Wind-tunnelling: stress-testing a strategy against scenarios

A scenario is a plausible story about how the outside world might look in five or ten years. It is not a prediction, and a set of scenarios is not a forecast with a high case and a low case. Wind-tunnelling means running the strategy through three or four genuinely different futures and asking, in each, whether it survives. The output is not a better forecast. It is a shorter list of choices. The moves that are right in every future. The moves that are right only in some, and need a trigger. And the bets the business is knowingly making.

Most strategies are built on one future, usually last year plus a trend, and the plan carries a sensitivity table that varies it by a few percent either way. That is not uncertainty. It is a forecast with a margin of error. The futures that break a strategy are the ones where the structure of the market changes, and a sensitivity table cannot reach them.

Scenarios versus forecasts

A forecast extrapolates from what is known and asks how far it might be wrong. A set of scenarios starts from what is not known and asks how differently things could turn out. The forecast is more useful for next year's budget. The scenarios are more useful for choices that take five years to pay back, because those choices are exposed to the futures the forecast cannot see.

Two rules keep scenarios honest. They describe the world outside the business, not the business itself. And they are built from uncertainties the business does not control. A scenario built around "we succeed" or "we fail" is not a scenario.

How to build a small scenario set

The process has seven steps and can be done in a few workshops with a week's preparation. The quantified version takes longer and is worth it when a large investment hangs on the answer.

1. Fix the question and the horizon. What decision is the exercise for, and over what period? "Should we build the second plant?" over ten years is a scenario question. "What will next year's volume be?" is not.

2. List the driving forces, from the outside in. Start with the forces the business cannot influence at all: economy, demographics, technology, regulation, resources, social change. Then the industry forces it can influence a little: demand, customer needs, competitors, channels. Only then its own environment. Most teams start from the inside and never reach the forces that matter.

3. Separate the trends from the uncertainties. A trend has a knowable direction and belongs in every scenario: an ageing customer base, a technology getting cheaper. An uncertainty could tip either way and matters a lot: whether a regulation lands, whether a substitute reaches price parity, whether the largest customer consolidates. Rank the uncertainties by impact and by how uncertain they really are.

4. Pick two critical uncertainties. The two with the highest impact and the least certainty, that are independent of each other, and that the business does not control. Two axes give four futures. Try several pairs before settling; the first pair is rarely the best.

5. Write the four scenarios. Each gets a name, a short story of how the world got there, and a description of what customers, competitors, suppliers and regulators are doing in it. A good set is plausible (each could happen), relevant (to the decision), divergent (they stretch in different directions), balanced (not three disasters and a dream) and challenging (not four versions of today).

6. Run the strategy through each. This is the wind tunnel. For each scenario, ask how the current strategy performs, which of its assumptions fail, and what a business that wins in that world looks like. Then ask across the set: are we prepared for any of them, or are we betting everything on one?

7. Sort the moves and set the signposts. What survives every scenario is a no-regret move: do it. What is right in some scenarios is a conditional choice: name the condition, give it an indicator and a trigger, and assign someone to watch it. What is right in only one is a bet: make it knowingly, size it accordingly, and say so. Those indicators become the strategy's assumptions and triggers.

Example: A building-products distributor's strategy assumes steady housing starts and stable import prices. Its two critical uncertainties are the direction of housing activity and whether a large overseas producer enters the market directly. In the scenario where housing falls and the producer enters, the strategy's second warehouse is a liability and its trade-counter network is the only thing that holds. The warehouse becomes a conditional choice with a trigger; investment in the counters becomes a no-regret move.

What wind-tunnelling changes

Three things, usually. Some initiatives turn out to be robust and get accelerated, because they no longer need the forecast to be right. Some turn out to depend on one future and get staged, with a trigger before the next tranche of money. And the strategy acquires a short list of things to watch, which is what turns an annual document into a living one.

It also changes the conversation. A leadership team that has walked through four futures argues less about whose forecast is right, because the point is no longer to be right about the future. It is to be ready for several.

Example: An agricultural-inputs business runs the exercise around water policy and commodity prices. Its stated strategy performs well in two scenarios and badly in two. The team does not change the strategy. It halves the size of the first step, adds two indicators to the board pack, and agrees the decision that will be taken if either trigger fires. Eighteen months later one does, and the decision takes a week instead of a year.

Pitfalls

Uncertainties that are really choices or outcomes: "whether we win the contract" is not an axis. Axes that are too extreme, so that two scenarios are implausible and get ignored. Axes that are correlated, which collapses four scenarios into two. Scenarios that are variations on today with different weather. A set built by one function, which produces four versions of that function's worries. And the commonest: building the scenarios well and never running the strategy through them, so that the exercise produces four stories and no decisions.

The five stages of handling uncertainty

How trends and uncertainty are handled is one of the practices in the facts-and-foresight dimension of a stages-of-excellence strategy assessment. The five stages read:

• Lagging: The plan assumes next year looks like last year; surprises arrive as crises.
• Basic: Trends get a slide in the planning pack and no consequences in the plan.
• Competent: The two or three outside forces that could reshape the business, a trend, a policy shift or a dependency, are named, with a stated position on each.
• Advanced: Key uncertainties are explored as scenarios and the strategy has been wind-tunneled against them.
• Leading: The strategy balances commitment and flexibility deliberately, taking big bets where confident, options where not.

Most businesses are at the second stage. The move to the third needs no scenarios at all, only a named position on the forces that matter. The scenarios are how the third stage becomes the fourth.

Frequently asked questions

How many scenarios? Four, from two axes, is the standard and it is enough. Three works if one quadrant is implausible. More than four and the team cannot hold them in mind.

How far out should the horizon be? Far enough that the structure of the market could change, and no further. Five years for most businesses; ten where the assets are long-lived. Beyond that the stories stop being useful.

Do we need a model? Not to start. A qualitative pass in a couple of workshops finds the no-regret moves and the bets. A model earns its place when the decision is large enough to justify quantifying each scenario's profit pool.

How often should the scenarios be revisited? When a trigger fires, and otherwise once a year, briefly, to ask whether the axes are still the right ones. The signposts are watched monthly; the scenarios are not rebuilt.

Isn't this just planning for the worst? No. Half of a good set is better than today, and the exercise finds opportunities the forecast hides as often as it finds risks. A business that only stress-tests for downside has built two scenarios, not four.

Have you wind-tunnelled your strategy, or just sensitised the forecast? Take the check. · The stages of excellence in strategy · How strategy processes fail