Atlas/Planning & Forecasting/Scenario planning

Scenario planning

Definition

Scenario planning means planning several possible futures rather than one. Instead of betting on a single forecast, it builds a small number of coherent scenarios, best, base, worst, event-driven, quantifies them and looks for a decision that stays good whichever one comes true.

Why it matters

Prepare, rather than predict.

A forecast, however carefully made, is still a bet on a single future. Yet the heaviest decisions, opening capacity, securing supply, committing stock, are taken precisely when the future is most uncertain. Scenario planning shifts the stance: it does not try to guess the future, but to prepare for it.

The principle is to replace a number with a fan. You identify the few variables that weigh most and whose uncertainty is highest, then derive a small number of coherent futures, each told as a credible story and translated into quantified assumptions. Not “costs go up”, but why, by how much, and with what knock-on effects.

The point is not to multiply futures, but to inform the decision. By testing choices against several scenarios, you see which hold everywhere and which only work in one world. You then look for the robust decision, the one that stays acceptable even in the worst case, rather than the one optimal in the single scenario you hope for.

What practice says

Operations research distinguishes two broad families. The stochastic approach assigns a probability to each scenario and optimises the expectation; the robust approach focuses on the worst cases and seeks a solution that holds across the whole uncertainty set. Both meet on one point: decide before uncertainty resolves, then adjust as it resolves.

The expert lesson is not to find the right scenario, but the right decision. A good scenario exercise is judged by its ability to reveal trade-offs and tail risks: stockout probability on critical references, worst-case delay, dependence on a constrained supplier. Scenario planning without a robust decision produces pretty stories with no operational value.

The mechanism

One present, several quantified futures.

Scenario planning follows a constant logic: isolate the uncertainties that matter, derive a few coherent futures, quantify them, then choose a robust decision across the fan.

It all starts with sorting uncertainties. Among all the factors that can vary, keep only those that combine high uncertainty and high impact on the decision: a demand, a material cost, a lead time, a regulatory shock. The rest are frozen at their expected value; otherwise the fan becomes unmanageable.

Next you build a small number of scenarios, rarely more than a handful. Each is a complete, coherent story, of the best, base and worst kind, or centred on a specific event such as a supply disruption. The schematic illustrates this fan: from one present, trajectories diverge, and the cone materialises the uncertainty that opens with the horizon.

Each scenario is then quantified and weighted, and from it you draw not a forecast but a decision. You can optimise the expectation, but you look above all at robustness: which decision stays acceptable across all scenarios, what the regret of being wrong would be. The right answer is often less ambitious and more resilient than the one the central scenario alone would have dictated.

Best · p 0.2Base · p 0.5Worst · p 0.3todaycone of uncertainty
Best scenarioBase scenarioWorst scenario

Figure 1. From one present, scenario planning traces several coherent futures, each with a probability. The cone materialises the uncertainty that opens with the horizon; the chosen decision aims to stay good across all scenarios. Illustrative schematic.

E[ x ] = Σs ps · xs ,   Σs ps = 1
Rather than a single forecast, scenario planning describes a small number of futures s, each with a probability ps and an outcome xs (demand, cost, margin). The expectation E[x] weights these outcomes, but the point lies elsewhere: you seek a robust decision that stays acceptable even in the worst scenario, and you track risk measures such as stockout probability or worst-case delay. Scenario planning is not predicting the future; it is preparing for several.

The traps

Three scenario-planning errors.

Too many scenarios, stories without numbers, or a fan that changes no decision: three ways to waste the exercise.

01

Too many scenarios

Multiplying futures does not make you clearer, only paralysed. Beyond a handful, scenarios look alike and dilute the decision. Three contrasted, quantified futures beat ten vague variants: value is in the contrast, not the count.

02

Stories without numbers

A scenario that stays literary is useless. Until “demand falls” is translated into quantified assumptions and effects on stock, cost and service, you have not planned scenarios, you have told a tale. The discipline is to convert each story into comparable numbers.

03

Planning without deciding

The ultimate trap is to produce a fine fan then fall back on the central plan, as if the exercise never happened. Scenario planning is only worth it if it changes a decision: coverage, capacity, contract, source. Otherwise it is one more report.

The rollout

Four steps to useful scenario planning.

A scenario exercise follows a progression where sorting uncertainties and the final decision matter more than the number of scenarios.

Choose the uncertainties that matter

List the factors, then keep only those with high uncertainty and high impact on the decision at hand. This sorting, more than the number of scenarios, makes the quality of the exercise.

Build a few coherent futures

Build three to five contrasted scenarios, each told as a credible story then translated into quantified assumptions. Naming each scenario helps to handle and communicate it.

Quantify and weight

Translate each scenario into comparable outcomes (demand, cost, margin, service) and, where possible, assign it a probability. Add tail-risk measures: stockout probability, worst-case delay, exposure to a supplier.

Choose a robust decision

Compare decisions across the fan and keep the one that stays acceptable everywhere, not just optimal in the hoped-for scenario. Link the result to short-term execution (S&OE) to adjust as uncertainty resolves.

Neighboring concepts

Read next.

Scenario planning extends probabilistic forecasting, feeds the load plan and loops back through execution.

From knowledge to action

One plan for an uncertain future, is that reasonable?

Rather than a single forecast, scenario planning prepares several futures and secures the decision. Our Planning, Forecasting & S&OP file equips your scenario-based planning.