Every strategic decision contains a prediction.

A company enters a market because it expects demand to exist. It launches a product because it expects customers to buy it. It acquires another company because it expects the combined business to create more value than the two businesses could separately. It builds a factory, hires people, signs a five-year contract, or invests in technology because it expects the conditions around that investment to hold long enough to make the decision worthwhile.

That makes strategy, at least in part, a prediction problem.

Executives rarely describe themselves as forecasters. They are strategists, operators, investors, or leaders. The distinction is mostly about vocabulary. Bain’s discussion of the 2026 CEO agenda frames executive ambition as a set of choices that must be converted into execution.^² The important question, then, is not whether management predicts. It is whether those predictions are explicit, testable, and good enough to guide where the company puts its money and people.

Consider a familiar strategy-class example: Kodak.

Kodak did not fail because its executives were unaware of digital photography. The company had developed an early digital camera. The strategic problem concerned the prediction attached to it. Kodak’s economics depended heavily on film, processing, and printing, so moving aggressively into digital meant betting that value would shift away from physical film quickly enough to justify cannibalising the existing business.

Suppose management had framed the decision explicitly as a prediction: By 2010, digital photography will account for X percent of consumer image-taking, margins on physical film will fall by Y percent, and consumers will value convenience and storage more than physical prints.

Now there is something management can watch, test, and revise. Without that framing, “digital is important” remains an observation. It becomes strategy only when management decides what it means for resource allocation.

The Quality of Strategy Is Declining

This matters because the quality of corporate strategy appears to be deteriorating at precisely the point when strategic decisions are becoming harder.

McKinsey’s 2025 research found that only 21 percent of executives reported that their strategies passed four or more of its Ten Tests of Strategy. That compares with 35 percent in its 2010 survey of more than 2,000 executives.^³ The proportion reporting strategies that met that threshold has therefore fallen by roughly 40 percent.

The environment has also become harder to read. McKinsey’s analysis of the World Uncertainty Index found that baseline uncertainty had more than doubled since 1990, while high-uncertainty events had become more frequent and successive major spikes more severe.^⁴ AI, geopolitical disruption, regulation, and new business models add further variables to decisions that were already based on imperfect information.

The consequences are becoming more uneven. McKinsey’s Economic Profit Power Curve shows that the top quintile captures nearly 90 percent of the economic surplus in its analysis.^⁵ The gap between the average gains of the top quintile and losses of the bottom quintile has doubled over two decades.^⁶

Uncertainty itself, however, is not the problem. Strategy has always involved uncertainty. The problem is pretending that uncertainty removes the need to make a prediction.

Netflix provides a useful counterexample. Rather than treating DVD rental as the permanent business, management bet that entertainment distribution would move towards digital delivery. The timing did not need to be perfect because Netflix could build capability gradually and increase investment as evidence changed. Good strategy does not require knowing exactly what will happen. It requires knowing what management is assuming will happen.

Prediction Is Inevitable, Even When a Company Opts Out

Companies often respond to uncertainty by trying to avoid prediction altogether. They diversify, delay decisions, maintain multiple options, ask for more data, or wait for the market to become clearer. Each of those choices, however, rests on an assumption about what will happen while the company waits.

Suppose a company decides not to enter a new market because demand is uncertain. That is a prediction that waiting will create more value than entering now.

Suppose it keeps an underperforming product because management does not want to make an irreversible decision. That is a prediction that the product may recover, or that keeping it costs less than abandoning it.

Suppose a company postpones investment in a new technology until competitors have demonstrated that customers want it. That is a prediction that the company will still be able to acquire the necessary skills and technology later, at a reasonable cost.

Bain’s Macro Trends Group makes the same analytical point: companies do not escape prediction by declining to forecast because every choice still implies an expectation about the future.^⁹ A decision to wait is not neutral. It is a bet on the value of waiting.

The practical issue is that implicit assumptions are harder to test than explicit ones. A leadership team that says, “We think AI will change customer-service economics within three years,” can identify what would need to happen for that belief to be true. It can decide which evidence to track and when to reconsider the investment. A leadership team that says, “We’re monitoring AI,” has not made its assumption visible enough to challenge.

This is why strategic discussions can become unproductive. Teams debate whether a market is attractive or a technology is important without identifying the assumptions underneath those claims. Once made explicit, executives can debate market size, timing, and probability instead of vague descriptions of the future.

The Value of Being Explicit

Making predictions explicit also changes how strategy is reviewed.

Suppose a company is considering a large investment in a new manufacturing facility. The business case might assume that demand will grow by 8 percent a year, capacity utilisation will reach 80 percent within four years, and input costs will remain within a particular range. Those numbers are not simply financial inputs. Collectively, they express the company’s view of how the market will develop.

If management treats them as fixed spreadsheet entries, the strategy can remain intact after its assumptions have changed. If it treats them as predictions, the figures become indicators to monitor. Management can compare demand, utilisation, and input costs with the original expectations, then accelerate, modify, or abandon the plan.

This creates a feedback loop between strategy and evidence. McKinsey’s work on Strategy Champions similarly places “testing assumptions and adapting” within execution, rather than treating execution as mechanical delivery of a fixed plan.^⁷ The same research argues that companies which fail to document their assumptions may be unable to distinguish an execution problem from the failure of the strategic hypothesis itself.^⁸

An execution problem may require different people, processes, or incentives. A failed hypothesis requires a different strategy. Confusing the two can lead a company to execute a weak idea with greater discipline.

Explicit assumptions are especially valuable when commitments can be staged. A company can invest on a smaller scale, learn, and commit more capital when evidence supports the thesis. This reduces the cost of being wrong while preserving the option to scale.

The same principle applies to the CEO agenda. Ambition only becomes strategically useful when it is translated into choices, resources, accountabilities, and evidence that allows management to judge whether execution is producing the expected result.^¹⁰

Forecasting Is Not the Same as Foresight

Traditional forecasting tends to ask: What is most likely to happen if current trends continue?

Strategic foresight asks a broader question: What different futures could affect our strategy, and what would we do in each one?

Consider Amazon’s expansion beyond books. If Amazon had looked only at its existing business, the obvious objective would have been to become a much larger online bookseller. Instead, the company made a broader bet: the systems it was building around technology, logistics, fulfilment, and customer trust could become valuable across many categories.

The prediction was not simply that online retail would grow. It was that the capabilities needed to win could become more valuable than Amazon’s starting category, leading to a different allocation of capital.

Scenario analysis extends this logic. Suppose an industrial company is deciding whether to build a new factory. A conventional forecast might produce one expected demand number for 2030. A foresight exercise might examine three plausible situations: demand grows rapidly, demand stagnates, or regulation and technology change the product category.

The question becomes less about one exact number and more about which investment remains defensible across several futures. BCG’s work on strategic foresight similarly presents foresight as a disciplined capability for anticipating change and informing action, not eliminating uncertainty.^¹¹

Harvard Business Review’s 2026 article, based on research involving 500 organisations, examines what companies with stronger strategic-foresight practices do differently.^¹² A central implication is that foresight needs to connect signals across different time horizons with decisions made in the present.^¹³ The purpose is not to predict everything. It is to recognise which developments could invalidate the assumptions supporting today’s strategy.

Tools such as the World Economic Forum’s Strategic Intelligence platform can widen the trends considered by decision-makers.^¹⁴ More signals do not automatically produce better strategy. They matter when management connects them to an assumption, indicator, or decision threshold.

The New Competitive Divide

Companies that are good at strategic foresight do not necessarily have better crystal balls. They have better processes for making assumptions visible, testing them against evidence, considering alternatives, and deciding when to change course.

McKinsey’s research on Strategy Champions points in the same direction. The strongest companies do not simply produce better strategies. They are also better at mobilising the organisation by assigning ownership, translating choices into initiatives, reallocating resources, and embedding strategy in plans and budgets.^¹⁵

Suppose two companies reach the same conclusion: AI will materially change their industry.

Company A commissions a strategy document, approves a three-year plan, and allocates a large budget. Company B states the prediction, identifies the assumptions behind it, defines what evidence would support or contradict them, runs smaller experiments, and decides when investment should increase, decrease, or stop.

Both companies have made the same basic prediction. Their difference lies in how they manage the possibility that the prediction is wrong.

The lesson is not that companies should eliminate the possibility of error. They cannot. The better question is whether a company can recognise that its original bet is failing before it has spent five years and hundreds of millions of dollars proving it.

Being wrong is not necessarily the greatest strategic failure. Continuing to allocate resources on the basis of an assumption that the evidence has already disproved is more damaging.

Strategy should therefore be treated less as a document describing the future and more as a system for making decisions when the future is uncertain. The best strategists make their assumptions visible, test them against evidence, consider what would happen if those assumptions were wrong, and decide in advance what would make them change course.

Strategy is a prediction problem. The advantage comes from getting better at making the prediction and better still at recognising when it no longer holds.