. Include at least one upside scenario. Being unready for a big new client is also a failure — it just feels nicer.
  4. Making the scenarios too similar. If all four worlds lead to the same decisions, your axes were not uncertain enough. Go back to step 2.
  5. Filing it and forgetting it. A scenario plan without trigger points and a quarterly review is a nice document, nothing more.
  6. Pasting sensitive data into a free tool. Use a business plan with training switched off, and anonymise client names where you can.

The bottom line

Scenario planning used to be a big-company luxury. With an AI assistant it becomes a two-hour exercise any owner can run before the next quarter starts: write an honest snapshot, pick two real uncertainties, turn four worlds into spreadsheet numbers, pre-decide your responses, and set trigger points you review every quarter. You will not predict the future. You will stop being surprised by it.

If you are building a full plan around this, pair it with our walkthrough on using AI to write a business plan — the scenarios become the risk section that lenders and partners actually read.

Where does your business stand on AI?

Take the free 3-minute AI Readiness Quiz and get a personalised score with your next steps.

Take the Free Quiz →