Bulletproof Problem Solving

Key Takeaways

1. Define the problem clearly and revisit it often

Good problem definition or framing can have a huge positive impact as we have shown in our examples.

Problem definition is critical. Take time to thoroughly understand the problem, its context, and the decision-maker's needs. Use a problem statement worksheet to capture key elements:

• Desired outcome
• Time frame
• Decision criteria and constraints
• Required accuracy
• Scale of aspirations

Be prepared to revise your problem statement as you learn more. Constantly iterate between your evolving understanding and new data or insights. This "porpoising" helps refine the problem and potential solutions.

Expand your perspective. Consider relaxing constraints or viewing the problem from different angles to unlock creative solutions. For example, the Avahan HIV project in India made a breakthrough by involving sex workers in developing solutions, rather than just focusing on traditional public health approaches.

2. Disaggregate complex problems into manageable parts

Any problem of real consequence is too complicated to solve without breaking it down into logical parts that help us understand the drivers or causes of the situation.

Use logic trees. Break down problems into components, factors, or hypotheses. Start with simple component trees and evolve to more sophisticated deductive or inductive logic trees as your understanding deepens. Aim for trees that are:

• Mutually Exclusive: Branches don't overlap
• Collectively Exhaustive: All relevant elements are included

Apply proven frameworks. For business problems, consider using:

• Price/Volume analysis
• Principal/Agent issues
• Assets/Options evaluation
• Collaborate/Compete strategies

For societal issues, explore frameworks like:

• Regulate/Incent approaches
• Equality/Liberty trade-offs
• Mitigate/Adapt strategies
• Supply/Demand analysis

3. Prioritize ruthlessly to focus on critical issues

Good problem solving is just as much about what you don't do as what you do, and good prioritization in problem solving makes solutions come faster and with less effort.

Use the 80/20 rule. Focus on the 20% of factors that drive 80% of the impact. Employ a 2x2 matrix to evaluate potential areas of focus:

• X-axis: Importance/Impact
• Y-axis: Ability to influence

Prioritize high-impact areas you can meaningfully affect. Be willing to prune off lower-priority branches of your logic tree to maintain focus on the critical path.

Conduct knock-out analyses. Quickly eliminate options or areas of inquiry that don't meet minimum thresholds. This prevents wasted effort on unproductive lines of investigation. For example, in a cost-reduction exercise, deprioritize units that can't meaningfully contribute to the overall target.

4. Create a structured workplan with clear hypotheses

Tools like Microsoft Project can be extremely helpful in workplanning. But they can also end up being like a beast that needs to be fed, generating enormous detail in workplanning that stretches out in time.

Develop hypothesis-driven workplans. For each analysis:

• State a clear hypothesis
• Specify the required output (e.g., chart type)
• Identify data sources
• Assign responsibility and deadlines

Use "chunky workplans" that focus on 2-3 weeks of critical analyses, coupled with lean project plans for overall milestones.

Employ one-day answers. Regularly summarize your current best understanding of the problem and potential solutions. Structure these in three parts:

1. Situation: Current state of affairs
2. Complication: Key tensions or changes
3. Resolution: Best current idea of the solution

This practice clarifies thinking, identifies gaps, and provides a basis for productive team discussions.

5. Start analysis with simple heuristics before complex models

Start all analytic work with simple summary statistics and heuristics that help you see the size and shape of your problem levers.

Leverage powerful heuristics. Use rules of thumb to quickly gauge direction and magnitude:

• Occam's Razor: Favor the simplest explanation
• Order of magnitude: Estimate relative sizes
• Compound growth: Use the Rule of 72
• Expected value: Probability x Impact
• Break-even analysis: Fixed costs / (Price - Variable costs)
• Marginal analysis: Consider incremental costs/benefits

Visualize data. Create simple charts and graphs to identify patterns and relationships. Heat maps, scatter plots, and histograms can reveal insights without complex statistical analysis.

Ask probing questions. Use frameworks like "5 Whys" or decision trees to structure your inquiry. For example, a three-question decision tree helped classify heart attack risk in patients more effectively than complex algorithms.

6. Employ advanced analytical tools when necessary

Which advanced tool you use is often dictated by whether you are seeking to understand drivers and develop an intervention strategy, or predict outcomes and plan accordingly.

Choose the right tool for the job:

• Multivariate regression: Understand complex relationships
• Bayesian statistics: Update probabilities with new information
• Randomized controlled trials: Test causal relationships
• Natural experiments: Leverage real-world quasi-experimental setups
• Monte Carlo simulations: Model complex, uncertain scenarios
• Machine learning: Identify patterns and make predictions from large datasets

Consider outsourcing. Platforms like Kaggle allow you to tap into a global pool of data scientists for specific analytical challenges.

Balance insight and prediction. Remember that while machine learning can produce powerful predictive models, it may not always provide clear explanations of underlying causal relationships.

7. Synthesize findings into a compelling narrative

Synthesizing brings together all the separate pieces of your analytic work and often yields new insights you didn't notice when you were in the weeds of analysis.

Use pyramid structure. Organize your key arguments and supporting evidence hierarchically:

• Top: Governing thought (main conclusion)
• Second level: Key supporting arguments
• Lower levels: Data and analysis backing each argument

Employ both inductive and deductive reasoning.

• Inductive: Build general principles from specific observations
• Deductive: Apply general principles to reach specific conclusions

Craft a compelling story. Structure your narrative to answer the decision-maker's key question: "What should I do?" Use techniques like:

• Situation-Complication-Resolution format
• Decision tree reveal for difficult conclusions
• Storyboarding to check the logical flow of your argument

8. Address uncertainty with flexible, iterative strategies

Uncertainty can be a good thing for strategic problem solvers! Hedge fund and other clever investors hope for uncertain and volatile markets—provided they have an analytic edge.

Assess uncertainty levels:

1. Clear enough future
2. Alternate futures
3. Range of futures
4. True ambiguity

Employ appropriate strategies:

• Buy information to reduce uncertainty
• Create strategic options
• Build capabilities for multiple scenarios
• Make "no regrets" moves
• Place "big bets" when you have high confidence

Use strategic staircases. For long-term uncertain environments:

1. Work backward from desired outcomes
2. Identify required capabilities and assets
3. Plan incremental steps to build knowledge and reduce risk
4. Remain flexible and adjust as new information emerges

9. Tackle "wicked problems" by reframing and systems thinking

Wicked problems are different, and we won't pretend they are easy to crack. But we believe the seven-steps approach sheds light on even these most difficult conundrums.

Characteristics of wicked problems:

• Multiple, interrelated causes
• Stakeholder disagreements on values
• Unintended consequences of interventions
• Require significant behavior change

Approaches for wicked problems:

• Reframe the issue to reveal new insights (e.g., using cost curves for climate change mitigation)
• Look for system-level interventions (e.g., addressing social networks in obesity)
• Find ways to internalize externalities (e.g., fishing rights in overfishing)
• Develop portfolios of strategies to address multiple facets

Example: Obesity. Consider holistic approaches:

• Link income, education, and health interventions
• Leverage social networks for positive influence
• Focus on high-leverage periods (pregnancy, early childhood)
• Redesign cities for walkability

10. Cultivate diverse teams and challenge biases

Good problem solving teams usually have an excellent lead or coordinator—somewhere between a musical conductor and an air traffic controller—who makes sure that the basic elements come together and on time.

Build diverse teams. Include members with varied backgrounds, experiences, and thinking styles. This enhances creativity and reduces blind spots.

Establish team norms:

• Hypothesis-driven, end-product oriented
• Frequent "porpoising" between hypotheses and data
• Seeking breakthrough thinking, not just incremental improvement
• Obligation to dissent
• Constructive confrontation

Combat cognitive biases:

• Confirmation bias: Actively seek disconfirming evidence
• Anchoring bias: Consider multiple reference points
• Loss aversion: Focus on opportunity costs, not sunk costs
• Availability bias: Look beyond readily available information
• Overoptimism: Conduct pre-mortem analyses

Encourage creative techniques:

• Role-playing to see different perspectives
• Analogical reasoning to apply insights from other domains
• Scenario planning to prepare for multiple futures

Last updated: January 23, 2025

Report Issue