“What will happen next?” can conceal two different requests. One asks for an estimate of a defined quantity. The other asks which changes could matter and how to prepare for them. Answering the second with a single forecast, or the first with a set of stories, leaves part of the decision unsupported.
Forecasting and strategic foresight are complementary practices. Their distinction is not simply short term versus long term, or quantitative versus qualitative. It is the purpose of the analysis and the kind of claim it makes.
Forecasting needs a defined target
A forecast might estimate weekly demand for a product at a specified horizon. It needs a target definition, an information cutoff, assumptions, and a way to compare predictions with later observations. Where appropriate, it should express uncertainty rather than only a point estimate.
Forecasts can be long-range and can incorporate judgment. What makes them forecasts is the attempt to estimate an outcome under stated conditions. Their usefulness depends partly on whether the conditions remain relevant and whether the estimate matches the decision horizon.
Foresight broadens the decision space
Foresight examines plausible developments, important uncertainties, and the implications of different futures. Its methods may use quantitative evidence, expert judgment, horizon scanning, and scenarios. The objective is often to challenge assumptions and identify actions that remain useful under more than one set of conditions.
A scenario is not automatically a forecast with a hidden probability. Unless probabilities have been estimated with a defensible method, a team should avoid treating the most vivid narrative as the most likely outcome.
Use the same decision to see the difference
Consider an illustrative product team planning a customer-support service. Forecasting can help estimate request volumes and staffing needs under an expected operating model. Foresight can explore what happens if customers move to a new channel, a key supplier becomes unavailable, or the service's responsibilities change.
The forecast informs a capacity plan. The scenarios expose the assumptions behind that plan and the options the team might preserve. A scenario can also become an input to a conditional forecast: what would demand look like if this particular change occurred?
Give each practice an appropriate evaluation
For forecasts, use relevant historical evaluation, meaningful error measures, and checks on the uncertainty estimates. A backtest is evidence about observed conditions, not proof of performance under every possible disruption.
For foresight, inspect the quality and diversity of evidence, internal consistency, attention to counterarguments, and connection to decisions. Did the exercise uncover an assumption worth testing? Did it identify an option or trigger that the team can actually use? These questions are more appropriate than scoring whether one scenario “won.”
Make the handoff explicit
- State the operational forecast and its main assumptions.
- Identify uncertainties that could make those assumptions fail.
- Explore plausible implications without disguising them as measured probabilities.
- Choose robust actions and conditional responses.
- Monitor evidence that would justify revisiting the plan.
The boundary between the practices is porous, which is useful. The important thing is to label the claims honestly: this is an estimate, this is an assumption, and this is a possibility we want to be ready for.
Related: forecasting around a concrete decision and designing for a longer horizon.
Conditional estimates are useful if named
A demand forecast conditioned on a scenario is an estimate under that scenario’s assumptions. It is not a forecast of the scenario’s probability. Keep the assumptions and operational response visible, and identify what evidence would invalidate them. Conversely, a quantified indicator inside a foresight exercise remains a measured indicator, not a validation of the complete future narrative. This separation lets the practices inform one another without overstating either.