Good Management Is the Root Cause

The Core Insight

Christensen catalogued every disk drive model sold between 1975 and 1994 and counted 116 new technologies. Established firms led in all 111 sustaining ones. The other five were disruptive, and the leaders lost every one of them.

The Innovator's Dilemma, published in 1997 from that database, argues one uncomfortable claim. Good management was the most powerful reason the leading firms failed. They listened to their best customers, invested where returns looked largest, and lost by doing exactly that.

There are two kinds of technology in the book. Sustaining technologies improve a product on the dimensions mainstream customers already pay for. Disruptive technologies arrive worse on those dimensions and better on ones the mainstream ignores: cheaper, simpler, smaller, more convenient. The trap closes because technology improves faster than customer need grows. Within a few years, the fringe product meets the mainstream requirement.

Most managers blame such failures on bureaucracy, arrogance, or slow engineering. Christensen argues the failures concentrated in the best-run firms. The discipline that made them the best rejected each disruptive project until the contest was over.

The Framework

The unit of analysis is the value network: the customers a firm serves, their ranked priorities, and the cost structure that serving them requires. Mainframe 14-inch drive makers needed gross margins near 60 percent, minicomputer 8-inch makers near 40, desktops 25, portables 15 to 20. A company tuned for 60 percent reads a 25 percent market as a pay cut, and kills proposals aimed there without instruction.

Two lines on one chart carry the argument. Performance demanded climbs on a shallow slope. Performance supplied climbs on a steep one. Past the crossing, the incumbent overshoots and climbs toward richer margins. The disruptor enters the vacated floor and rides the same steep slope into the incumbent's market.

Five principles govern the collision.

• Companies depend on customers and investors for resources, so allocation processes evolve to kill whatever the paying base rejects.
• Small markets do not solve the growth needs of large companies.
• Markets that do not exist cannot be analyzed, and expert forecasts about them are always wrong.
• Capabilities live in processes and values, and the same two define what an organization cannot do.
• Technology supply can outrun market demand, so today's inadequate product meets tomorrow's requirement.

Key Ideas

Every Drive Generation Toppled Its Leaders

The casualty list is exact. Of the seventeen drive makers active in 1976, every one but IBM's operation failed or sold by 1995. Of 129 later entrants, 109 failed.

Mainframe buyers absorbed 15 percent more capacity a year while 14-inch technology supplied 22. Entrants such as Shugart, Micropolis, Priam, and Quantum sold 8-inch drives of 10 to 40 MB into minicomputers, while mainframes demanded 300 to 400. Two-thirds of the 14-inch makers never shipped an 8-inch model, and every one left the industry. Seagate then opened the 5.25-inch generation in 1980 with 5 and 10 MB drives that bored minicomputer buyers wanting 40 and 60.

Seagate's own engineers built the industry's second working 3.5-inch drives in 1985, some eighty prototypes, before anyone sought formal approval. Marketing showed them to IBM, forecast under 50 million dollars, and executives shelved the program for a 5.25-inch line earning 300 million a year. Conner Peripherals shipped the small drive in 1987 and booked 113 million dollars in year one, a United States manufacturing record. Seagate entered in 1988, after the industry shipped nearly 750 million dollars of them. Its drives went to desktop customers rather than portables.

The 1.8-inch generation repeated it. By 1995, entrants held 98 percent of that 130 million dollar market, and the largest first application was portable heart monitors.

Excavators and Steel Ran the Same Script

Hydraulic excavators appeared in 1947 lifting a quarter cubic yard, while cable shovels moved one to five yards per scoop. Entrants sold them to residential builders, and hydraulics reached two cubic yards by 1965 and ten by 1974. Of roughly thirty cable shovel makers, four survived the transition. Bucyrus Erie aimed a cable-hydraulic hybrid at existing customers in 1951, and it failed. The leaders logged record profits until 1966, the year the lines crossed.

Steel ran slowest and clearest. A minimill melted scrap at about 15 percent lower cost than an integrated mill, and building one cost about 400 million dollars against 6 billion. Minimills took rebar first, then bars, then beams, then sheet. Their North American share went from nothing in 1965 to 19 percent in 1975, 32 in 1985, and 40 in 1995. Not one integrated producer in the world built a minimill in those thirty years.

Sheet steel compressed the book into one decision. Thin-slab casting arrived in 1987 at about one-tenth the capital of a conventional sheet mill, with surface blemishes premium buyers rejected. Bethlehem and USX studied it and bought conventional casters instead, to serve buyers who paid for perfect surfaces. Nucor, with no such customers to protect, poured first at Crawfordsville in 1989 and held 7 percent of the sheet market by 1996.

Your Best Customers Veto the Future

Resource dependence names the engine. Customers and investors supply the money, so allocation evolves to serve them. Interviews with more than eighty industry managers yielded one six-step script. Engineers inside the incumbent build the disruptive prototype first. Marketing shows it to flagship customers, who reject it. The forecasts come back small, and the firm accelerates sustaining work instead. Frustrated engineers quit, start the entrant, find a market by trial and error, and march upmarket. The incumbent ships its shelved design in self-defense and wins nothing new.

The 1.8-inch drive shows the veto working below the CEO. In 1994 the head of one large maker showed Christensen a fourth-generation 1.8-inch drive and declared the market absent. Disk/Trend measured that market at 40 million dollars for 1993 and projected 80 million for 1994. A month later one of his students described buying 1.8-inch drives for car navigation from a small Colorado startup. The big makers stocked the drive, and their computer-account salespeople had no reason to sell it.

Small Markets Cannot Feed Large Companies

The growth arithmetic is short. A 40 million dollar company grows 20 percent by finding 8 million dollars. A 4 billion dollar company needs 800 million, and no emerging market is that size. Waiting costs the lead: entrants in a network's first two years succeeded six times as often as later ones.

A census of 83 entrants prices the choice. Firms that attacked established markets reached 100 million dollars in annual revenue 6 percent of the time. Firms that led into new markets got there 37 percent of the time. Leaders booked 62 billion dollars of cumulative revenue against 3.3 billion for followers, and 1.9 billion per firm against 64.5 million.

Scale also grades the outcomes. Apple sold 43,000 Apple II units in its first two years and went public on the result. The Newton sold 140,000 units across 1993 and 1994, three times that pace, and observers called it a flop. At 5 billion dollars of revenue, a win that size cannot register.

The Cure Is a Separate Company

The survivors share one structure. Quantum missed the 5.25-inch generation, financed its departing employees, kept 80 percent of their spinoff, and left it alone. The spinoff's revenue replaced the dying core, and its executives took over the parent. By 1994 the rebuilt Quantum led the world in drives shipped.

Control Data ran its 5.25-inch program from Oklahoma City, away from the Minneapolis mainframe business, where a million dollar order turned no heads. The remote unit celebrated 50,000 dollar orders and took a 20 percent share of high-capacity 5.25-inch drives.

Kresge and Woolworth entered discounting in 1962, three months apart. Kresge built Kmart apart and closed a tenth of its variety stores each year to feed it. Woolworth ran Woolco inside the parent, whose 35 percent margin rules pulled the discounter's margins up and its inventory turns down. A decade in, Kmart's sales neared 3.5 billion dollars against Woolco's 0.9 billion, and Woolco died in 1982. One condition rides along: in every success the chief executive personally protected the unit's right to different margins, with no reported exception.

Overshoot Resets What Buyers Pay For

Performance oversupply triggers the invasion and re-prices the product. By 1988 the average 5.25-inch drive exceeded desktop capacity demand by nearly 300 percent. Desktop makers paid 20 percent more per megabyte to get the smaller 3.5-inch drive. Within four years it carried 60 percent of desktop sales. The premium then vanished: a cubic inch of size reduction commanded 4.72 dollars in 1986 and 6 cents in 1989.

The basis of competition steps down a fixed ladder: functionality, reliability, convenience, price. Contractors switched to hydraulics once bucket size sufficed, because a snapped cable kills. Intuit's Quickbooks stripped the accounting rigor out of small business software and took 70 percent of that market within two years. Eli Lilly spent close to 1 billion dollars perfecting 100 percent pure insulin and met a tepid market at a 25 percent premium. Novo's insulin pen cut a syringe routine of one to two minutes to ten seconds and sustained a 30 percent premium.

Disk/Trend forecast sustaining generations within 8 percent and missed two disruptive ones by 265 and 550 percent. Markets that do not exist cannot be analyzed, and the expert number is always wrong.

Practical Applications

Run the disruption test first. Plot what your median customer absorbs each year against what the technology supplies. Christensen ran it on the electric car, after California mandated that electrics reach 2 percent of 1998 sales. In 1997 drivers needed 125 to 150 miles of range, and electric vehicles delivered 50 to 80. Range demand grows under 1 percent a year, and the technology improves 2 to 4 percent a year. Parallel lines mean a niche. Converging lines mean a countdown.

Sell the product where its weaknesses read as strengths. Chrysler packed 1,600 pounds of batteries into a minivan and produced a 100,000 dollar vehicle against a 22,000 dollar gasoline twin. The demand for a battery breakthrough was the tell. Breakthroughs turn necessary only when a disruptive product aims at the mainstream. Christensen's alternative is agnostic marketing: assume nobody, customer included, knows the application, and run cheap expeditions until real buyers appear. His candidate markets were teenagers' first cars and taxi fleets in Southeast Asian traffic, where slow acceleration and short range cost nothing.

Plan to learn, and budget to be wrong. Discovery-driven planning writes down the assumptions under the forecast and tests the deadliest ones before capital commits. Honda entered America in 1959 to sell big road bikes, and the engines leaked oil on the highway. The market that worked appeared when its staff rode Supercubs in the dirt on weekends. Honda kept enough reserve to chase the accident, and by 1975 the market it opened reached 5,000,000 units a year. Hewlett-Packard's Kittyhawk spent its whole budget on a palmtop market that never arrived. Video game makers then wanted a 50 dollar version and found nothing left.

Give the project a home whose survival depends on it. Spin out when the values do not fit, meaning margins too low or a market too small for the mainstream to rank first. Keep the work inside when the change sustains the current business, because incumbents win those fights. Fund the unit to stay hungry, so small orders feel like wins and positive cash arrives before patience runs out.

Who This Is For

Founders attacking incumbents get the most, because this is the attacker's manual wearing a defender's cover. Entrants won every disruptive transition in the study, protected by the incumbent's rationality. Operators inside successful companies get the mirror lesson. The upmarket drift is unconscious, and improving margins during a losing war is its signature.

Skip it if this quarter's problem is selling a known product to known buyers. The book picks where to compete and how to organize, never how to sell.

This is 1997 research with a 2000 revision, so date-stamp everything. The evidence is one industry's data plus case studies chosen after their outcomes were known. The mechanism aged well. Minimills kept taking share, and low-end entry into overserved markets became the standard lens on incumbent failure. Specific calls aged worse. Flash memory, waved off here as remote from the drive makers' networks, later ate their business from below. Kmart, the featured discounting winner, lost the following decades to Wal-Mart. Apple held an integrated premium position for two decades against cheaper good-enough rivals, a position this theory expects to fall. Treat the mechanism as durable and every named prediction as dated.

The Decision

Run the two-line audit this week. Write down the metric your buyers pay on, what your median customer absorbed last year, and what your roadmap adds this year. Then name the cheaper, simpler product your best customers dismissed last quarter, and check its slope. If it improves faster than your customers do, the clock runs. Charter a separate unit with its own customers and margin rules, or accept the retreat upmarket and put a date on it. The firms in this book retreated one rational quarter at a time, and none of them wrote the decision down. Write yours down.