Why Good Companies Make Bad Decisions

Key takeaways

• Capable companies often fail not through incompetence but by doing more of exactly what made them successful. Clayton Christensen called this the innovator's dilemma.
• His account is contested. The historian Jill Lepore argues it rests on cherry-picked cases and shaky history, and that several of the disrupters he celebrated went on to fail themselves.
• Whatever you make of the theory, the day-to-day mechanism is easy to recognise: the processes that serve your best customers and protect your margins are the same ones that reject the cheap, scrappy bets that later matter.
• Dan Davies' idea of the accountability sink captures the other half of the problem: structures that soak up responsibility until decisions get made that no single person would own.
• The real danger is rarely bad people. It is good process tuned for yesterday, sitting inside a structure where nobody is accountable for the slow wrong turn.

Capable companies often fail by doing what made them succeed

The comforting story about corporate failure is that the people in charge were fools. It is usually not true, which is what makes the subject worth taking seriously. Plenty of well-run companies, staffed by intelligent people making defensible decisions, walk steadily into decline. The previous tier of this series looked at how individuals stall on data and misjudge their own decisions. This post moves up a level, to how a whole organisation, full of capable people, can make a chain of sensible choices that add up to a bad one.

Christensen's diagnosis: good management is the root cause

Clayton Christensen's "The Innovator's Dilemma" argued that great firms are often toppled not because their managers made bad decisions but because they made good ones, the same disciplined, customer-focused decisions that had made them great for decades. Mainframe makers listened to their mainframe customers, improved their mainframes, and protected their healthy margins, all textbook good management, and in doing so they missed the personal computer, a worse and cheaper product that served customers they did not value, until it grew up and ate the industry. Christensen's uncomfortable phrase was that "doing the right thing is the wrong thing." The processes that make an established company succeed, listening to customers, chasing higher-margin products, backing the markets that look substantial, are precisely the processes that lead it to reject a disruptive technology that offers lower profit and underperforms today.

Lepore's objection: the theory is tidier than the history

It would be dishonest to present that as settled, because it is not. The historian Jill Lepore wrote a pointed critique arguing that disruption theory behaves more like a faith than a finding. Many of the entrant firms Christensen celebrated as triumphant disrupters did not endure; their success was in some cases brief and in others illusory. She questioned the case selection, the history, and the way the theory has become close to criticism-proof, since doubters can be dismissed as enemies of progress. Her deeper claim is that "disruption" became the early twenty-first century's anxious theory of history, an account of change founded on a fear of financial collapse and surprisingly thin evidence. You need not accept all of it to find it a healthy corrective to a theory that often gets quoted as law.

The everyday mechanism survives the academic fight

Here is what interests me as someone who works with operators rather than theorists. Whoever wins the academic argument, the mechanism underneath is real and observable. I watched a healthy ecommerce brand run a small version of the innovator's dilemma in about eighteen months. Their best customers wanted more of the premium hamper they already loved, so the team, sensibly, kept improving the premium hamper. A cheaper, simpler subscription box kept getting proposed and kept dying in planning, because it had thinner margins, served a less valuable customer, and made the headline figures look worse in the short term. Every individual decision to deprioritise it was defensible. A competitor built exactly that box, used it to acquire a generation of younger customers cheaply, and grew past them inside two years. Nobody was stupid. The process was doing its job, and its job was to protect the present.

Accountability sinks: where decisions lose their owner

Dan Davies, in "The Unaccountability Machine", supplies the half that Christensen leaves out. Large systems, he argues, regularly produce decisions that no individual inside them actually intended or wanted, because responsibility has been spread across so many hands that it lands in nobody's. He calls the structures that do this accountability sinks: arrangements that absorb negative consequences and complaints so that no specific person ever has to own them. The airline that strands you has built one deliberately. The staff you can reach have no power to fix it, and the people with the power are arranged so that they never have to meet you. Inside a company, the same sink swallows the slow wrong turn. Ask afterwards who decided to let the subscription box die and you will find that no one did, exactly. It died in the gaps between a budget meeting, a roadmap review, and a quarter when everyone was busy.

The danger is good process plus no owner

Put the two ideas together and you get a fair picture of how good companies make bad decisions. Good processes, optimised for the customers and margins you already have, quietly screen out the bets that threaten the present. And accountability sinks ensure the screening happens without anyone deciding it, so there is no single moment to point at and no single person to argue with. This is why telling a company to "be more innovative" rarely works. The problem is structural rather than a failure of will. Tony Blair once described government as "a conspiracy for inertia", and large companies grow their own version, in which the safe path is always to keep serving the known customer and let the awkward idea expire from neglect.

There are partial defences. Gary Klein points out that insights are inherently disruptive, that they make things less smooth, and that organisations therefore tend to suppress them without meaning to, so protecting the awkward idea has to be deliberate. You can give a small, threatening bet its own owner, its own budget, and shelter it from the metrics that would kill it in its first quarter, the way the previous post argued you sometimes have to. You can revisit the decisions nobody remembers making, and ask who actually owns them now. None of this is a cure. The forces are strong and they reassert themselves the moment attention drifts.

Helping founders see these patterns before they harden is a fair description of what we do at Mean Decisions: not handing over a framework so much as sitting in the actual decisions, the box that keeps dying, the bet nobody will own, and naming what is really happening so the team can choose with their eyes open. What I cannot promise, and would not, is that naming it is enough to beat it. The most honest thing I can say is that the companies which resist this fate are usually the ones willing to keep having the uncomfortable conversation about who decided, and why, long after everyone would rather move on.