In 1971, a small team inside Royal Dutch Shell began rehearsing for a crisis that had not yet happened. Led by Pierre Wack, they weren’t trying to forecast an imminent oil embargo, and never claimed they could.

What they did was harder. They walked Shell’s executives through how the company would respond if oil prices suddenly quadrupled, repeating the exercise until the choices no longer felt hypothetical. When OPEC throttled the global economy in October 1973, Shell acted while its rivals froze. Within a decade, it had risen from the weakest of the seven oil majors to one of the strongest.

That’s as close as corporate strategy gets to a controlled experiment, and the lesson has nothing to do with oil. Shell won because it had built a way of deciding before the panic arrived.

Part One set out the six foundational settings of the global economy that have broken. Part Two showed how the rules of business are shifting from prices to politics and ended on the three questions every board now has to answer: what do we own, what do we depend on, who can take it from us.

Part Three: what to do about it

Most boards can answer the first question from the balance sheet. The second and third are harder, because they test judgement rather than proficiency in the P&L. This instalment is about building that judgement: how a board decides, rather than what it decides, in a world none of us can predict.

The paperwork problem

A board’s job is judgement under pressure. Walk into most boardrooms and you find paperwork. Strategy arrives in one pack. Audit and risk in another, with its own jargon and committee. Sustainability in a third, dense with reporting standards and climate scenarios. Each comes from a different team, with its own consultants.

The strategy paper does not reflect the risk register; the risk register does not reflect the scenarios; the scenarios sit in a compliance folder no director opens twice. The board ends up busy and exposed, grappling with three byzantine documents, each balancing risk and reward but none assembling the whole. When the shock hits, there is no single position to act from.

The problem runs deeper than structure. Daniel Kahneman’s work on “noise” shows that even experienced professionals, given the same information, scatter far more than anyone assumes. At one insurance company, two underwriters given the same risk file produced estimates that differed by 55%.

Boardrooms are no exception. Give the same strategic brief to each director independently before a meeting and compare the answers. The spread is the noise in your decision-making.

What antifragility means

Start with the packs. Require the executive to bring strategy, risk and sustainability to the board as one paper, because the company has one future, not three. Rebuilding the company is harder, and for that the aim should be antifragility, a term coined by options trader Nassim Taleb. A resilient system survives stress and carries on unchanged; an antifragile one feeds on it and comes out stronger.

Think of the Australian bush. When the fire comes, a fragile forest burns to the ground. A resilient one survives the flames and recovers. The eucalypt does something stranger. It needs the fire. The heat splits open its woody capsules, dropping seed onto a fresh bed of ash. The tree resprouts from buds buried under its own bark while the ground is still warm. A fire that levels its rivals clears the ground for new growth. The board’s task is to build a company with more eucalypt than kindling, and to know which is which.

Take a utility with an ageing coal plant it must close. The fragile operator treats the closure as a loss to be minimised, defends the asset to the last legal day and is left with a stranded plant and no plan. The resilient operator manages an orderly exit, meets its obligations, and pivots late into clean energy. The antifragile operator builds grid-scale storage on the same connection, a virtual power plant in partnership with its customers, solar arrays feeding a data centre where the smokestack used to be. One event, three responses; only one ends the decade stronger.

Start where the forces collide

Boards track risks and opportunities in isolation: separate line items for AI, energy prices, supply chains, geopolitics. The forces that decide whether a business survives do not sort themselves so neatly. They collide where risk and opportunity meet, and a board whose orienting assumption is wrong will misread the signals that follow.

Consider AI, a standing item on most agendas. The global build-out of AI runs on data centres, and data centres run on power. That demand is hitting grids that cannot add capacity fast enough, so wholesale prices and emissions rise together. A technology story has become an affordability problem and a climate problem at once.

Go one step further. The batteries and transmission needed to firm that new load rely on critical minerals. China processes close to 90% of the world’s rare earths and has already shown it will restrict flows when it suits Party interests. A board that starts by talking about AI as a productivity tool should also be discussing energy security, the cost of living, and dependence on a strategic rival. Four trends, one collision, on a single balance sheet.

Identifying the overlap is the easy half. The harder work is a discipline of scenario analysis for any shock, run as a rehearsal the board performs, not a report it files. It has six steps:

1. 1. Map the outside forces that could reshape the business.
   2. Find your own weak points: the dependencies and exposures a shock would hit.
   3. Build three honest scenarios: best, worst, and base case.
   4. Put the board in the room and rehearse each one.
   5. Set the triggers: the early signals, and the action each one commits you to.
   6. Refresh quarterly, as the world moves.

Done well, the result is what military strategists call pattern-based decision-making: the ability to act fast by matching a live signal to a rehearsed scenario. Directors stop sifting through dense reports and start having real debates, weighing trade-offs, testing hypotheses, learning to stress-test a decision before the stakes turn existential.

Step five matters most. A trigger is a decision you make before you need it, agreed while everyone is calm. “Monitor critical-mineral exposure” is not a trigger. “If Chinese export licensing reaches the magnets in our storage supply chain, we move to the pre-signed second source that week, at higher cost” is.

One refinement from military planning. A single signal can mislead, so set several, and agree in advance what their tripping together commits you to: if these four indicators are met, we change course to plan B. Planners call these branch plans. The new branch is already there, budded under the bark, waiting for the fire.

The pre-mortem

The trigger tells the board what to do when a signal fires. The pre-mortem, a technique developed by psychologist Gary Klein, surfaces the signals no one has thought of. Before committing to a strategy, the board assumes it has already failed, and each director writes down why; only then does the room discuss. Imagining a future event as if it has already occurred – prospective hindsight – makes people about 30% better at identifying causes of failure, because the brain is better at explaining facts than predicting possibilities. Kahneman called the pre-mortem his single most useful debiasing technique. A trigger pre-commits the response; a pre-mortem tests whether it is aimed at the right risk.

Antifragility in action

Three companies show what it looks like in practice.

Start with Toyota. When the 2011 Tohoku earthquake knocked out key suppliers and stalled production, Toyota did more than rebuild. It drew up a list of several hundred critical components and required suppliers to hold two to six months of extra stock, a deliberate retreat from the just-in-time orthodoxy it had spent decades perfecting. Those buffers sat unused for nine years. When the 2020 chip shortage hit, Toyota kept its lines running while Ford and GM slashed output. Cover was not total, but through the worst of it the carmaker that had pre-committed its response was the one still building.

In 2021, in the most profitable year in its history, BHP sold out of oil and gas and reinvested the capital into copper, the principal metal of the energy transition. Reshaping a portfolio at the top of a cycle is not, strictly, antifragility, because BHP did not gain from a shock. Instead, it killed a winning position before it could damage the company’s future. By December 2025, a little over half of BHP’s earnings came from copper, roughly 30 percentage points higher than three years earlier. The oil exit looked mistimed when the 2022 energy crisis sent crude soaring, but the economics of copper are more compelling. The lesson is the discipline to fold a hand most boards would play to the last card.

Finally, Berkshire Hathaway, which has made antifragility its whole design. Warren Buffett holds a vast cash reserve as deliberate optionality, ready to deploy when everyone else is a forced seller. In the 2008 panic, while banks were begging for capital, Berkshire wrote a US$5bn cheque to Goldman Sachs and US$3bn to General Electric, on terms only a buyer with cash and nerve in a crisis could command. The Goldman investment alone earned Berkshire billions. The cash earns little when markets climb, and Buffett wears years of criticism. That is the price of optionality.

Boards want a method, not a lecture

Three moves are enough to begin, and a board can make them at the next meeting.

1. Put geopolitics inside the strategy.

   The question of who can take what you depend on belongs in the same paper as the growth plan, priced into capital allocation, not quarantined in a risk appendix. A growth plan that ignores the dependencies it rests on mistakes exposure for strategy.

2. Turn efficiency into optionality wherever a single chokepoint exists.

   The value of an option rises with volatility. In a stable world, redundancy is expensive slack. In the world described in Parts One and Two, it is cheap insurance.

3. Write your first trigger and your first kill criterion.

   Take one collision the board can already see forming. Decide what signal would force a move and what you would do, then decide what would make you walk away. Agree both at the table while the decision is still cheap to get wrong.

A shock is coming for every company. The only thing a board controls is whether it has built something that burns down, or something that, like the eucalypt, comes back from the fire thicker than before.