The future is not a dataset waiting to be downloaded. It is partly constrained by what already exists, partly shaped by choices, partly driven by systems already in motion, and partly open to events that cannot be known in advance. Strategic foresight exists because serious decisions must often be made before certainty arrives.

Forecasting asks what is likely to happen under a model or set of assumptions. Foresight asks a wider question: what different futures are plausible, what would make them happen, what would they mean for us, and what should we do now so that our strategy remains useful across more than one future?

This is not prediction theatre. Done well, foresight is disciplined uncertainty management. It makes hidden assumptions visible, searches for emerging change, builds alternative worlds, tests strategies against them and creates signposts so plans can adapt as evidence arrives.

The foresight loop

DECISION HORIZON
→ DEFINE FOCAL QUESTION
→ MAP CURRENT SYSTEM
→ SCAN HORIZON
→ IDENTIFY DRIVERS + UNCERTAINTIES
→ SEPARATE SIGNAL FROM NOISE
→ BUILD PLAUSIBLE FUTURES
→ TEST CONSEQUENCES
→ STRESS-TEST STRATEGY
→ IDENTIFY ROBUST ACTIONS
→ SET SIGNPOSTS + TRIGGERS
→ ACT
→ MONITOR
→ UPDATE

The loop matters because foresight is not a one-off workshop. It becomes valuable when it changes how an organisation notices change and revises decisions.

1. Start with a focal question

“What will the future look like?” is too broad. A foresight exercise needs a focal question that identifies the decision, time horizon, system and stakes. For example: What conditions could make our current education strategy ineffective by 2035? How might freight networks change if energy, trade and automation evolve differently? What capabilities should a city build now if heat, demographics and technology remain uncertain?

The focal question prevents horizon scanning from becoming an endless collection of interesting news.

2. Choose the time horizon deliberately

A useful horizon is long enough for meaningful change to occur but close enough that present decisions can influence preparedness. Infrastructure may require decades. Technology portfolios may use shorter horizons. Workforce and education systems often need to think across generations because capability takes years to build.

Different horizons can coexist: near-term operational uncertainty, medium-term strategic change and long-term structural transformation.

3. Map the present before imagining the future

Foresight begins with a current-state model. Who are the actors? What resources flow? Which institutions control decisions? What dependencies exist? Which trends are already measurable? Which constraints are slow to change?

Without a current system map, scenarios can become fiction disconnected from real mechanisms.

4. Horizon scanning searches for emerging change

The UK Government’s Futures Toolkit describes horizon scanning as systematic collection of insights about emerging trends and weak signals in order to identify possible threats, risks and opportunities. The discipline is systematic because a narrow information diet produces narrow futures.

• science and technology;
• demography;
• economics and finance;
• environment and climate;
• politics and regulation;
• social values and behaviour;
• infrastructure;
• security and geopolitics;
• education and capability;
• unexpected cross-domain interactions.

5. Weak signals are clues, not forecasts

A weak signal is an early, ambiguous indication that something may be changing. It may be a new research result, unusual behaviour, regulatory experiment, niche technology, emerging vocabulary or local practice.

Most weak signals do not become dominant trends. Their value lies in widening attention. A foresight team should ask what larger change a signal could indicate and what evidence would confirm or falsify that interpretation.

6. Trends are not destiny

A trend describes observed movement. Extrapolation assumes the movement continues. Foresight asks what could accelerate, slow, reverse or transform the trend.

Population ageing may continue while retirement behaviour changes. Computing power may increase while energy constraints reshape deployment. Urbanisation may grow while remote work changes central business districts. Trends create momentum, not inevitability.

7. Megatrends operate across systems

Megatrends are broad, long-term shifts that affect many sectors: demographic transition, climate change, digitalisation, geopolitical rebalancing, urbanisation or changes in resource demand. Their importance lies in interaction. Climate affects migration, infrastructure, food, health and insurance. AI affects labour, education, security, science and governance.

The OECD’s Strategic Foresight programme treats megatrend analysis, horizon scanning, scenario planning and backcasting as complementary methods for exploring multiple plausible futures.

8. Drivers are mechanisms of change

A driver is a force capable of changing the system. Good foresight identifies mechanism, not merely topic. “Technology” is too broad. “Falling cost of high-quality automated translation” is more useful because it implies specific consequences for communication, education and services.

Drivers can be external or internal, slow or sudden, controllable or largely uncontrollable.

9. Predetermined elements are different from critical uncertainties

Some future conditions are relatively constrained. A population cohort already born will age. A rail tunnel under construction will still shape the network years later. Other conditions are deeply uncertain: regulatory response, technology adoption, geopolitical alignment or public behaviour.

Scenario design improves when it separates what is already strongly constrained from what could plausibly diverge.

10. Uncertainty has different kinds

• Risk — outcomes can be described and probabilities may be estimated.
• Ambiguity — people disagree about what the problem means.
• Deep uncertainty — models, probabilities or future structures are themselves contested.
• Ignorance — important possibilities may not yet be recognised.

Foresight is particularly valuable in the last two categories, where a single forecast creates false precision.

11. Scenarios are structured alternative futures

A scenario is a coherent description of how the future could unfold under a connected set of assumptions. It is not a prediction and should not be written merely as an optimistic, pessimistic and middle case.

Useful scenarios differ in causally meaningful ways. They show how drivers interact, how institutions respond and what consequences follow.

12. Plausible is not the same as probable

Scenario planning expands the decision space beyond the most likely future. A low-probability but high-consequence future may deserve attention if preparation is cheap or if delay would make response impossible.

The scenario should therefore state its purpose: exploration, stress test, contingency planning, innovation or strategic challenge.

13. The 2×2 scenario method is useful but not mandatory

A common method chooses two high-impact, high-uncertainty drivers and places their opposing states on two axes, creating four scenario spaces. This is memorable and forces contrast.

It can also oversimplify. Important futures may depend on more than two uncertainties. Morphological analysis, branching scenarios, archetypes, quantitative ensembles and narrative combinations can be better for other questions.

14. Scenarios need internal coherence

A scenario should not simply contain a list of dramatic events. Elements must fit. If energy prices fall sharply, industrial patterns, transport and political responses should reflect that condition. If demographic ageing accelerates, workforce, care and fiscal effects should connect.

Coherence is what turns a collection of trends into a usable world model.

15. Quantitative models and narrative scenarios can work together

Narratives can represent institutional and behavioural change that is hard to quantify. Models can test physical, economic or operational consequences under explicit assumptions. Used together, they constrain one another.

See How Models and Simulations Work: scenario assumptions should enter models visibly rather than being mistaken for observed facts.

16. Futures wheels trace second- and third-order consequences

A change rarely stops at its first effect. A futures wheel begins with an event or driver and traces direct consequences outward, then consequences of those consequences. This helps reveal system interactions.

For example, autonomous delivery does not only change drivers’ jobs. It can alter warehouse design, curb use, insurance, energy demand, urban noise, retail location and regulation.

17. Cross-impact analysis tests interactions

Drivers are not independent. Regulation can accelerate or suppress technology. Climate can reshape migration. demographic change can alter political priorities. Cross-impact analysis asks how one development changes the likelihood or consequence of another.

18. Backcasting starts from a desired or unacceptable future

Backcasting imagines a defined future state and works backwards to identify what must happen earlier. It is especially useful when incremental extrapolation is insufficient—for example, net-zero infrastructure, major education transformation or long-term resilience.

The method can also begin from an undesirable future and ask what early actions would prevent it.

19. Visioning is not enough without pathways

A vision creates direction. It becomes strategic only when translated into capabilities, milestones, dependencies and evidence. “Become future-ready” is not a pathway. “Develop these capabilities by these dates under these trigger conditions” is closer.

20. Strategy should be tested across scenarios

The purpose of scenarios is not to choose the favourite future. It is to ask how the current strategy performs in each one.

Strategy behaviour	Meaning
Robust	Performs acceptably across many plausible futures
Fragile	Works only if one assumption holds
Adaptive	Can change direction as evidence arrives
Hedged	Maintains multiple options
Contingent	Activates when a defined trigger occurs

21. Stress testing asks how the strategy breaks

Instead of asking whether a strategy works in the base case, stress testing asks what conditions cause failure. This is close to engineering. The strategy has an operating envelope. Foresight deliberately pushes beyond comfortable assumptions.

The OECD has explicitly linked strategic foresight with systemic resilience and stress testing of policy strategies under possible disruption.

22. Robust actions differ from optimal actions

An action that maximises performance in one forecast may fail badly if the forecast is wrong. A robust action sacrifices some theoretical optimum to remain useful across a wider range of conditions.

Examples include modular infrastructure, diversified suppliers, transferable skills, open standards and reversible pilots.

23. Options have strategic value

An option is the ability, not the obligation, to act later. In uncertainty, preserving options can be more valuable than committing early. Pilot programmes, reserved land, interoperable systems, flexible contracts and cross-trained staff can all preserve future choice.

24. Signposts connect scenarios back to reality

A scenario exercise becomes operational when it identifies observable signposts: measurements that would indicate the world is moving toward one pathway rather than another.

• technology cost thresholds;
• adoption rates;
• demographic milestones;
• regulatory changes;
• trade patterns;
• capacity utilisation;
• skill shortages;
• environmental indicators.

Signposts should be measurable enough that strategy reviews can use them.

25. Triggers define when the plan changes

A trigger links observation to action. If a signpost crosses a threshold, an option is activated, investment accelerates, a contingency plan begins or a strategic assumption is reopened.

Without triggers, organisations can observe change for years without changing behaviour.

26. Foresight must avoid availability bias

Recent dramatic events dominate imagination. This can produce scenarios that are vivid but narrow. A robust scan should include slow structural changes and domains outside the team’s usual expertise.

27. Groupthink can corrupt scenarios

If the same leadership team that owns the current strategy also controls the scenario process, uncomfortable futures may be edited away. Diverse participants, external challenge and independent evidence help prevent scenarios from becoming reassurance exercises.

28. Wild cards should test resilience, not entertain

A wild card is a low-probability, high-impact event or structural break. It is useful when it reveals hidden dependencies or missing resilience. It is less useful when chosen only because it sounds dramatic.

29. Black swans and unknown unknowns cannot be enumerated

Foresight cannot list every surprise. Instead, it can build general resilience: slack, redundancy, modularity, sensing, learning and authority to adapt. This is the answer to uncertainty that cannot be described in advance.

30. Indicators can become targets and distort behaviour

When signposts are tied strongly to incentives, actors may optimise the indicator rather than the underlying capability. Foresight monitoring should therefore use multiple signals and periodic interpretation, not one dashboard number.

31. Foresight and forecasting are complementary

Forecasts are useful where stable relationships and measurable probabilities exist. Foresight is useful where structural change and deep uncertainty matter. A mature strategy can use both: quantitative forecasts inside each scenario and scenarios around the boundaries of the forecast model.

32. Foresight needs a revision cycle

A scenario set becomes stale as evidence accumulates. Some uncertainties resolve. New drivers appear. Signposts change direction. The organisation should therefore set review dates and event-driven review triggers.

This connects to Strategic Adaptation: the objective can remain stable while the route changes as the world changes.

33. A practical foresight protocol

1. Define the decision and horizon.
2. Map the current system and dependencies.
3. Scan broadly for evidence of change.
4. Separate trends, weak signals, drivers and uncertainties.
5. Identify predetermined elements.
6. Rank uncertainties by impact and uncertainty.
7. Build several coherent futures.
8. Check scenario plausibility and internal consistency.
9. Test current strategy in every scenario.
10. Identify robust actions and options.
11. Specify signposts.
12. Specify triggers.
13. Assign owners for monitoring.
14. Review scenarios as evidence changes.
15. Record which decisions changed because of the exercise.

34. Foresight in education

Education is unusually exposed to long horizons. A child beginning primary school may enter adulthood into labour markets and technologies that do not yet exist in stable form. Foresight should therefore focus less on guessing exact future jobs and more on capabilities that remain valuable across futures: literacy, mathematics, scientific reasoning, communication, adaptability, ethics and the ability to learn new systems.

35. Foresight in infrastructure

Infrastructure decisions are partly irreversible and often last decades. Scenario planning helps test demand, climate, technology, land use and regulation. Flexible staging can preserve options when uncertainty is high.

36. Foresight in organisations

Organisations should ask not only what product to build next, but what capabilities could become strategic under several futures. Data governance, trusted relationships, adaptable processes, modular technology and institutional learning often retain value even when particular forecasts fail.

37. Foresight belongs beside evidence, not instead of evidence

Scenarios are structured hypotheses about possible future conditions. They should use evidence about trends and mechanisms, but they are not empirical observations of the future. Research Methods and Source Evaluation remains the evidence owner; foresight owns the disciplined exploration of uncertainty beyond what current evidence can settle.

38. The deepest foresight question

The useful question is not “Can we predict the future?” It is “Can we notice enough, imagine enough, test enough and preserve enough flexibility that the future does not arrive as a complete surprise?”

Good foresight turns uncertainty from a reason to freeze into a reason to design strategy differently.

Sources and further reading

• OECD — Strategic Foresight
• UK Government — The Futures Toolkit
• OECD — Strategic Foresight and Systemic Resilience

Continue through eduKate

• How Models and Simulations Work
• Strategy Under Uncertainty
• Strategic Adaptation
• Strategic Positioning
• How Comparative Systems Research Works
• How Statistical Inference and Uncertainty Work

Wintour House return: foresight is not the art of sounding certain about the future. It is the discipline of making uncertainty visible early enough that present decisions can remain intelligent when the world changes.