Scenario planning: preparing for several futures

Forecasts give one answer about the future, and they are often wrong. Scenario planning takes a different approach. Instead of betting on one forecast, you describe several plausible futures, then check whether your plans would hold up in each of them. The aim is to prepare well, not to predict correctly.

This guide explains what scenarios are, where the method comes from, a four-step process based on the UK government’s futures toolkit, a worked illustration and the common mistakes.

What scenarios are, and what they are not

The UK Government Office for Science publishes a Futures toolkit for policymakers and analysts. It is direct on this point: “Scenarios are not plans or predictions of the future; elements of your scenarios will happen; some will not.” A scenario is a coherent, plausible story about how the world around you could develop. A set of scenarios covers a range of futures, and each one is different enough to test a plan against.

Scenarios are most useful when the future depends on a few big uncertainties you cannot control, such as regulation, technology, customer behaviour or the economy. They are less useful for questions that a good forecast or a simple sensitivity analysis can answer.

A short history

Most writers credit Herman Kahn, who worked at RAND in the 1950s and coined the term “scenarios” for stories about possible futures, as summarised on Wikipedia. Gaston Berger developed similar methods in France around the same time.

In business, the method is linked with Royal Dutch Shell. Strategy-focused scenario work began at Shell in 1971 under Pierre Wack, and the 1973 oil shock gave it more impetus. Wack, who headed the business environment division of Shell’s planning department and co-developed its scenario system with Edward Newland, described the approach in Harvard Business Review in 1985. Shell’s experience in the oil shocks of the 1970s is the best-known success story, and accounts of Wack’s work say Shell anticipated both shocks of that decade. Critics note that much of the evidence for the method is anecdotal and comes from Shell itself.

Step 1: frame the question and the time horizon

Start with a decision, not a general wish to “look at the future”. Examples: “Should we open a second warehouse?” or “Which markets should we build up over the next ten years?”

Then choose a horizon far enough out for major change to happen. The futures toolkit suggests looking at least five years ahead, and generally 10 to 20 years for policy work. A small business may use a shorter horizon, but it should still be longer than its normal budget cycle.

Step 2: find the drivers and the critical uncertainties

List the forces that could shape your question. The toolkit’s driver mapping sorts them into categories such as PESTLE (political, economic, social, technological, legal and environmental). Gather them from horizon scanning, interviews and reading beyond your usual sources.

Next, plot each driver by how important it is to your question and how certain its direction is. Drivers that are important and fairly certain, such as an ageing population, belong in every scenario. Drivers that are important and highly uncertain are your critical uncertainties, and they shape the scenarios. The toolkit suggests starting with at least ten critical uncertainties.

Step 3: build and write the scenarios

The toolkit’s 2×2 method has five steps:

1. Cluster the critical uncertainties into two independent axes, each running between two opposite outcomes.
2. Describe the world in each of the four quarters the axes create.
3. Give each scenario a short, memorable name.
4. Present and discuss the scenarios with the wider group.
5. Write them up.

Sources differ on how many scenarios to use. The 2×2 method gives four. The Wikipedia summary of Shell-style practice describes drafting seven to nine mini-scenarios and narrowing them to two or three, and says around three is often easiest to discuss. Either works if each scenario is plausible, clearly different and relevant to the decision. Avoid labelling them “best”, “worst” and “most likely”: people then focus on the middle one and ignore the rest.

Step 4: use the scenarios

Scenarios earn their keep when you test decisions against them. The toolkit pairs them with policy stress-testing, “a method for testing policy, strategy or project options against a set of scenarios.”

• Test each option. For every strategic option, ask how it performs in each scenario.
• Find no-regret moves. Actions that help in every scenario can go ahead now.
• Find hedges and options. Some actions protect you in a bad scenario at low cost, or keep a door open.
• Set signposts. For each scenario, name early signs that it is starting to happen, and who will watch for them.

A worked illustration

An example (illustration only). A UK online retailer of home fitness equipment is deciding whether to sign a ten-year lease on a larger warehouse. Its team picks two critical uncertainties: whether home exercise stays popular or people return mainly to gyms, and whether online advertising costs stay manageable or rise sharply.

Testing the ten-year lease, the team sees it only works well in one scenario. A no-regret move in all four is building an email list, which lowers dependence on paid advertising. They choose a shorter lease with an option to expand, and set signposts: monthly gym membership news, their own repeat-purchase rate and cost per acquisition.

Common mistakes and limits

• Treating scenarios as forecasts. Assigning probabilities tends to make people plan for one scenario only.
• Choosing axes that are not independent, which produces scenarios that cannot happen.
• Stopping at the stories. An early Shell review led by Arie de Geus found that problems came more from the decisions after the scenarios than from the scenarios themselves.
• Using only internal views. Bring in outsiders, customers and people who disagree.

The method is subjective, and its long horizons make it hard to prove that it works. Use it for big, hard-to-reverse decisions where the outcome is highly uncertain, combine it with a PESTLE scan and financial modelling, and revisit the scenarios when your signposts start to move.