Why Good Management Fails

The idea in plain words

The companies that lose to new technologies are rarely badly run. They are often the best-run companies in their industry. They listen closely to their most profitable customers, invest in improving products those customers want, and avoid small markets with thin margins. That is what good management means. And it is exactly what leaves them exposed when a cheaper, simpler technology, at first good enough only for customers they do not care about, improves until it is good enough for everyone.

Clayton Christensen called this the innovator’s dilemma. Its tragedy is that the incumbent’s failure follows from its virtues.

Where it comes from

In 1942 the economist Joseph Schumpeter described capitalism as a process of creative destruction: new goods, new methods and new forms of organisation continually destroy old ones. Competition, he argued, comes less from rivals lowering prices than from new combinations that make existing firms obsolete. Schumpeter saw this as the essential fact of capitalism, and as a reason why large firms’ advantages were always temporary.

Christensen, a former consultant studying at Harvard Business School, asked why incumbents so often failed to ride these waves. His research focused on the disk drive industry, where successive generations of smaller drives, 14-inch, 8-inch, 5.25-inch, 3.5-inch, each displaced the leaders of the previous generation. The leaders were technically capable of building the new drives. They did not, because their customers, makers of larger computers, did not want them, and the new markets were too small to matter to a large company.

Christensen distinguished sustaining innovations, which improve products for existing customers and which incumbents usually win, from disruptive ones, which start out worse on the dimensions mainstream customers value but are cheaper, simpler or more convenient, and which appeal first to fringe or new customers. Because performance improves faster than mainstream customers’ needs grow, the disruptive product eventually becomes good enough, and the incumbent’s customers move.

Rebecca Henderson and Kim Clark had shown a related mechanism in 1990: architectural innovations, which reconfigure existing components in new ways, defeat incumbents because the incumbents’ organisation and knowledge are built around the old architecture.

What it does to the mind

The dilemma works through resource allocation, which is where strategy actually happens. Every manager competes for budget by showing returns, and the new, small, low-margin opportunity always loses to the established, large, high-margin one. No single decision is wrong. The sum of correct decisions is a strategic blind spot.

There is a perceptual side as well. Disruptive products look like toys to experts. Their flaws are obvious, their customers seem unimportant and their margins are unattractive. The very expertise that makes incumbents good makes it hard for them to take the newcomer seriously.

Where it shows up

The operating systems essay and the compute essay describe repeated cases of smaller, cheaper machines displacing larger ones, from minicomputers to PCs to phones. The Lisp machines road is an example from the other side: superior specialised machines displaced by cheaper general-purpose workstations. Microsoft’s Internet Tidal Wave memo and Nadella’s “Bold Ambition” are attempts by an incumbent to escape its own dilemma, a decade and a half apart.

What it asks of someone deciding

Look below you. Ask which customers you serve badly or not at all, and who is serving them with something cheaper.

Separate what the core cannot do. Christensen’s own prescription was to place disruptive bets in independent units with cost structures and metrics suited to small markets, so they are not starved by the core business’s priorities.

Take the objection seriously. Later studies, including one by Andrew King and Baljir Baatartogtokh, found that many celebrated examples fit the theory poorly. Use the dilemma as a lens for seeing threats, not as a law that predicts them.