Free guide · 02
Scenario planning does not predict the future. It shows you which conditions matter, tests your choices against several of them, and tells you in advance what evidence would justify changing course.
The AI conversation rarely stays still for long.
One week brings an impressive technical demonstration. The next brings warnings about employment, safety or regulation. A new product launches, investors rush towards a fresh category, and another expert explains why the market has misunderstood everything.
For a founder or independent professional, the pressure to react can be exhausting.
Should you change your offer?
Learn a new skill?
Automate part of your service?
Launch an AI product?
Wait until the technology becomes more reliable?
Making a long-term decision from the latest headline is a poor planning method. Ignoring change altogether is no better.
A more useful response is to plan for several credible futures instead of betting everything on a single forecast.
That is the purpose of scenario planning. It does not claim to predict the future. It helps you identify the conditions that matter, test your choices against different outcomes and decide what evidence would justify changing course.
The following four-prompt sequence turns that process into a practical decision exercise:
Use the prompts sequentially in one conversation with an AI system that can search the web. Review each response before moving to the next stage.
“Where is AI heading?” may produce an interesting discussion, but it is unlikely to help you make a specific choice.
A better question would be:
“Should I launch an AI-assisted research service for independent estate agents within the next six months?”
This establishes an action, an audience and a timeframe.
Before starting, decide:
The quality of the final plan will depend heavily on the quality of this starting question.
The first stage is designed to prevent a persuasive headline, company announcement or dramatic forecast from controlling the entire analysis.
Complete the context section before submitting the prompt.
Help me assess the following professional or commercial decision without assuming that any single prediction about AI will become true.
My situation
Industry and market:
Role, company and revenue model:
Decision I need to make:
Desired result and target date:
Options currently being considered:
Available time, budget, skills, assets, audience, information and relationships:
Existing obligations and constraints:
Acceptable level of risk:
My current beliefs about AI and its likely effect:
Ask no more than five clarification questions. Only ask a question when the answer could materially alter the analysis. If sufficient information is already available, list your assumptions and proceed.
Conduct current research before reaching conclusions. Support every material factual statement with a direct link, publication date and source name. Give priority to official statistics, peer-reviewed research, regulatory documents, company filings, original announcements and other primary evidence. Use reputable secondary analysis when interpretation is needed or primary material is unavailable.
Do not use a benchmark, laboratory result, product demonstration, funding announcement or level of media attention as proof of customer adoption, dependable deployment or commercial value.
Do not cite material you have not opened. When credible sources disagree, describe the disagreement, evaluate the quality of the underlying evidence and explain which conclusion is better supported.
Organise the evidence into four categories:
Investigate the factors that could materially affect my decision, including:
Provide a concise map of the parts of my industry that are relevant to this decision. Explain where AI could alter costs, demand, bargaining power or the value customers expect. Exclude background information that would not influence the choice.
Return:
Distinguish between visible news events and the deeper systems influencing them. Identify the assumptions held by important stakeholders and the narratives shaping their behaviour.
Suggest three possible pairs of scenario variables. Recommend the two variables that are most consequential to my decision, genuinely unresolved and sufficiently independent to create substantially different futures.
Define the high and low end of each variable precisely.
Finish with an evidence register containing:
Stop after presenting the evidence register. Do not create the scenarios yet.
Before continuing, examine the sources yourself. Correct weak evidence, unsupported claims and any invented citations.
The second prompt uses the two chosen uncertainties to produce a four-part scenario framework.
The exercise is useful only when the scenarios would create meaningfully different business conditions. “Good”, “bad”, “moderate” and “delayed” are not sufficient.
Continue from the evidence base I have approved. Do not repeat the earlier research or replace established findings without explaining why.
Apply the same evidence standards to any new factual claims. Include direct sources and dates, and distinguish sourced information from working assumptions.
Use the two selected uncertainties to construct a four-part scenario matrix.
Give each scenario a short, memorable name. Ensure that the four scenarios represent meaningfully different operating environments, rather than one prediction occurring at four different speeds.
Use time intervals appropriate to my target date. If no target date was supplied, consider conditions at 12, 36 and 60 months.
For each scenario, describe:
Add one separate wildcard: an unlikely development with sufficiently high impact to disrupt all four scenarios.
Do not introduce numerical estimates simply to create an appearance of precision. Quantify only variables that could alter the decision. Where credible evidence allows, provide a cautious estimate, a central estimate and an upper estimate. Label every figure as either sourced or assumed.
For each scenario, identify:
Preserve material disagreements in the evidence. Do not manufacture consensus.
End with a concise comparison of the four scenarios and identify the uncertainties with the greatest influence on my choice.
Stop before recommending a strategy.
Review the output carefully. You should be able to describe observable differences between the scenarios and recognise the early signs of each one.
A plan may look attractive because it performs brilliantly under one forecast. That does not make it robust.
The third stage compares the same choices across all four environments and pays particular attention to downside, reversibility and opportunities to learn.
Use the four scenarios I have approved. Do not repeat the previous stages.
Evaluate my existing options under every scenario. If I have not supplied clear options, propose a small set that fits my resources, constraints and tolerance for risk.
Create a comparison table containing:
Do not optimise the recommendation solely for the scenario you judge most likely. Give preference to choices with a tolerable downside across all four environments.
Classify the recommended actions under five headings:
Identify the decisions that should require human approval. Separately identify the research, monitoring and preparation tasks that could be delegated to AI.
Challenge your own recommendations:
Finish with a prioritised list of actions, but do not create the 90-day schedule yet.
Stop and allow me to choose which actions I am prepared to take.
Do not approve a plan merely because it appears thorough. Remove actions that do not fit your actual capacity.
The final stage works backwards from the desired outcome.
It turns broad strategic choices into immediate actions, measurable milestones and predefined reasons to change direction.
Use only the strategy and actions I approved in the previous stage. Do not repeat the research or rebuild the scenarios.
Begin with my desired position at the target date and work backwards.
Define:
Adjust these intervals if my entire planning horizon is shorter than one year.
Produce a 90-day plan containing no more than five priorities.
For each priority, provide:
Distinguish between decisions that are easy to reverse and decisions that create lasting commitments. Where possible, place reversible actions earlier when they can generate useful information.
Next, create a monitoring dashboard with no more than eight signals. Include only signals that could change a decision.
For each signal, show:
Finally, write a decision memo of no more than 500 words covering:
End with a list of actions for the next seven days.
Do not treat the finished analysis as a one-off report.
Keep the following items together:
Review the dashboard monthly. Repeat the full exercise quarterly, or sooner when an agreed threshold is crossed.
The point is not to become indifferent to new developments. It is to decide in advance which developments deserve a response.
You cannot eliminate uncertainty from a changing market. You can, however, stop uncertainty from making every decision for you.
02 Next step
If you want this built into how your business actually decides, rather than left in a document, let's talk.