Why Avoiding Predictions Is Itself a Risky Prediction

Challenging Conventional Planning Under Uncertainty

Karen Harris and Martin Toner from Bain's Macro Trends Group deliver a provocative challenge to conventional wisdom about planning under uncertainty. Their central argument—that companies cannot avoid making predictions because every business decision represents an implicit bet on the future—deserves serious consideration. But it also requires critical examination, particularly regarding when explicit prediction helps versus when it hinders organizational effectiveness.

The authors are right that we've entered what they call The Great Transformation, a period where foundational assumptions about globalization, capital availability, labor markets, and demographic trends are reversing. In such environments, defaulting to historical patterns becomes dangerous. Yet their prescription—making implicit predictions explicit and testing conviction around them—while valuable in principle, oversimplifies the complex relationship between prediction, decision-making, and organizational action.

The Compelling Case for Conscious Prediction

The article's most powerful insight is the exposure mapping framework. Harris and Toner correctly observe that most companies operate like investors who don't know their portfolio positions. A business may think it's diversified across regions and sectors, but deeper analysis reveals concentrated exposure to a single macroeconomic variable—perhaps commodity prices, Chinese growth, or interest rate trajectories.

Consider the period between 2010 and 2015, when numerous Western companies expanded aggressively into emerging markets. On the surface, this appeared to be geographic diversification. In reality, many were doubling down on the same bet: that the commodity supercycle would continue, Chinese infrastructure spending would accelerate, and emerging market consumers would rapidly climb the wealth ladder. When these assumptions reversed simultaneously, companies discovered their seemingly diverse portfolio was actually a concentrated position.

• This pattern validates the authors' call for making exposures visible. Philip Tetlock's research on forecasting, documented in "Superforecasting," demonstrates that forecasters who explicitly articulate their assumptions and track their prediction accuracy improve over time. The key mechanism isn't prophetic ability but rather the discipline of making fuzzy intuitions concrete enough to evaluate.
• The distinction between conviction and confidence similarly illuminates a common source of strategic confusion. Statistical confidence requires sample sizes and testable hypotheses—luxuries rarely available when making singular strategic bets. A CEO deciding whether to build a multi-billion dollar production facility in Vietnam cannot run multiple trials or calculate confidence intervals. They can only assess their conviction based on judgment, experience, and available evidence.
• Annie Duke explores this territory in "Thinking in Bets," arguing that good decisions sometimes yield bad outcomes while bad decisions occasionally produce good results. What matters is the quality of the decision-making process given the information available at the time. Harris and Toner's framework for assessing conviction levels provides a structured approach to this challenge, helping leaders distinguish between decisions warranting full commitment versus those requiring smaller exploratory moves.

Where the Prediction Framework Needs Qualification

Despite these strengths, the article's recommendations require important qualifications that the authors acknowledge only briefly.

• First, there's a meaningful risk that making predictions explicit triggers psychological commitment, reducing adaptive capacity rather than enhancing it. Social psychology research on cognitive dissonance suggests that people who publicly commit to a prediction become invested in being right, filtering subsequent information to confirm their initial view. Robert Cialdini's work on commitment and consistency demonstrates how small initial commitments can lock people into escalating patterns of behavior.
• The aviation industry provides a cautionary example. Throughout the 2000s, major airlines made explicit predictions about fuel prices and hedging strategies, often announcing them publicly to shareholders. But this transparency sometimes created organizational rigidity. When oil prices moved contrary to predictions, airlines that had articulated strong conviction found it politically difficult to reverse course quickly, even when new information warranted it. The explicit prediction became an organizational albatross.
• Second, the article's treatment of scenario planning as merely a complement to explicit prediction understates the value of genuine scenario thinking. The authors criticize "overbuilt central cases" with arbitrary sensitivity analyses, and they're right to do so. But the strongest scenario planning doesn't start with a base case at all. Instead, it identifies critical uncertainties and maps genuinely different futures without privileging any single outcome.
• Shell's scenario planning approach, pioneered in the 1970s and refined over decades, explicitly avoids forecasting which scenario will occur. Instead, Shell develops multiple coherent narratives about how the future might unfold, then tests strategies against each. This approach proved prescient before the 1973 oil crisis and again during the 2008 financial crisis, not because Shell predicted these events but because it had already thought through how to respond if they occurred.
• The distinction matters because Harris and Toner's approach—make explicit predictions, then use scenarios to stress test them—still privileges the prediction. An alternative approach treats multiple scenarios as equally worthy of consideration, deliberately avoiding the anchoring effect that comes from designating a "base case" even if you plan to test alternatives.
• Third, the framework may be less applicable in domains where experimentation and learning matter more than prediction and planning. Rita McGrath's research on "discovery-driven planning" demonstrates that in highly uncertain environments—new markets, novel technologies, emerging customer segments—attempting to predict outcomes is less valuable than structuring low-cost experiments that generate learning.
• Consider how Amazon approaches new initiatives. Rather than making explicit predictions about which new business lines will succeed, Amazon operates through rapid experimentation with clear decision gates. AWS didn't emerge from an explicit prediction about cloud computing demand; it emerged from internal infrastructure projects that revealed external demand. Amazon's success stems not from prediction accuracy but from creating organizational systems that learn quickly and scale successes.

The Critical Role of Organizational Context

The applicability of Harris and Toner's framework depends significantly on organizational context—a dimension the article touches on but doesn't fully develop.

• For capital-intensive industries with long investment horizons—energy, infrastructure, aerospace—making predictions explicit is nearly mandatory. When BP decides to invest billions in offshore drilling projects that won't produce returns for a decade, it must articulate assumptions about oil demand, price trajectories, regulatory environments, and technological trajectories. These decisions cannot be easily reversed, so the discipline of explicit prediction with stress testing is essential.
• The same logic applies to industries facing structural transformation. Automotive executives cannot avoid making predictions about electric vehicle adoption rates, autonomous driving timelines, and changing ownership models. These predictions should be explicit because they drive capital allocation decisions with multi-year consequences. GM's explicit prediction that EV demand would accelerate—leading to its commitment to launch 30 electric models by 2025—represents exactly the kind of strategic clarity Harris and Toner advocate.
• But in industries with shorter cycle times and lower switching costs—software, e-commerce, digital media—the balance shifts toward adaptation over prediction. Netflix's evolution illustrates this well. While the company made an explicit bet on streaming over DVD rental (high conviction, long-term commitment), its content strategy relies more on continuous experimentation and measurement than on predicting hit shows. Netflix runs thousands of A/B tests and rapidly adjusts based on viewing data rather than making multi-year predictions about content preferences.

Addressing the Transformation Challenge

The authors' emphasis on The Great Transformation deserves particular attention because it highlights when their framework becomes most valuable and most difficult simultaneously.

• Periods of structural transition invalidate historical relationships between variables, making traditional forecasting methods unreliable. During such periods, the instinct to avoid explicit prediction becomes stronger—there's simply more uncertainty. But Harris and Toner argue this is precisely when explicit prediction matters most, even if conviction is lower.
• This argument has merit but requires careful implementation. Consider three different approaches companies might take during structural transitions:
• Approach One: Wait and See
  Some companies respond to fundamental uncertainty by delaying major commitments until patterns clarify. This sounds prudent but represents an implicit prediction that the transition will unfold slowly enough to observe and respond. For Kodak, this assumption about the pace of digital photography adoption proved fatal.
• Approach Two: Place Multiple Bets
  Other companies respond by diversifying their commitments, essentially hedging against uncertainty. Automotive manufacturers pursuing this strategy invest simultaneously in electric vehicles, hybrid powertrains, and advanced internal combustion engines. This represents a prediction that the transition timeline remains uncertain enough to warrant multiple paths.
• Approach Three: Make a Committed Choice
  Some companies read the signals and commit fully to one vision of the future. Tesla's all-electric strategy represents this approach, as does Microsoft's bet on cloud computing over perpetual software licenses. These decisions reflect high conviction about directional trends even amid uncertainty about specifics.
• Harris and Toner's framework helps clarify which approach a company is taking and whether that approach aligns with actual leadership beliefs. But the framework alone doesn't determine which approach is correct—that depends on industry dynamics, competitive positioning, and organizational capabilities.

The Practical Implementation Challenge

Even when explicit prediction is clearly valuable, implementing it effectively poses significant challenges that the article addresses only briefly.

• First, surfacing implicit predictions requires skilled facilitation and psychological safety. Most leadership teams develop implicit consensus through a gradual process of discussion and decision-making that never makes underlying assumptions fully explicit. When a facilitator attempts to articulate those assumptions directly, leaders often discover they don't actually agree—they've been operating with an illusion of consensus.
• This discovery can be valuable but also destabilizing. One consumer products company we studied brought in external facilitators to map their implicit predictions about channel evolution, only to discover fundamental disagreement between the CEO (who believed physical retail would remain dominant) and the chief digital officer (who expected e-commerce to reach 40 percent market share within five years). Making this disagreement explicit was necessary for strategic clarity, but it also triggered a political crisis that took months to resolve.
• Second, the discipline of tracking predictions against outcomes requires organizational commitment that extends beyond initial enthusiasm. Tetlock's research shows that forecasters improve through rigorous score-keeping and calibration over multiple prediction cycles. But maintaining this discipline in a corporate environment—where leaders rotate, priorities shift, and selective memory operates—proves difficult.
• Very few companies systematically track their strategic predictions and evaluate accuracy. When predictions prove wrong, the typical response is to move on rather than examine what went wrong and why. This represents a missed learning opportunity, but it's also psychologically understandable. Leaders face strong incentives to focus on the future rather than conduct autopsies on past predictions.

The Value of Strategic Humility

The most important qualification to Harris and Toner's framework is one that Nassim Taleb emphasizes throughout his work: in domains characterized by genuine complexity and fat-tailed distributions, prediction becomes not just difficult but potentially dangerous if it creates false confidence.

• Taleb's concept of antifragility suggests that in highly uncertain environments, building systems that benefit from volatility and disorder matters more than predicting specific outcomes. For example, rather than predicting which specific risks will materialize, a company might structure its operations to maintain optionality—keeping multiple paths open, avoiding irreversible commitments, and positioning to benefit from surprises.
• This approach doesn't eliminate prediction entirely, but it changes the nature of what you're predicting. Instead of forecasting specific outcomes ("oil prices will reach $100 per barrel"), you predict the character of the environment ("volatility will remain elevated") and structure your organization accordingly.
• The financial services industry learned this lesson painfully during the 2008 crisis. Banks that made explicit predictions about housing price trajectories and default correlations felt confident in their risk models—until those predictions proved catastrophically wrong. Banks that instead emphasized balance sheet strength and liquidity without claiming to predict specific outcomes weathered the crisis better.

A Synthesis for Practice

So how should leaders respond to Harris and Toner's call for explicit prediction?

The answer depends on decision context, but several principles emerge:

• Make Predictions Explicit When They Drive Irreversible Commitments
  For decisions involving significant capital investment, long time horizons, or difficult reversibility, the discipline of explicit prediction with stress testing is essential. Even if your predictions prove wrong, the process of making them explicit improves decision quality and enables faster recognition when circumstances change.
• Use Scenario Thinking to Challenge Predictions, Not Replace Them
  Develop alternative scenarios that genuinely differ from your baseline prediction, then test your strategy against each. This isn't about avoiding commitment but about understanding what could make you wrong and preparing appropriate responses.
• Distinguish Conviction Levels and Match Action Accordingly
  High conviction warrants concentrated commitment; low conviction suggests exploratory investments and structured learning. The key is matching resource allocation to conviction level rather than treating all decisions uniformly.
• Build in Monitoring and Adaptation Mechanisms
  Whatever predictions you make, establish clear signals that would indicate you're wrong and decision rules for how you'll respond. This addresses the valid concern that explicit prediction creates rigidity.
• Maintain Strategic Humility
  Even as you make predictions explicit, acknowledge the limits of predictability in your domain. For some decisions, building optionality and resilience matters more than prediction accuracy.

Conclusion

Harris and Toner have made an important contribution by highlighting that companies cannot avoid prediction—they can only avoid making their predictions conscious and testable. The exposure mapping framework provides a valuable tool for surfacing implicit strategic bets that often go unexamined.

But the strongest approach combines their call for explicit prediction with genuine humility about predictive limits. Leaders should make their key assumptions visible and test their conviction, as the authors recommend. But they should also build organizations capable of rapid learning and adaptation, recognizing that in periods of transformation, the ability to change course quickly may matter more than the accuracy of initial predictions.

The goal isn't to become better fortune tellers but rather to make better decisions under uncertainty. That requires the structured thinking that explicit prediction provides, combined with the humility to acknowledge when prediction must give way to experimentation, and the organizational capability to learn quickly from mistakes.

In the end, Harris and Toner are right that clarity is a superpower in times of transformation. But that clarity should extend not just to what we predict but also to the limits of our predictive capacity and the mechanisms we've built to thrive despite inevitable prediction errors. The companies that will navigate The Great Transformation most successfully aren't those with the most accurate predictions but rather those that combine conscious strategic bets with genuine organizational adaptability.

For more insights on predictions and navigating uncertain futures, explore this Bain article.