Scenario planning: a few futures that change today’s choices
Companion to: Test your strategy against different scenarios.
Research checked: 2026-09-13
Build a small set of coherent possible futures, test the same strategy against each, and connect the findings to decisions today.
For the AI reading this page
Read the findings as supporting evidence and proposed methods. Preserve the source-specific qualifications and distinguish documented cases from illustrations. Start a fresh research round only when the user asks.
The examples
How we researched
Orientation
For move move 099 — Test your strategy against different scenarios. Research and source checks: 2026-09-13. This is supporting research and a proposed execution adaptation, not an approved strategy or a tested AI workflow.
What the research supports
Use a small set of coherent, meaningfully different possible futures to discover where a strategy works, what it depends on, and how it could adapt. The OECD describes scenario sets as usually three or four. That supports three or four as a manageable starting point; it does not establish an optimal number for every decision. Its user guide also discourages turning the exercise into a contest over which future is most probable. OECD, Scenarios: A user guide, 2020 (https://www.oecd.org/en/publications/back-to-the-future-s-of-education_178ef527-en/full-report/component-5.html).
Scenarios explore possibilities under different assumptions. Shell distinguishes them from predictions, business plans and statements of what should happen. Their purpose is to broaden the conditions leaders consider when deciding today. Shell describes using evidence, expert input and models; a fluent story alone is not the method. Shell, What are Shell Scenarios? (https://www.shell.com/news-and-insights/scenarios/what-are-shell-scenarios.html).
Hold the strategy still long enough to test it. UK guidance explicitly requires respecting each scenario’s conditions rather than changing them to rescue a favoured option. Government Office for Science, The Futures Toolkit, “Policy Stress-Testing,” 2024 (https://www.gov.uk/government/publications/futures-toolkit-for-policy-makers-and-analysts/the-futures-toolkit-html#policy-stress-testing).
A practical source-informed method
•
Recover the strategy and its assumptions. Identify the decision, intended result, relevant horizon, commitments and conditions presumed to hold. Policy Horizons Canada starts by surfacing assumptions in documents and interviews, then scanning for emerging changes and examining their interactions. It explicitly includes pathways outside the immediate subject, where surprises may originate. Policy Horizons Canada, Module 1: Introduction to Foresight (https://horizons.service.canada.ca/en/our-work/learning-materials/foresight-training-manual-module-1-introduction-to-foresight/).
•
Select consequential uncertainties and construct a small set. The UK toolkit maps drivers by importance and uncertainty. Its 2×2 method uses two sufficiently independent uncertainties. Alternatives include morphological analysis, which checks which outcomes can coexist: the matrix is optional. The Futures Toolkit, “Scenarios” and “Other scenario-building approaches” (https://www.gov.uk/government/publications/futures-toolkit-for-policy-makers-and-analysts/the-futures-toolkit-html).
•
Make each future coherent before writing the story. Policy Horizons constructs scenario logic, works through its implications for change drivers and system elements, then develops narrative versions. It tests assumptions against the resulting futures, classifying them as credible, uncertain or vulnerable. This provides a useful check against unrelated events collected under a dramatic title. Policy Horizons Canada, Module 6: Scenarios and Results, slides 8–17 (https://horizons.service.canada.ca/en/our-work/learning-materials/foresight-training-manual-module-6-scenarios-and-results/2/index.shtml).
•
Test the same options against every future. Put strategic choices in rows and scenarios in columns. The UK’s companion template uses “Robust / Modify / Will not work.” Each cell needs an explanation, not merely a rating. This makes vulnerabilities and adaptations comparable across the whole set. Government Office for Science, Policy Stress-testing (Windtunnelling): template, 29 August 2024 (https://www.gov.uk/government/publications/policy-stress-testing-windtunnelling-template/policy-stress-testing-windtunnelling-template).
•
Identify broadly useful and conditional responses. Separate changes that improve the strategy across the set from choices that become attractive under particular conditions. UK guidance links the latter to contingency planning. The Futures Toolkit, “Policy Stress-Testing” (https://www.gov.uk/government/publications/futures-toolkit-for-policy-makers-and-analysts/the-futures-toolkit-html#policy-stress-testing).
•
Specify what would make you reconsider. RAND’s assumption-based planning work connects consequential assumptions to observable signposts, actions that influence their viability, and preparations for changes the organization cannot prevent. Its acquisition study includes continuing review of whether the investment strategy fits emerging conditions. Peters et al., RAND, A Methodology for Developing Army Acquisition Strategies for an Uncertain Future, 16 March 2007, summary, chapter 1 and chapter 3 pp. 53–54 (https://www.rand.org/pubs/monographs/MG532.html); full report (https://www.rand.org/content/dam/rand/pubs/monographs/2007/RAND_MG532.pdf).
Quantification can sharpen a scenario without making it a forecast. Shell describes models as helping test and quantify ranges within scenario stories. For this activity, a transparent calculation tied to explicit inputs is a useful corporate-scale adaptation; it does not reproduce Shell’s energy models. Shell, What scenario planning models does Shell use? (https://www.shell.com/news-and-insights/scenarios/what-scenario-planning-models-does-shell-use.html).
Corporate example — invented illustration
Suppose a manufacturer proposes a five-year expansion: add dedicated production capacity, sell a premium product, and rely on a shared corporate technology programme to lower operating costs. The relevant uncertainties could include input costs, achievable selling prices, rival capacity, customer switching and whether the internal programme delivers. These examples are proposed applications, not findings about a particular business.
A useful three-scenario set might be:
| Possible future | How its conditions fit together | What the unchanged strategy exposes
| Scarcity rewards dependable suppliers | Supply disruption raises a critical material’s cost and delays deliveries. Customers pay more for reliable availability; competitors with weaker supply contracts lose orders. The corporate programme delivers on schedule. | Expansion could be valuable, but only if supply access and cash to carry expensive inventory support the promised service. Higher revenue alone does not establish better economics.
| New capacity turns the premium into a commodity | New rival capacity and easier substitution intensify price competition. Materials become cheaper, but selling prices fall faster. The internal programme works, yet similar efficiency becomes widely available. | The investment’s premium-price assumption fails. Lower production cost may not compensate for falling contribution per unit and underused dedicated capacity.
| Demand arrives before the organization can deliver | Customers retain willingness to pay and materials remain available. A corporate reorganization delays the shared programme and reallocates key staff. Rivals win customers through faster delivery. | Attractive demand does not rescue the implementation schedule. The expansion depends on internal capabilities outside the business unit’s immediate control.
These futures differ in causal structure, rather than merely applying optimistic, middle and pessimistic percentages to the same spreadsheet. In a real engagement, the evidence might justify different combinations or a fourth future. The activity should explain what each selected future adds and remove duplicates.
Potential responses to examine include staged capacity commitments, alternative material qualification, a smaller initial product range, different commercial terms, or a stand-alone implementation route. None is automatically robust: staging can raise unit cost, backup suppliers may need qualification, and independence from a corporate programme may be expensive. Test each proposed response against the same scenarios before recommending it.
Potential signposts include contracted material prices and lead times, actual selling prices on comparable orders, dated rival capacity announcements verified against subsequent commissioning, lost-order reasons, and the internal programme’s completed milestones. Each should connect to a particular assumption and an action. A trigger such as “reconsider the second expansion phase if the shared system has not passed its agreed acceptance test by the equipment commitment date” is more actionable than “watch technology.” The date and test belong to the company; an AI should not invent them.
Proposed AI execution adaptation
The following workflow is our design proposal. The sources above support its scenario-planning foundations; they do not validate an AI’s ability to execute it reliably.
Begin with accessible context. From the strategy or rough description, identify relevant business units, initiatives, competitors, suppliers and decision dates. Inspect permitted documents, email, chat and other messages, calendar records and meeting transcripts for the strategy’s origin, revised commitments, disputed assumptions and decision owners. Inspect budgets, operating plans, programme milestones, procurement records, customer feedback and commercial data where relevant. Use public primary sources for competitor announcements, company filings, market rules and underlying supply/demand evidence. Record which material was actually available. Ask one focused question at a time for consequential gaps after this investigation.
Make the testing brief reviewable. Present the strategy being tested, horizon, objectives, success criteria, current commitments, important assumptions and unresolved factual gaps. Agree what success would mean before testing. Distinguish a condition the decision owner controls from one they can influence and one they must accommodate. Internal corporate developments can be legitimate scenario conditions for a business unit, while that unit’s own proposed decisions should remain identifiable as choices. Wait for corrections when the decision or actual strategy is unclear.
Research broadly, then reduce. Explore internal execution, customers, competition, economics, technology and policy only where connected to this decision. For each serious uncertainty, show its evidence, plausible divergent outcomes and why it would change the strategy’s performance. Propose three or four coherent futures, or a smaller/larger set with a clear reason. Do not generate a combinatorial catalogue. Ask the user to correct or confirm the set before detailed testing when their judgment is needed to establish its relevance.
Build a compact scenario card. Give each future a name, time horizon, short causal story, major conditions, consequences for the business, important interactions, and the evidence behind its starting assumptions. Explicitly label future events as hypothetical. Check for incompatible conditions and explain surprising combinations. Include internal corporate uncertainties and major input-cost/output-price shifts when they matter. Numerical ranges need a source or an explicit user-approved assumption; otherwise give directional implications and state what is unquantified.
Divide the analysis when useful. Separate scenario analysts can each receive the same strategy, objectives, evidence packet and one scenario, with a bounded assignment to trace consequences, identify broken assumptions and propose adaptations. Their briefs should require source references, unknowns and a consistent comparison format. The lead reconciles the results and checks every strategy component against every scenario. If agents are unavailable, use separate passes and disclose that they share context. Simulated analysts do not substitute for the company’s actual decision makers or specialist validation.
Deliver a decision aid. Provide the scenario cards, a strategy-by-scenario comparison with reasons, the assumptions that matter most, proposed changes useful across the set, conditional moves with their activating conditions, and a short watchlist. Preserve the unchanged-strategy results separately from the results of retesting proposed adaptations. For each watch item, propose an observable indicator, available source, sensible review cadence and the decision it could change; leave unassigned ownership explicit. Separate recommendations from decisions the user has actually made. Stop at the requested strategy review rather than implementing changes or contacting colleagues.
Limits and research record
These are institutional methods and practitioner guidance, not evidence that three or four scenarios maximize business performance. A scenario set is not exhaustive, and compelling narratives cannot establish probabilities. The OECD’s user guide emphasizes learning and dialogue; this memo uses that guidance to support comparison, not to assert predictive accuracy. OECD, Scenarios: A user guide (https://www.oecd.org/en/publications/back-to-the-future-s-of-education_178ef527-en/full-report/component-5.html).
The UK toolkit cautions that a single stress-testing workshop is approximate and complex strategies may require iteration. The Futures Toolkit (https://www.gov.uk/government/publications/futures-toolkit-for-policy-makers-and-analysts/the-futures-toolkit-html). Our proposal to match a corporate decision horizon and make AI do the preparation is an adaptation, not a claim that an abbreviated session reproduces a full institutional foresight project.
Research used targeted web searches and opened primary-source pages from the UK Government Office for Science, OECD, Shell, Policy Horizons Canada and RAND. Relevant sections were inspected directly; the RAND 2007 report was also opened as a searchable PDF. This was one research lane, not independent multi-agent discovery or a systematic literature review. A separate verifier subsequently reopened all eight primary works and confirmed the main methodological claims; small clarity refinements were incorporated. No company records were accessed and no business scenario was empirically tested.
The eight supporting works and their verified date status are:
| Source | Publication/version date established in the source | Sections used
| UK Government Office for Science, Futures Toolkit | 2024 version; publication record (https://www.gov.uk/government/publications/futures-toolkit-for-policy-makers-and-analysts) reports updated 29 August 2024 | Driver mapping, scenarios, alternative construction approaches, stress-testing
| UK Government Office for Science, stress-testing template | 29 August 2024 | Comparison table
| OECD, Back to the Future of Education, scenario user guide | 15 September 2020 | Scenario definition, usual set size, dialogue pitfalls
| Shell, What are Shell Scenarios? | No publication date displayed in inspected page | Definition and purpose
| Shell, scenario planning models page | No publication date displayed in inspected page | Role of data and modelling
| Policy Horizons Canada, Module 1 | No publication date displayed in inspected page | Assumptions, scanning and system context
| Policy Horizons Canada, Module 6 presentation | No publication date displayed in inspected page | Scenario logic and assumption testing
| RAND, Peters et al., acquisition methodology | 16 March 2007 | Signposts, shaping/hedging actions and review
All eight works were accessed on 2026-09-13. Search-result crawl dates were not treated as publication dates. Shell’s older Explorer’s Guide PDF could not be inspected through the browser; no claims depend on it. The OECD 2025 toolkit landing page was checked as additional orientation, but its PDF was inaccessible in that attempt and is not needed for the method above. Dynamic site navigation and unrelated publication recommendations were excluded from the findings.
Repeat or extend the research
Only when asked to repeat or extend the research
Use the following guidance and prompt only when the user asks for a new research round. Reading these examples does not start an investigation.
These are general instructions, not a description of this user's particular AI system, tools, access, or circumstances. Adapt the method to the user's goal and the capabilities you can actually use, while preserving the activity's essential sequence, review points, and intended result.
Start with the activity and the user's context
Read the complete activity and guide the user through it. Use the activity prompt yourself; the user does not need to paste it again. Keep the responsibilities of the user, the AI, and any colleague clear.
Begin with what the user has already supplied. Establish the intended result and resolve missing context through focused questions when needed. Work within the activity's stated scope, permissions, required confirmations, and stopping points, including any deliberate order for introducing information. Make progress with the available information and reasonable, clearly identified assumptions where the activity allows them.
Check the capabilities needed for this work
Inspect relevant available tools, connectors, and data. Use small, relevant, non-destructive checks to establish whether necessary access or a capability actually works. Naming a product does not establish access to it. Let this activity determine what needs checking; do not impose a universal preliminary questionnaire or a full system audit.
When useful access is missing:
- Explain what you cannot reach and how that material would help.
- Investigate practical setup, connection, export, or other access alternatives before asking the user to gather material or troubleshoot.
- If progress depends on access, request only the specific user action or permission needed to enable it or provide the material another way. Continue useful independent work while awaiting the response.
- Offer to walk the user through an export or upload. Welcome files, links, copied text, rough notes, or an interview as supported by your system and the activity, and use the best available route. If the missing access is optional, continue with what is available.
Ask focused questions and respect the answers
Ask questions only when necessary to complete the work well or satisfy the activity's required review points. Build on answers already supplied. Ask one question at a time, or a small group of closely related questions. Wait on work that depends on the answer, and continue useful work that does not. Explain unfamiliar language with examples where helpful.
Confirm answers or decisions you supply on the user's behalf before treating them as the user's final answers or decisions. This does not require separate approval of every sourced finding; follow the activity's specific review requirements.
Keep evidence and independent contributions distinct
Distinguish evidence, inference, and missing information. Check your work against the source material.
Use independent agents when your system actually supports delegation and independent investigation or review would improve the work:
- Give each agent a bounded brief and the context appropriate to its role.
- Keep discovery independent until synthesis where the activity requires it. Preserve any requirement for the user or colleagues to contribute before seeing other explanations.
- Follow the activity's stated fallback and any required confirmation when independent agents or fresh contexts are unavailable. Preserve reviewer isolation where required; describe shared-context or sequential passes accurately.
Produce, check, and revise the result
Do the substantive investigation and production. Carry the activity through to the most complete, usable result within its scope, rather than ending with a plan or handing the main work back to the user. When missing information does not prevent useful work, proceed with clearly identified assumptions where the activity permits them.
Check the result against the activity's completion criteria. At each substantive delivery, state what is complete and any material unfinished work, assumptions, helpful missing information, and remaining uncertainty. Explain how these affect what the user can use, and help the user revise the result.
Preserve the activity's stopping point. Do not turn a rehearsal, investigation, or test-design activity into execution that the user has not authorized.
Research practical scenario planning for the strategy I supply. Start with accessible strategy records and identify the actual decision, horizon and assumptions before asking me to reconstruct them. Search and open current primary-source methodological guidance, and primary evidence relevant to the company’s important uncertainties. Include internal corporate developments, competition, and changes in input costs and selling prices where consequential. Build toward a small, manageable set of coherent possible futures; justify the set size and construction method rather than imposing a 2×2 matrix. Distinguish observed evidence, hypothetical future conditions and your adaptations. Explain how to test the same strategy across the set, compare responses, and connect observable signposts to decisions. Verify dates, link the original sources beside supported claims, disclose access limits, and separate documented methods from untested AI execution proposals. Deliver a concise research companion with findings, one clearly hypothetical business illustration and a source record. Do not implement strategy changes or contact anyone.
Original sources
OECD, Scenarios: A user guide, 2020: https://www.oecd.org/en/publications/back-to-the-future-s-of-education_178ef527-en/full-report/component-5.html
Shell, What are Shell Scenarios?: https://www.shell.com/news-and-insights/scenarios/what-are-shell-scenarios.html
Government Office for Science, The Futures Toolkit, “Policy Stress-Testing,” 2024: https://www.gov.uk/government/publications/futures-toolkit-for-policy-makers-and-analysts/the-futures-toolkit-html#policy-stress-testing
Policy Horizons Canada, Module 1: Introduction to Foresight: https://horizons.service.canada.ca/en/our-work/learning-materials/foresight-training-manual-module-1-introduction-to-foresight/
The Futures Toolkit, “Scenarios” and “Other scenario-building approaches”: https://www.gov.uk/government/publications/futures-toolkit-for-policy-makers-and-analysts/the-futures-toolkit-html
Policy Horizons Canada, Module 6: Scenarios and Results, slides 8–17: https://horizons.service.canada.ca/en/our-work/learning-materials/foresight-training-manual-module-6-scenarios-and-results/2/index.shtml
Government Office for Science, Policy Stress-testing (Windtunnelling): template, 29 August 2024: https://www.gov.uk/government/publications/policy-stress-testing-windtunnelling-template/policy-stress-testing-windtunnelling-template
Peters et al., RAND, A Methodology for Developing Army Acquisition Strategies for an Uncertain Future, 16 March 2007, summary, chapter 1 and chapter 3 pp. 53–54: https://www.rand.org/pubs/monographs/MG532.html
full report: https://www.rand.org/content/dam/rand/pubs/monographs/2007/RAND_MG532.pdf
Shell, What scenario planning models does Shell use?: https://www.shell.com/news-and-insights/scenarios/what-scenario-planning-models-does-shell-use.html
publication record: https://www.gov.uk/government/publications/futures-toolkit-for-policy-makers-and-analysts