In 1970, a planner at Royal Dutch/Shell named Pierre Wack started telling the company's managers a story in which the price of oil quadrupled. Almost no one believed it. Three years later the 1973 oil shock arrived, and the firms that had merely forecast a stable future were caught flat-footed, while Shell, having rehearsed the shock on paper, moved faster than rivals many times its size. That is the whole promise of scenario planning and its sharper cousin, wargaming: not knowing the future, but being less surprised by it.

The quick version

  • Scenario planning builds a few plausible, sharply different stories of the future, then asks: would our strategy survive each one? It exists to change how leaders see, not to predict.
  • Wargaming is role-play: teams act as your competitors, regulators and customers, then react to your plan, surfacing the counter-moves your own optimism hides.
  • The point of both is the rehearsal, not the document. The value is in the argument the room has, and the early-warning signals you agree to watch for afterwards.
  • Neither predicts. Both fail when they become a box-ticking ritual, when scenarios are too timid to threaten the plan, or when no one acts on what surfaced.

The idea in depth

Start with the problem both tools were built to solve. Traditional planning runs on forecasts, a single best-guess line projected forward. That works while the world is stable, and fails exactly when it matters: at the discontinuities. Pierre Wack made this case in two now-classic Harvard Business Review essays, "Scenarios: Uncharted Waters Ahead" and "Scenarios: Shooting the Rapids" (both 1985). His central, counter-intuitive claim is that good scenarios don't aim at better forecasts at all. They aim at the mental model of the decision-maker, the picture of the world a leader carries in their head and quietly defends. A scenario succeeds when it changes that picture.

So the move is: stop asking your planning process "what will happen?" and start asking "what would have to be true for us to be badly wrong?" Write two or three futures that each make a different one of your core assumptions false. If your strategy looks fine in all of them, either it is genuinely well-built or, more often, your scenarios are too polite to do their job.

The discipline got its modern, teachable shape from Paul Schoemaker. In "Scenario Planning: A Tool for Strategic Thinking" (MIT Sloan Management Review, Winter 1995), he frames scenarios as a deliberate antidote to two well-documented biases: overconfidence (we think we know more than we do) and tunnel vision (we see one future and plan only for it). His method is a sequence: define the scope and time horizon, list the key stakeholders, identify the basic trends and the critical uncertainties, then bundle those uncertainties into a handful of internally-consistent stories. The output isn't a prediction; it's a set of plausible worlds detailed enough to argue with.

flowchart TD
    A(["Scope & time horizon"]) --> B(["List trends you're sure of"])
    B --> C(["List the critical uncertainties"])
    C --> D(["Bundle uncertainties into 2-4 stories"])
    D --> E(["Stress-test the strategy in each"])
    E --> F(["Agree early-warning signals to watch"])
Schoemaker's scenario method, simplified, the work ends not with a story but with signals. Leaders Loop

Wargaming attacks a different blind spot: your rivals have a vote. Most plans quietly assume the competition stands still. Business wargaming, adapted from military simulation and codified for managers by Benjamin Gilad in Business War Games (2008), fixes this by putting people in the room to be the competitors. In a typical session, one team presents the company's own plan while other teams role-play named rivals, a regulator, or a disruptive new entrant, using real competitive intelligence to react as those parties actually would. Gilad's argument, drawn from running these games for Fortune 500 firms for over thirty years, is that the exercise reveals two things at once: how the market will likely respond to your move, and the internal politics and culture that would stop you responding back.

So the move is: before you commit to a big launch or price change, spend a day with your plan on trial. Assign your sharpest people to play the two competitors you most fear, brief them to win, and let them attack. The objection that surfaces in that room for free is the one that would otherwise surface in the market for money.

"Scenarios are stories about the way the world might turn out tomorrow.", Peter Schwartz, The Art of the Long View

An honest limitation. It is tempting to claim these methods "predict" disruptions, and the Shell legend encourages that. They don't, and the academic record is careful about it. Researchers including George Wright, Ron Bradfield and George Cairns, writing in Technological Forecasting and Social Change, have noted that scenario planning rests on plausibility, not probability: a scenario is judged by whether it is coherent and challenging, not by how likely it is, and the field still lacks strong controlled evidence that scenario work reliably improves decision outcomes. That's not a reason to skip it. It's a reason to treat the output as a thinking aid, not a forecast, and to be suspicious of any vendor who sells it as foresight-as-a-service.

A worked example

Take a mid-sized UK software firm, call it a payroll-and-HR platform for small businesses, weighing whether to bet the next two years on an AI-assistant feature. The forecast view says: usage is up, competitors are slow, ship it. Here's the same decision run through both tools. (Figures below are illustrative, to show the shape of the work, not real market data.)

First, the scenarios. The team picks two critical uncertainties: will regulators tighten rules on automated payroll decisions? and will the dominant accounting suite bundle a free rival feature? Crossing those gives four plausible worlds.

quadrantChart
    title Two uncertainties, four futures
    x-axis "Light regulation" --> "Tight regulation"
    y-axis "Incumbent stays out" --> "Incumbent bundles free rival"
    quadrant-1 "Race to the bottom"
    quadrant-2 "Compliance moat"
    quadrant-3 "Open field"
    quadrant-4 "Squeezed both sides"
A 2×2 of the two uncertainties that most threaten the bet. Each quadrant is a world to stress-test the plan against. Leaders Loop

"Open field" (light regulation, incumbent stays out) is the forecast world everyone already pictured. The useful quadrants are the others. In "Squeezed both sides," a compliance burden raises the cost of the AI feature and a free bundled rival caps what customers will pay, the bet loses money in roughly the illustrative range of a 30–40% revenue shortfall against plan. That world is plausible. So the strategy adapts: the team agrees two early-warning signals, a named regulator's consultation paper, and any hiring by the incumbent for "payroll AI", and pre-commits that if both trip, they slow the spend and lead instead on the auditable, compliance-friendly angle the regulation would reward.

Now the wargame. A half-day session puts four people in the incumbent's seat, briefed to crush the upstart. Within an hour they land on a move the home team hadn't priced in: not building a better feature, but giving a basic one away to make the category a checkbox, not a product. That single insight reframes the whole bet, the question is no longer "can we build it?" but "can we be valuable enough that free isn't good enough?" No forecast would have produced that sentence. A rival, played by your own people, did.

Frequently asked questions

Isn't this just expensive guessing?

It would be, if the goal were to guess the future. It isn't. The deliverable is a better-prepared management team and a short list of signals to monitor, the kind of preparation that let Shell react to the 1973 shock faster than peers. You're buying readiness and a wider field of vision, not a prophecy.

How many scenarios should we build?

Two to four. One is a forecast in disguise; five or more and the team can't hold them in mind or argue about them seriously. Schoemaker's guidance is a small set of internally-consistent stories that differ on the uncertainties that actually matter, not every variable you can think of.

What's the real difference between a scenario and a wargame?

A scenario rehearses the environment, regulation, demand, technology, the macro weather your strategy must survive. A wargame rehearses the actors, the specific competitors and parties who will react to your move on purpose. Many strong strategy reviews use scenarios to set the stage and a wargame to populate it with adversaries.

We're a small team with no budget. Can we still do this?

Yes, and you should resist the consultant-grade version. A lightweight scenario session is two hours, a whiteboard, two uncertainties and a 2×2. A lightweight wargame is one afternoon and people willing to argue against the house. The cost is honesty and attention, not money.

How do we stop it becoming a ritual that changes nothing?

Tie every session to a live decision and end it with commitments: the signals you'll watch, who watches them, and the pre-agreed move if a signal trips. A scenario exercise that produces no change to the plan and no trigger to monitor has failed, however good the stories were.

Related in the Toolkit

Where to go next