Strategy Explore

Scenario Planning

Scenario planning builds several plausible futures and stress-tests a decision that has to survive more than one of them.

The questionWhat would we do differently if the outside forces we cannot control went one way and not another, and which decisions hold up either way?

Process · Herman Kahn, then Pierre Wack at Shell, 1950s to 1970s

1
Focal question The decision and a matching horizon
2
Driving forces Predetermined elements and uncertainties
3
Critical uncertainties Two to four that matter most
4
Build scenarios Consistent stories with a path from today
5
Test the decision Run each option against each scenario
6
Signposts Indicators, owners and review dates

Simplified sequence of the scenario planning steps described in the note.

When to use it

A large, hard-to-reverse commitment, such as a plant, a long contract, or entry into a regulated market, depends on conditions the client cannot predict.
The leadership team's plans all assume one view of the future, and nobody has written down what would make that view wrong.
Executives disagree about what will happen in the industry and are arguing positions, so the work needs a place to put different views side by side and see what each implies.
The client's forecasts have missed by a wide margin more than once.

When to avoid it

The outcome is mostly under the client's control, or the uncertainty is narrow enough for a sensitivity analysis or a base case with ranges.
The decision is due within weeks and nothing the exercise could show would change it.
The team wants a probability-weighted number to put into a financial model. Scenarios do not carry probabilities, so a valuation with explicit ranges is the better tool.
No senior person will act on the results. The workshop will be pleasant, and the report will go on a shelf.

Overview

Scenario planning replaces a single forecast with a few different, internally consistent accounts of how the outside world could develop, and then tests a decision against each. It supports choices that are expensive to reverse and that depend on forces the client can neither predict nor control. The output is a view of which options hold up across the stories, and which signs would show that one of them is under way.

Scenarios are not forecasts and carry no probabilities. The common misuse is a quick two-by-two that gives four labeled quadrants and no insight about the decision. The supporting evidence is mostly case reports, especially from Shell, and not controlled studies, so present the method as a way to improve decisions under uncertainty and not as proven.

What it is not

A forecast with three versions. Best, base, and worst cases vary the size of an outcome along one line, and each usually carries a probability. Scenarios are different accounts of how the outside world could develop, and none of them is the expected one.
The mechanical two-by-two. Drawing axes for two uncertainties and naming four quadrants is one way to build scenarios. It often produces neat labels with no logic behind them, and the axes are chosen for convenience and not from the decision at hand.
A way to find the most likely future. Choosing one scenario and planning for it turns the exercise back into a forecast.

The steps

Fix the focal question and horizon

State the decision the scenarios will inform, such as "should we build this plant," and set a horizon that matches the life of the commitment. A vague question such as "the future of the industry" gives scenarios that interest everyone and help no one.
Gather the driving forces

Interview executives and outsiders, such as customers, regulators, and technologists, and list the forces that will shape the answer, across social, economic, technological, political, and environmental areas. Split them into predetermined elements, which will happen regardless, such as demographic shifts already under way, and uncertainties.
Choose the critical uncertainties

Pick the two to four uncertainties that matter most to the decision and are hardest to predict. Check that they are reasonably independent of each other. This is the step where the design is won or lost, so spend the time here.
Build the scenarios

Combine the uncertainties into a small number of internally consistent stories, each with a path from today and not only an end state. Name each scenario so people remember it. Aim for scenarios that call for different decisions, not for the same decision at a different scale.
Test the decision

Run each option against each scenario and note where it fails. Look for moves that hold up in all of them, options worth keeping open, and steps that bet on a single scenario. Ask whether any scenario makes the executives uncomfortable. If none does, the set is probably too close to the existing view.
Set signposts and review dates

Name specific, observable indicators that would show which scenario is unfolding, who watches each one, and what level triggers a decision. Add a date to review the scenarios themselves.

In practice

The focal question for a regional water utility was narrow: should it commit to a thirty-year desalination plant, or expand reservoir capacity and demand management? The team identified rainfall patterns, the regulator's approach to allowed returns, and growth in industrial demand as the uncertainties that mattered most, and built three scenarios from them. One of the three was a decade of persistent drought with a regulator hostile to new debt, which the executives had not wanted to discuss.

Tested against each scenario, the reservoir option held up in two of them and failed in sustained drought. The plant was costly and idle in the wet scenario. The utility chose a staged path: fund the design and permitting for the plant, which was cheap, and expand the reservoir. It agreed in advance on signposts, including multi-year rainfall deficits below a stated level and industrial connection requests above a stated level, that would trigger the decision to build.

What you leave with

A focal question, with a list of predetermined elements and critical uncertainties.
Two to four scenario narratives, each with a path from today.
A table showing how each option behaves in each scenario.
A decision on what to do now, what to keep open, and what to hold until a trigger is met.
A short list of signposts, each with an owner and a review date.

Failure points

Scenarios that converge

All three are variations of the future the client already expects, one slightly better and one slightly worse. The team chose uncertainties it was comfortable with. Test the set by asking which scenario the leadership would least like to be true.

A favorite scenario

The executives pick the most probable story and plan for it, and the others become decoration. The pull toward a single number is strong, and the exercise is consumed as a forecast.

Narratives without a decision

The workshop produces vivid stories but no table of options against scenarios and no signposts. People leave energized and nothing changes, because the stories were never tied to a choice.

Axes chosen by the facilitator

Two axes are picked early, often by the sponsor, and the scenario set follows a template. The uncertainty that actually matters, such as a rival's technology, never appears. Derive the axes from the focal question and the interviews.

Signposts that nobody watches

The signposts are listed without owners or dates. A year later nobody can say which scenario is closest, and the exercise gets repeated from scratch.

The failures above are usually Framing: the wrong problem was chosen before this framework was applied.

Origin

Herman Kahn, working on US military problems at the RAND Corporation in the 1950s, adopted the word scenario for stories of how events might unfold. In 1961 he founded the Hudson Institute, where he applied the method to social and policy questions. Business use for strategy is usually traced to Pierre Wack in the planning group of Royal Dutch/Shell, from about 1971. The 1973 oil shock is commonly cited as the event that made Shell's management take the work seriously. Wack described the method in two 1985 Harvard Business Review articles, "Scenarios: Uncharted Waters Ahead" and "Scenarios: Shooting the Rapids."

Peter Schwartz, who headed scenario planning at Shell from 1982 to 1986 and later co-founded Global Business Network, wrote The Art of the Long View (1991), which spread a step-by-step version of the method. The two-axis grid that many consultants use today is one common variant. The evidence that scenarios improve decisions is mostly anecdotal and comes largely from Shell's own accounts. The method is not trademarked or licensed.

Scenario Planning

When to use it

When to avoid it

Overview

What it is not

The steps

Fix the focal question and horizon

Gather the driving forces

Choose the critical uncertainties

Build the scenarios

Test the decision

Set signposts and review dates