Most discussions around decision making focus on individual decisions, as if decisions are homogeneous and standalone. The truth is, every person, every team and every organisation is making a large number of decisions with different risk-return profiles, together forming what I call a decision portfolio.

This is easier to understand in investing, where we generally agree that a well-constructed portfolio should include both safe assets with predictable returns and risky assets with outsized upside potential. The mix depends on the goal, whether it’s aiming for preservation or growth.

Business decisions resemble asset class returns remarkably well as each decision carries different risk-return trade-offs. I’d argue that in business, it’s far rarer to see a well-constructed decision portfolio than in the world of investment management. Part of this is probably a lack of conscious thinking around the idea itself. But the structural tendency is also very clear.

When AI blasts us with threats and opportunities, the need for decision portfolio optimisation becomes all too clear.

The risk in investment management is usually quantified as the volatility of returns. In business, risk is better understood as the uncertainty of outcomes.

There are decisions with a high degree of certainty, such as automation or continuous improvement initiatives. These are usually consensus bets. They improve efficiency, reduce operational friction and create reliability, but the payoff tends to be limited because everyone else can see the same opportunity. Then there are decisions with much higher uncertainty, but when they work, the outcomes can be generational.

Intel’s decision to shift away from memory chips and focus on microprocessors looked risky at the time, but it fundamentally changed the trajectory of the entire industry. Amazon building AWS initially looked like an odd distraction from retail. Netflix moving aggressively into streaming cannibalised its own existing business before the market was fully ready. These decisions often appear irrational or uncomfortable in the moment precisely because the upside is uncertain and the path is unclear. If the opportunity were already obvious, the returns would likely already have been competed away.

One important difference between investment management and business decision making is that investment managers are largely at the receiving end of risk and reward. Regardless of who manages the portfolio, the underlying risk-return profile of an asset does not fundamentally change.

However, business is different. In business management, the quality of strategy and execution can dramatically influence both risk and payoff. Great organisations are not simply choosing between fixed opportunities. They actively reshape the probability distribution of outcomes.

A strong team can reduce unnecessary risk through better execution, faster learning, tighter feedback loops and clearer strategic focus. In other words, talent density matters because exceptional people are often able to remove risks that carry little payoff, leaving behind only the irreducible risks associated with genuine innovation and asymmetric opportunities. That is where breakout outcomes often come from.

The other thing I noticed was that the defining returns of great companies rarely come from consensus decisions. They usually come from a small number of low-P decisions that looked uncertain, uncomfortable or even irrational at the time.

If low-P decisions are responsible for so much long-term value creation, why are they so rare inside large organisations?

Interestingly, the fate of most products and services is commoditisation.

This feels almost inescapable, but I could not help wondering why. As my understanding deepened, I arrived at a surprising realisation: commoditisation and bureaucracy share the same root cause of consensus-seeking. And consensus-seeking is an inevitable tendency as industries and organisations scale.

When an organisation matures, a standard playbook gradually gets written and followed, best practices emerge, processes solidify, and coordination costs rise. Over time, the common denominator shrinks to the point where

The result is often surface-level optimisation where organisations become increasingly efficient at competing in areas where everyone already knows how to compete. This is how industries slowly drift toward commoditisation.

Andrew Chen described this tendency well:

Consensus is not irrational. In large organisations, consensus reduces friction. It reduces visible failure. It creates alignment across increasingly complex systems. The problem is that consensus is also deeply correlated with crowdedness. Once an idea becomes widely accepted internally, competitors are often already pursuing similar opportunities externally. The upside becomes competed away.

This creates an uncomfortable reality: group settings are usually not the natural soil for contrarian ideas. A non-consensus idea often appears socially irrational long before it appears economically irrational. Not aligning with the group can look like being unreceptive to feedback. Challenging the prevailing view can create interpersonal friction without any immediate visible payoff for the individual taking the risk. As a result, many low-P decisions die before they are properly explored.

Marc Andreessen captured this dynamic succinctly:

The same increasingly applies to business strategy itself. Once a decision becomes obvious consensus, its returns are often already commoditised. There are probably two broad ways organisations can counteract this tendency.

The first is allowing small autonomous teams to own outcomes.

In my previous article Scaling Growth Team in the AI Era, I described the idea of small strike teams reducing coordination debt. Small teams can pursue unconventional ideas without requiring organisation-wide consensus upfront. Most experiments will fail, but the cost of experimentation remains bounded while the upside can be disproportionately large.

However, structure alone is insufficient. Incentives matter too. Not every manager’s remuneration can realistically be tied directly to long-term commercial outcomes, but at minimum, psychological safety must exist for people willing to risk looking foolish in pursuit of asymmetric outcomes. This is often easier said than done, especially in environments with strong consensus cultures and tall poppy syndrome dynamics.

The second mechanism is talent density. Talent density is almost a magical concept repeatedly mentioned by founders of legendary companies. Exceptional people tend to understand that the defining opportunities in business are often exceptions themselves. When talent density is high, non-consensus thinking can become surprisingly common because strong individuals are better able to distinguish genuine risk from social discomfort.

This may explain why greatness tends to cluster in certain organisations. The most exceptional organisations are often not merely better at execution. They are better at allowing intelligent non-consensus ideas to survive long enough to compound.

In investment management, the concept of the efficient frontier describes the optimal mix of assets that maximises expected return for any given level of risk.

The idea is again surprisingly applicable to business decision making. A company making only high-P decisions may achieve operational stability, but it also risks converging toward commoditised consensus behaviour. On the other hand, a company making only low-P decisions will likely descend into chaos, constantly pursuing asymmetric upside without enough stability to sustain itself.

The goal is therefore not maximising certainty, nor maximising risk-taking, but rather constructing a decision portfolio that produces the highest long-term asymmetric upside while remaining survivable. This is effectively the efficient frontier of business decisions. The interesting thing is that the frontier itself is not fixed.

A growth-stage company should probably have a much higher concentration of low-P decisions in its portfolio. At that stage, survival itself often depends on discovering non-consensus opportunities before competitors do. Without asymmetric bets, a young company risks becoming just another participant in an already crowded market. A stable profitable company, on the other hand, can afford to hold a much larger mix of high-P decisions. Operational excellence, reliability and incremental optimisation become increasingly important once scale is achieved. In many cases, this is exactly what shareholders expect.

However, the frontier itself is shifting because AI is rapidly commoditising knowledge, workflows and even execution capability. Paradoxically, this is happening at the same time large organisations command more power than ever. This creates a strange dynamic. Scale has never mattered more, yet the risk of strategic convergence has also never been higher. The same AI tools, benchmarks and best practices spread across industries almost instantaneously, pushing organisations toward increasingly similar behaviours and compressing returns further on high-P decisions. This is what I believe sits at the core of the SaaSpocalypse.

As AI accelerates the spread of knowledge and best practices, consensus forms faster than ever. Truly non-consensus opportunities therefore become rarer, but also more valuable. In other words, as the left side of the efficient frontier becomes increasingly crowded and compressed, the asymmetric upside on the right side expands. In a world drifting toward strategic convergence, the greatest rewards may increasingly flow to organisations capable of making intelligent low-P decisions before they become obvious to everyone else.