Search This Blog

Wednesday, August 26, 2009

Clayton Christensen Explains Disruptive Innovation at the American Enterprise Institute

Clayton Christensen, who has observed and written about how disruptive innovation changes markets and industries made three presentations at the American Enterprise Institute explaining disruptive innovation and then showing how it applied to healthcare and education. His talks give an extraordinary advantage defining what to look for when analyzing a situation for innovative opportunities. (Long post)

Before attending Clayton Christensen’s multiple presentations at the American Enterprise Institute, I watched some You Tube videos to get his context. From the videos:

Disruptive Change – If you try to innovate inside an existing system, the innovation will be co-opted by the existing system. It must compete and win from the outside to be accepted. (Chrome!)

Innovator’s Dilemma – Microsoft has overshot what its customers need. (feature creep) Scott McNealy says early technological leaders were hairballs, non-standard, heavily tweaked builds to provide or improve performance.

“I learn better by analogy than by having a mirror shoved in my face”

Clay Christensen at AEI. (6’8”, Reagan Administration alumni, genuinely pleased to be with the group). Frederick (Rick) Hess, Director of AEI Policy Studies introduces.
In industry after industry, new technology starts expensive and complicated. Then along the way there is a breakdown between value and cost. Someone can provide better quality at less cost. That is the disruption.

There are two lines of progress. Performance improvement that customers can use is usually less than the pace of technological progress. Markets open in underserving customers.

Progress can be incremental or breakthrough. Analog to digital phones was a breakthrough. On incremental change, incumbents nearly always win. The rich get richer.

A disruptive breakthrough starts as not nearly as good, but more affordable. Then the entrant becomes the new leader.

In case studies, they find the upmarket leader successively gives up the low profit, more competitive market to the new technology and the new technology improves to move steadily upmarket over time. There is no stupidity and no villains. Each group is doing what makes sense for them.

If the upmarket leader tries to introduce an offering down market, it seldom works. Each management function wants to impose their current best model of how to solve their problem. Sales will recommend its current model, design will share its current model, engineering will share its current model, distribution will want to use its current model.

To go down market, you have to allow that the product will not be solving the same problems as the market leader. It will have a different user base, (less experienced, lower expectations) a different business model (faster and cheaper to market), a different distribution system (less convoluted and expensive) than the originator. They are solving different problems, so the experience of the industry leader is of limited value to the down-market challenger.

Case studies – Steel (Bethlehem and Nucor), Airplanes (Boeing and Embraer), Airlines, Automakers, Computing Models (IBM, Microsoft, and Linux}

Similar slide presentation: Disruptive Innovation in Education & Health Care found at

Steel Industry Case Study – Electric furnace enables the mini-mill (Nucor) to make steel 20% cheaper than the integrated mill (Bethlehem).

Mini-mill starts making rebar, lowest quality steel (it is buried in concrete) 7% margin, 4% of market. They have a price advantage, eventually Bethlehem leaves the market. Prices drop 20%. Note – You can only win with a low price strategy when you have a high price competitor.

Nucor improves technology, moves up to angle iron, 12% margin, 8% of market. Integrated mills pull out, 20% drop in commodity price.

Nucor improves technology, starts making structural beams. 18% margin, 22% of market. Integrated mills pull out, price drops 20%.

Nucor improves technology, starts making sheet steel, 25 – 30% margin, 55% of the market, Bethlehem goes out of business.

Guy sitting next to me leans over and relates he was selling health benefit packages to the steel industry during that period. Many meetings with Bethlehem executives, committees, unions, couldn’t get a decision. Nucor bought quickly.

Clay’s maiden analysis was the disk drive industry. He has old disk drives as trophies all over his office, like an anthropologist has statues and totems.

He wrote a book about disruption in disk drives and got a call from Andy Grove’s office. Would he come out and brief the Chairman on innovation and disruption?

Flies across country for a half hour meeting. Gets there, Intel’s hair is on fire. New offer, ten minute meeting. Clay does ten minutes on the concept. Grove says, “I get it, so how does this affect Intel?”

Clay tells the steel story for ten minutes. Grove says, “I get it, so we have to cannibalize our existing products so we stay in business.”

Clay tells us, “How lucky I was telling the steel story. If I had told him how to fix his business, he would have never believed me. I don’t tell them what to do, I give them the tools and templates to think about what they have to do.”

Clay and Grove went on to present 18 retreats to discuss this with 2,000 Intel execs. They had breakout sessions and planning sessions, etc. Clay says the ideas generated in those shows have been responsible for $16B in new sales.

How transistors replaced the vacuum tube. Early transistors could not handle enough power to work in vacuum tube radio and television chassis. Existing radio and television manufacturers were working to develop higher voltage transistors, going nowhere. Transistors were first proven on hearing aids, where vacuum tubes were too big to use.

Boeing vs. Embraer – Embraer started out building 12 passenger aircraft. Boeing couldn’t make one at market price, that was not their market. Embraer is now selling aircraft that carry over 100 passengers, Boeing keeps innovating in both size and is now pioneering fiber material airplanes.

Airlines – Short haul routes have highest turnaround expense. Southwest dodged federal CAB jurisdiction by flying short routes inside Texas until they were ready to compete.

Retail – Hard goods like paint are less profitable, more commodities, people come into the store wanting to buy paint. Clothing and cosmetics increase margins when merchandised. Target, a disruptive spin-off of Dayton Hudson, took over hard goods and department stores got out of the market. As Target added upmarket inventory, chased the regional department stores out of business. Now Target is being chased by Wal-Mart.

Television supply chain – When Sony introduced the transistor TV, appliance stores, which sold vacuum tube TVs, would not sell transistor models, since they made substantial money replacing tubes in televisions. Kmart was just coming up and was glad to distribute Sony TVs. How fortunate that the television that needed no service was being sold by the stores that could provide no service.

In every case, there was no stupidity or villainy on either side. The existing paradigm paralyzed best practices for meeting the down-market challenge. The industry leaders were beaten by an asymmetry of motivation.

As the new, cheaper, disrupter gains market, their customers are not previous users, but new users who have access to the product benefit for the first time. Mainframes were operated by dedicated teams of professionals. PC’s were for people who would take them home and learn how to complete work on them. Linux allows the hobbyist to write their own software, create the functionality of the system.

You need a new business model to take on an entrenched competitor. Transistors couldn’t replace vacuum tubes. Although everyone knew it would happen, they couldn’t make transistors work in a 50,000 volt chassis. Transistors had to establish a beachhead in hearing aids.

When Clay and his brother pooled their funds to buy a $2.00 Sony transistor radio, it was staticky and they had to face the Great Salt Lake to hear anything. You couldn’t sell that radio to someone who owned a vacuum tube radio. You had to sell it to someone who had nothing, to teenagers, the “rebar” of the social order.

A disruptor just has to be better than nothing when the group to be served has nothing. When you compete against non-consumption, all you have to do is make it better than nothing and consumption soars.

Solar only works where you don’t have a ubiquitous, dependable, inexpensive power grid. Solar power is being adopted in the third world, but it hasn’t succeeded competing against the grid.

Slide design note – Clay’s slides started with a “high and to the right over time” graph to show each established industry’s market segments. He would then show a similar graph for the disruptor in the foreground of the first chart to show there is no interaction.

As the disrupter grew and absorbed the predecessor industry, he would refer to the previous industry as the “backplane,” showing his experience in disk drives and circuit board design.

One of Clay’s students returned to Japan and spent 2 years building the plan for the future of Japan’s economy. Called Clay in despair, said he didn’t see any hope. Clay said, “Well, come on back to Harvard. We can usually figure something like that out in a couple of days.”

After a half hour, the former student had outlined that Japan’s incredible growth in industry after industry had come from disrupting the established order…and now they were getting disrupted by the third and fourth world. Toyota loves competing with Mercedes with the Lexus. There is plenty of margin in that space…but not growth. Clay is still thinking about that.

From the audience, “What about the US, then?” Clay says we have a fluid job market, financing for new companies in place, a tradition of startups. (The previous day I saw Clay on a video saying an innovator was someone who had parents who took stuff apart and fixed things, giving the innovator permission to do the same.) Japan doesn’t have these things. There are microeconomic factors for a country’s macroeconomic optimism or malaise.

The right product architecture depends on the basis for competition. (I think that gets back to not overbuilding when you just have to be better than nothing. Establish how good you have to be and manage feature creep to your strategic advantage. D2)

Mainframes were individually wired at the start, went through “all Blue” then “plug compatible” phases. PC’s were even looser (It was a PC if it could run MS Flight Simulator). Now Linux systems take different modules from multiple distros and with some fitting they all work.

Linux is more open source than the PC which was more open source than the Mini, which was more open source than the IBM mainframe, which was more open source than the original sitewired, hairball mainframe computers. D2

Clay (and others) compute the investment to create the Windows environment is over $1B. That is possible because so many companies invested and got a return from the Windows environment. This openness means there was a lot of freedom for subsequent engineering, as they provided a basic set of standards. These standards also made them back off the frontier edge of what’s possible from their successors.

In contrast, the early MP3 was an open system which didn’t work due to different components, both in the player, in the downloading systems, and in the content. Ipod/Itunes came in and provided a complete solution which performed well and took over the market.

When something gets to be “good enough” it goes to an open architecture giving speed, responsiveness and customization. However, if the system doesn’t work, centrally controlling the user experience to make sure it works can create a market boost.

Three Enablers of Disruption
1. Simplify Technology – Foolproof and idiot simple.
2. Business model innovation
3. Value Network – Different model for retailers and suppliers

Disruption is facilitated when historically valuable and expensive expertise becomes commoditized. Three stages of this process are:
1. Experimentation and problem solving (Art)
2. Probabilistic pattern recognition solutions
3. Rule based solutions

DuPont had four molecules to build an empire, acetate, nylon, Kevlar and Teflon. They were created by experimentation and problem solving. After those molecules were created, there was a good deal of puttering and pothering to figure how to best use them. Today, there is software to predict the behavior of new molecules and also software where you can specify behavioral properties and the software will recommend a molecule, whether it already exists or not. When I wwas a Master Carpenter at DuPont in the 70’s, the Chairman, Mr. Shapiro, said DuPont could no longer afford basic research, and had moved to applied research. D2

There is a problem in medicine of diagnosing by symptom. We only have so many symptoms to present, so multiple diseases express with the same symptoms. There are currently 21 diseases that have the symptom of elevated glucose, all lumped under “diabetes.” One is fat in the liver which causes inflammation of the liver. This is cured with ibuprofen, an anti-inflammatory. Diabetes Type 1 is easy to solve with insulin.

The current state of medicine is intuitive medicine. The next step is “empirical medicine,” or pattern diagnosis, which will be replaced by “precision medicine” which is rules based.

Current diagnosis is largely by physician intuition of expressed symptoms. The next step will be understanding molecular diagnostics, and then using imagery technology.
The traditional general hospital is not a viable business model. They can’t survive as currently set up, and are surviving on philanthropy and financial rescues. Their current value proposition is “We don’t know what is wrong. We can address any problem you bring.”

There are three underlying business models that apply to hospitals and if we are going to get health costs under control, we need to understand and correctly use them.

Business Models
1. Solution shop – Fee for service. Diagnosis and experimentation to define the problem
2. Value Added Process Business – Fee for outcome. Medical procedures after a definitive diagnosis
3. Facilitated Network – Fee for membership. Caring for chronic disease.

Four Elements of a Business Model – Value Proposition, Resources, Processes, Profit Formula. These are not only a cycle, but each part becomes interlocked and interdependent with each other.

Improvement in medicine happens disease by disease, and as a case is diagnosed, it can go to treatment, then go to chronic care/ongoing therapy. (Note – If initial diagnosis is incorrect and you are passed off to a procedure, you are hosed with this model, but understanding when you are trying to form the diagnosis would help everyone on the case behave better. D2)

Clay has also observed this process in other industries. The University of Michigan now gets the majority of its students from feeder state schools, where the cost of credits are lower, since the state schools are not paying for faculty research. As this path has become better defined, the state schools are moving up and providing four year and graduate curricula, disrupting the traditional university market. However, the universities are increasing their graduate education and other services. Everyone is moving upmarket.

New disruptive solutions are usually aimed at the toughest problems, but they should be started on the simplest problems.

You can’t redo a whole system when you are making a profit from it. You should apply the disruptive solution to the simpler, unprofitable, low margin, and commodity services.

There are no villains and no stupidity. People are doing what it makes sense for them to do.

Powerful monopolies, network effects (like appliance shops selling tube televisions) and stifling regulation are most easily broken through disruption than through head-to head competition. (I see Google with Chrome, GPhone, and FCC bandwidth auctions using this. D2).

Changing Class – How Disruptive Innovation Will Change The Way The World Learns
Michael Horn

The biggest problem are conflicting mandates vs. the way we teach and learn. Computers have failed because we put them in conventional classrooms.

Howard Gardner says there are multiple intelligences, linguistic, logical/mathematical, spatial, bodily kinesthetic, musical, interpersonal, intrapersonal, naturalist.

A second filter would be learning styles, visual, written aural, playful, or deliberate.

A third filter would be pace of learning, fast, slow, or medium.

People don’t learn all things the same way.

Chester Finn, an education manager and bureaucrat did not support Gardner’s ideas, but in the end agreed education was paralyzed by existing regulation and mandates. Said this was a special situation where disruption wouldn’t work because of the social and regulatory power of the unions. He kept talking and realized that disruptive education would flourish outside of the education industry, which would just siphon off the funding.

Parents and the low cost of the disruptive technologies would be pitted against new laws limiting the amount of disruptive education, for instance the number of students allowed to enroll in alternative, disruptive education, or the certification of disruptive education. Finn was accurate and it was sad. In the end they all agreed that the education unions formed a tougher problem than Finn had been able to fix, and Clay and Michael Horn didn’t press.What will happen is clear in spite of regulation and educational industry muscle.

There are other problems attempting to block disruption in education. Temporal (educators say you have to finish the seventh grade before learning the eighth grade material), lateral (if you have a breakthrough in teaching French, you may have to go apply it to English and History) Physical (educators agree that project based learning is superior, but classrooms are not set up for it), and hierarchical (current education is the established monopoly and has the headcount to prove it.).

Audience Observation – Finn was passionate that disruption wouldn’t happen to his education industry.

Christensen and Horn didn’t argue. They had made their point previous that education was a poster child for disruption and soon. They didn’t have to respond. They had been making their case all day.

After the dustup, they got back to “wouldn’t it be nice” if there was customization instead of standardization for the good of the students.

If you are going to start the disruption, what are the lower value parts of US education? It turns out the higher values have nothing to do with education, but with maximizing political and financial leverage. Standardized testing is the current value leader and will be the last to be disrupted. The unimportant areas that are open to disruption are credit recovery (remedial education), dropouts, AP Courses, Home school and home study, small rural and urban schools, tutoring, pre-K. (and electives like “German” the rebar of the education industry.) Lack of money for educational areas the buyers think is important or that the students want to learn will propel the disruption.
Chester coins “Private Pay Parent Purchase.” (However the cost of internet distribution will approach zero. D2)

Clay explains that progress of a disruptor replacing an incumbent in a market follows an “S” curve, which comes out to a straight 45 degree line if your vertical scale is increasing log values. Clay says they can look at the very early stages of a disruption and precisely plot market gain over time.

The established giant flees to greener pastures of what is politically most important, so it is important for the disruptor to compete against non-consumption. Private Pay Parent Purchase is looking for a simpler model with better results. Start in the easiest, most underserved part of the market.

A district can evolve, a school cannot is like an culture can evolve, an organism cannot. (This reminds me of “To make a better decision, first stop implementing your worse decisions.” D2)

Clay’s models aren’t to give a solution. They give us a common language and a common way to frame our reality so we can come up with the solutions.


  1. Nice summary. Hoping to get up to speed on Christensen, I found this blog post, and ended up buying Christensen's "Seeing what's next" - touted as a more "how to" summary of his early innovation books.


  2. Clay at TEDxBoston "How will you measure your life?