Why Good Companies Kill Good Ideas

The engineer who built the world’s first digital camera worked for Kodak. Steve Sasson assembled it in 1975 — eight pounds, 0.01 megapixels, twenty-three seconds to record an image to cassette tape — and demonstrated it internally. Kodak went on to invest seriously in digital imaging research for decades, held foundational patents, and even fielded early consumer digital cameras. Then it went bankrupt in 2012, and the internet settled on its verdict: Kodak was too blind to see digital coming.

That verdict is wrong, and the way it’s wrong is the subject of this post — the opening of a series on how innovation actually survives inside large companies. Kodak saw digital coming with more clarity and earlier than nearly anyone else on earth. Its managers ran the numbers, compared a business with legendary gross margins on film and paper against a nascent digital business with thin margins and no razor-and-blades economics, and made the call any well-trained manager would make: protect the profitable core, treat digital as a hedge. Every individual decision was defensible. The sum of them was death.

That pattern — smart people, sound process, rational choices, catastrophic outcome — is what Clayton Christensen spent his career explaining, and The Innovator’s Dilemma (1997) remains the best structural account of it. The book’s claim is more unsettling than the pop version that circulates in pitch decks. It’s not that big companies fail to innovate because they’re slow, bureaucratic, or complacent. It’s that the very practices that make a company good — listening to customers, investing where returns are highest, killing weak projects early — are the mechanism by which it kills the ideas that would have saved it. The failure isn’t a bug in good management. It’s a feature, pointed the wrong way.

If that’s true, then the standard prescriptions — hackathons, culture decks, exhortations to “think like a startup” — are aimed at the wrong layer. You can’t culture your way out of a structural problem. You have to restructure, and the rest of this series is about how: where new bets should live in the organization, how to run the units that house them, who actually does the work, and how you fund and measure things that don’t fit a P&L. But the fixes only make sense once the diagnosis does, so let’s get the diagnosis right.

Why disk drives, of all things

Christensen didn’t start with a theory and hunt for anecdotes. He started with a puzzle and picked the industry that would let him solve it rigorously, and his choice is worth understanding because it’s what separates the theory from the buzzword it became.

He studied the disk-drive industry — every firm, every product, every generation, from the mid-1970s through the early 1990s — because disk drives were the fruit flies of business research. Product generations turned over every few years instead of every few decades, so a single study could observe many complete cycles of technological change and watch which firms lived and died through each one. Fast generations, complete data, repeated natural experiments.

And the pattern the data showed was strange. When the technological change was sustaining — better performance on the dimensions existing customers already valued, like more capacity in the same form factor — the incumbents almost always won, even when the change was technically radical. Established firms were excellent at hard engineering in service of their existing customers. But when the change was disruptive — a new architecture, typically a smaller drive, that was worse on the metrics existing customers cared about but opened a new market that valued something else — the incumbents lost, generation after generation. The 14-inch drive makers who owned the mainframe market missed 8-inch drives for minicomputers. The 8-inch leaders missed 5.25-inch drives for desktops. The 5.25-inch leaders missed 3.5-inch drives for portables. Each generation’s winners were mostly new entrants, and each generation’s losers had been the previous generation’s disruptors. Nobody stays smart for two rounds.

The mechanism, and this is the part the buzzword usage flattens, is about trajectories. Technology improves faster than any given market’s ability to absorb the improvement. So a product that starts out laughably inadequate for the mainstream — the 3.5-inch drive was small, low-capacity, expensive per megabyte — can serve some marginal market (laptops) that values its other attributes (ruggedness, size, power draw), ride the improvement curve, and arrive at “good enough” for the mainstream while the incumbent’s products have overshot into performance nobody will pay extra for. Disruption is a story about two lines crossing on a graph: the trajectory of performance improvement versus the trajectory of market demand. It is not a synonym for “startup,” “digital,” or “thing I find impressive.” Uber, to take Christensen’s own famous 2015 correction in HBR, was by his definition not disruptive to taxis — it entered the mainstream market with a better service, a sustaining move that happened to come from outside. The word gets used ten thousand times a day and means something specific about once.

The murder weapon is the resource-allocation process

Here’s the question the disk-drive data forces: the incumbents saw every one of these transitions. They frequently had prototypes of the next-generation drive in their labs before the entrants shipped. Why didn’t they act?

The answer Christensen documented is quietly devastating for anyone who imagines innovation dies in dramatic boardroom vetoes. It doesn’t. Ideas die in the resource-allocation process — the everyday, mostly invisible mechanism by which middle managers decide which projects get engineers, funding, and their own career capital. No executive ever has to say “kill the small disruptive project.” The organization’s immune system handles it several layers down, one reasonable prioritization decision at a time.

Think about the incentives of a capable middle manager choosing which projects to champion. Backing a project for your biggest customers, with a forecastable market and healthy margins, is good for the company and good for your career. Backing a project for a market that doesn’t exist yet, whose customers can’t be surveyed because they aren’t customers yet, whose margins look worse than the core’s, and whose honest revenue forecast rounds to a rounding error — that’s a career gamble with no institutional support. So the disruptive project doesn’t get vetoed; it gets starved. It’s ranked eleventh on a list where the top ten get staffed. Its best engineers get pulled onto the big-customer escalation. Its funding survives, technically, at a level guaranteeing irrelevance. In my experience the tell is always the same: leadership sincerely insists the initiative is a priority, and the people assigned to it can’t get a designer for two quarters. Watch where the next contested engineer actually lands — that’s your real strategy, and no offsite changes it.

Two of Christensen’s sharpest lines describe the jaws of the trap. First: small markets don’t solve the growth needs of large companies. A $50 million emerging market is a company-making opportunity for a startup and a distraction for a $10 billion incumbent that needs to find $800 million of new revenue this year just to keep growing at 8%. The bigger and more successful you are, the more rational it becomes to ignore every market while it’s still small — which is the only time you can enter it cheaply.

Second, the one that inverts the most sacred rule in business: listening to your best customers is what leads you off the cliff. Your best customers are, by construction, the ones most invested in your current trajectory. Ask them about the disruptive thing and they’ll tell you, accurately, that it’s worse — slower, lower-capacity, lower-quality — and that they’d rather you improved what they already buy. They’re not lying. The disruptive product genuinely is worse for them, today. Seagate’s marketers showed 3.5-inch prototypes to their desktop customers, heard “not interested, we want cheaper 5.25-inch capacity,” and shelved the program — a textbook-correct customer-driven decision that handed the portable market to Conner Peripherals. Customer-centricity is a sustaining-innovation superpower and a disruptive-innovation blindfold, simultaneously.

Resources, processes, values — the asset that disqualifies you

Christensen’s later refinement (with Michael Overdorf, in the 2000 HBR piece “Meeting the Challenge of Disruptive Change”) explains why the trap isn’t escapable by hiring better people. An organization’s capabilities live in three places: its resources (people, cash, technology, brand), its processes (the patterns by which it turns resources into value — how it does market research, budgeting, development), and its values (not ethics — the criteria by which everyone from the CEO to the sales rep decides what’s worth doing, most importantly what gross margins are acceptable and what market size is interesting).

Resources are flexible — you can move people and money to a new opportunity, and this is why executives keep believing they can will an innovation into existence by assigning stars and budget to it. Processes and values are not flexible, and here’s the cruel symmetry: they can’t be, because their inflexibility is precisely what makes them valuable to the core business. A process is, by definition, a way of doing something repeatedly and consistently; a process that changed every time would not be a process. A margin threshold is only useful as a decision filter if it actually filters. A company that has spent twenty years perfecting a stage-gate process for forty-percent-margin products sold through enterprise channels has built a machine that reliably produces exactly that — and that machine will process a low-margin, unknown-market, wrong-channel idea the way any well-built filter processes something outside its specification. It rejects it. Correctly, by its own lights.

This is why “we acquired a startup and integrated it” so often destroys exactly what was purchased: the acquirer maps the startup’s resources into its own processes and values, and the resources were never where the magic lived. And it’s why the RPV frame turns the whole problem from a morality tale into an engineering problem. The question stops being “why is our culture so anti-innovation?” — it usually isn’t — and becomes “which of our processes and values disqualify this idea, and where could the idea live so that they don’t apply?” That second question has real answers, and they’re structural. The ambidexterity literature is essentially a body of evidence about which structures let one company run two sets of processes and values at once without the stronger one digesting the weaker.

Two autopsies the pop history gets wrong

Kodak, properly told. The lazy version — film executives too dim to see digital — collapses on contact with the record. Kodak’s own 1981 internal assessment of digital’s threat (commissioned after Sony announced the Mavica) was strikingly accurate about the technology’s trajectory and timing; the company poured billions into digital R&D over the following decades; by the early 2000s its EasyShare line was among the best-selling digital cameras in the US. Kodak did the innovation. What it could not do was commit to a business that cannibalized the most profitable product economics of the twentieth century — film’s consumables model, where the camera was nearly a loss leader and the recurring margin lived in film, paper, and chemistry. Every dollar of digital success was a dollar of high-margin film revenue destroyed, and the resource-allocation process priced that trade honestly every quarter. Willy Shih, who ran Kodak’s digital business and later wrote the inside account for MIT Sloan Management Review, is blunt that the problem wasn’t perception but the economics of substitution: digital offered razor-thin hardware margins governed by commodity semiconductor dynamics, replacing a vertically integrated annuity. Kodak’s processes and values, world-class for the film business, evaluated digital accurately — and accurately concluded, year after year, that it was a worse business. It was. It was also the only business that was going to exist.

Blockbuster, properly told. The lazy version says Blockbuster laughed Netflix out of the room in 2000 (the famous declined $50 million acquisition) and dozed until bankruptcy. The interesting version is what happened next: Blockbuster’s management did respond, and the response was good. Under CEO John Antioco, the company launched Blockbuster Online in 2004, and in 2006 rolled out Total Access — rent online, return in any store — which used the store network as a weapon Netflix couldn’t match. It worked. Netflix’s subscriber growth wobbled; Reed Hastings reportedly approached Blockbuster about a deal; Netflix’s own executives later acknowledged how dangerous the program was. Then Blockbuster’s own structure killed it. Total Access and the elimination of late fees were enormously expensive bets against current earnings — late fees alone had been a material chunk of profit — and the franchise system meant thousands of store owners whose economics the online pivot directly attacked. Activist investor Carl Icahn, on the board since 2005, fought Antioco over the strategy’s cost; the fight ended with Antioco’s exit in 2007, and his successor Jim Keyes explicitly re-centered the company on the retail stores and scaled back the online push. Blockbuster filed for bankruptcy in 2010. The company didn’t lack the idea, the strategy, or even the executive will — it had all three. The idea was killed by the values of the institution: the earnings expectations, the franchise economics, the board’s fiduciary attachment to the current P&L. Which is to say it was killed by the resource-allocation process, operating exactly as designed, one level above the CEO.

Notice what both autopsies share. In neither case was the fatal deficiency vision, technology, or talent. In both cases the disruptive response was invented inside the incumbent and then dismantled by the incumbent’s own — entirely functional — machinery for protecting its core economics. The call was coming from inside the house, and the house was working correctly.

The dilemma is the point

I want to close the loop on why Christensen called it a dilemma rather than a mistake, because the distinction carries everything. A mistake implies a right answer that was missed. A dilemma means both horns gore you. If Kodak had aggressively cannibalized film in 1985, it would have traded the best margin structure in consumer goods for a commodity hardware business decades early, and its board would have — reasonably — fired the CEO who did it. If Blockbuster’s franchisees had been forced to eat Total Access’s costs, they’d have revolted, and Icahn’s earnings case against Antioco was not irrational. The disturbing conclusion of the disk-drive study was that Christensen could find no management error to correct: the failed incumbents were, by conventional measures, the best-managed firms in their industry. That’s the finding that should reorganize how you think about this problem. When good management reliably produces the failure, more management quality is not the cure.

Neither is culture, which is where most corporate innovation energy goes to die. Posters about bold thinking do not change the margin threshold in the business case template. An empowerment workshop does not change whose customers the roadmap prioritization meeting listens to. If the killing mechanism is structural — allocation processes, margin-based values, growth arithmetic — the countermeasures have to operate at the same layer: separate units with their own processes and values and the executive wiring to survive contact with the core; programs that give individual employees allocation power the middle layer can’t intercept; funding and accounting mechanisms that let a small market be judged as a small market rather than starved for failing to be a large one. Those are, in order, the subjects ahead — starting with the oldest structural question of all: where in the organization a new bet should live so the antibodies can’t reach it.

The dilemma doesn’t dissolve. Kodak’s managers and Blockbuster’s board would recognize the pressures in your company’s next planning cycle, because the pressures are the same and they are legitimate. The only durable move is to build structures that don’t ask good managers to act against their own correct judgment — structures where the disruptive bet is somebody’s core business, however small. Everything else is asking the immune system to please stop working.

Further reading

• Clayton Christensen, The Innovator’s Dilemma (1997) — read the disk-drive chapters, not just the summary; the method is the argument.
• Christensen, Raynor & McDonald, “What Is Disruptive Innovation?” (HBR, 2015) — the authors’ own correction of twenty years of misuse, including the Uber analysis.
• Christensen & Overdorf, “Meeting the Challenge of Disruptive Change” (HBR, 2000) — the RPV framework, and the most practically useful twelve pages in the canon.
• Willy Shih, “The Real Lessons From Kodak’s Decline” (MIT Sloan Management Review, 2016) — the inside account from the man who ran Kodak’s digital business.