Harvard Business School professor and author Clayton M. Christensen has attracted a lot of fans who are heavy hitters. Fans like Andy Grove, chairman of Intel Corp. In a review of Christensen's 1997 business best-seller, The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail, on Amazon.com, Grove wrote: "This book addresses a tough problem that most successful companies will face eventually. It's lucid, analytical, and scary." Then there's George Gilder, of Gilder Technology Report, who called it "the most profound and useful business book ever written about innovation ...." And when we interviewed Michael Bloomberg recently, the man who is now mayor of New York City credited Christensen with helping to shape his ideas about innovation in running his successful media company, Bloomberg LP.

Christensen's thesis: that the best run of the large corporations -- those that are well managed, pay attention to their customers, and invest in new technology -- are vulnerable to being outwitted by disruptive innovators.

Christensen, a professor of management at Harvard Business School, has, in his writings and his work as a consultant, been focused largely on helping big companies to understand disruptive innovation and to avoid becoming victims of it. Recently, he's looked at the process from the point of view of the disrupters themselves. He has come up with a series of tests to help entrepreneurs judge whether their ideas are likely to succeed in the marketplace. To find out more, senior editor Nancy J. Lyons interviewed Christensen at his Harvard office.

Let's get right down to brass tacks here. When I first heard you speak, you threw out this phrase: "If I wanted to start a new growth business, I'd ..." What is it that you'd tell entrepreneurs to do?

No idea for a new growth business ever comes fully shaped. When it emerges, it's half-baked, and it then goes through a process of becoming fully shaped. I've developed tests that I'm hoping can help entrepreneurs manage that shaping process, so that the business plan that comes out the other end has a very high probability of success.

So you have been able to identify the strategies that allow new growth companies to succeed?

The whole story is about motivation. The leaders in every industry have vast resources at their disposal. If I try to grab a piece of real estate that the established leaders want, where the customers are attractive and the business is attractive, the evidence is overwhelming that the leaders will win. So what I want to do is to craft a strategy that takes advantage of what I would call asymmetry of motivation. That is, a situation where I'm motivated to go after the business of the market leaders, but the piece of their business that I can most naturally go after is the one that they're the least motivated to defend.

The established companies are always looking ahead to products that better satisfy their prime customers, where their profit margins are highest. Does that mean they're not that concerned about the lower-margin end of their business? And that's where the opportunity lies for start-ups with new ideas?

Remember that when a new idea emerges in an established company, it needs to get funded. And the only ideas that get funded are those that help the established company make more money. That process favors the ideas that create improved products for existing customers, and tends to reject more innovative, or disruptive, ideas. That is what creates disruptive entrepreneurial opportunities.

Tell me more about what a disruptive idea or innovation is.

A disruptive innovation is a technologically simple innovation in the form of a product, service, or business model that takes root in a tier of the market that is unattractive to the established leaders in an industry. Very often this occurs at the low end of a market -- that is how Toyota attacked General Motors, for example. Or it takes root by providing a simple and inexpensive product that enables a new population of customers to begin participating in a new application in the market -- as was the case with personal computers.

I don't feel that this concept of disruptive technology is the solution for everybody. But I think it's very important for innovators to understand what we've learned about established companies' motivation to target obvious profitable markets -- and about their inability to find emerging ones. The evidence is just overwhelming.

How would you go about coming up with an idea for a disruptive technology?

The first thing is to find out whether there is potential to create a new growth market, a new application, within the general industry dominated by an established company. The first litmus test is, Is there a larger population of less skilled, or less wealthy, customers who could be pulled into that market?

You mentioned personal computers. Can you give us some other examples of companies that have brought new customers into a market?

If you look back into history, many of the most exciting disruptive growth businesses have done this. The available products on the market were so complex you had to be an expert in order to use them, and they were expensive. Take stocks. Prior to 1910, to buy stocks you had to be a J. P. Morgan-type person. Charles Merrill's strategy was to bring Wall Street to Main Street, and he created a business model of salespeople who were not commissioned but were salaried, so they could afford to sell stocks to ordinary Americans. That was a ridiculous business proposition at the time -- that the unwashed masses would be interested in owning stock -- but Merrill made it so simple and cheap that he brought a larger population into the investing market. Then E*Trade and Charles Schwab have done it again, enabling even college students to manage their own portfolios, and day traders -- not just George Soros -- to speculate. That's how Ricoh and Canon disrupted Xerox. They made copiers that just were so cheap you could stick one on a table in a corner, and people could do the copying for themselves. If there isn't a larger population of less skilled and less wealthy people whom you can identify to pull into the market, it's a strike against the disruptability of that market.

But not all successful disruptive strategies bring new people into the market, do they? For instance, you've cited the online educational market, which some custom publishers are going after by putting supplemental materials on the Web. Their idea is to give college students the opportunity to learn in-depth about a particular subject. Having observed the study behavior of most college students, you suggested somewhat jokingly that the custom publishers would do better to put up a site called Cram.com. How does that fit in with your theories about disruptive innovation?

If an online publisher targeted college students, it would meet that first litmus test because the present customer who chooses the textbook is the professor, not the student. But this a great example that relates to my second litmus test for disruption. An innovation will get traction only if it helps people get something that they're already doing in their lives done better. In the case of college students, that would be cramming for exams. The safest assumption is that people won't change the fundamental things they want out of life. An online service that helps students cram more easily and more effectively has a much higher probability of success than an online capability that helps students explore and learn more deeply the things that are only touched upon in textbooks and professors' lectures. That means that you need to be very observant about what people are trying to do -- not what they say they wish they were doing. If you can facilitate what they're already doing -- maybe make it cheaper, easier, and more convenient to do -- then your disruption has a much better chance of success.

Can you give us some some real-world examples to illustrate that assumption?

Think about digital photography versus film. What are people right now trying to do in their lives? We take our film to CVS and almost always order double prints. We do that because it's cheap and just in case a photo turns out great, we want a convenient copy to send to Grandma. Then we bring the prints home, we sit down at the dining-room table, and we go through them. And then what do we do? We put them back in the envelope, and put them in a box or a drawer. More than 95% of all prints get looked at only once. Very rarely is somebody conscientious enough to go back and put the most memorable in a photo album for posterity. But we do put some in an envelope and send them to Grandma.

The digital-photography people come to market with the proposition of "If you'll just take the time to learn this software, you can upload all these images onto your computer, and you can edit out the red-eye in all of those photographs you only ever looked at once. And you can keep those images organized in these online photo albums, and we'll store the albums for you in data warehouses." People who own digital cameras just don't do that stuff. There's no point in making it easier for people to do what they aren't prioritizing in the first place; they aren't going to do it. On the other hand, people do send digital images of their babies and their vacations all over the Internet. That helps them get something done that they're already trying to do.

There are a lot of companies -- not just Sony and Kodak -- that have spent a lot of money trying to make the quality of the digital images comparable with film. But when you're sending these things over the Internet, they don't have to be high quality. In fact, what do you do when somebody sends you images? You click on them, look at them once, and then you close them -- just the way we put photos back in the envelope. You don't print them and store them in your photo album. Almost always, you just close them. You can get faked out by asking people what they want. You've got to watch what they really do and how they prioritize things in their lives.

Your third test goes against everything we hear in these days of quality. You talk about crummy products' being a necessary characteristic of disruptive innovation. What's that all about?

The mistake that makes launching a venture expensive is when you try to make a disruptive technology so good that it can compete on a quality basis with an established product. If, instead, you find a new set of customers who are already trying to get something done, and you can facilitate that, they'll be delighted to have something that's actually not very good but is better than nothing.

If you go back through the history of disruptive innovations, they've all smelled like that. The first crude transistors took root in hearing aids. Then transistors got better and popped into radios that teenagers carried around to listen to rock and roll. Then we put them in portable TVs, and by the mid-1960s transistors were sophisticated enough to use in floor-standing televisions. The earliest personal computers were toys. Kids were delighted to have an Apple to use, and most businesspeople just had no use for them. People were delighted to have the PalmPilot even though it couldn't do most of the things that a notebook computer could do. And residential-housing contractors were delighted to have hydraulic backhoes, even though they could lift only a quarter cubic yard of earth, because now they didn't have to dig ditches by hand.

Do any obvious new markets come to mind -- even an example where you'd be willing to put up with a crummy product?

Sure. Think about voice-recognition technology. IBM tried at the outset to make voice-recognition technology good enough so that you could speak your word-processing documents rather than type them. That's a very complex problem. Simple voice commands are a much more likely first market. And then what about developing voice-recognition software targeting kids in chat rooms so they could speak rather than type? After that, an embedded algorithm that allowed us to speak E-mail into our BlackBerries or our Palm VIIs -- I'd be thrilled to have a voice-recognition algorithm that was 80% accurate in a mobile E-mail world, even though I wouldn't tolerate 80% accuracy when word processing on my desktop computer.

So far your tests are about expanding markets, either by bringing new people into a market -- what you call creating consumption from nonconsumption -- or by making it easier and cheaper for people to do what they're already doing or wish they could do. Are those the two primary ways to create a viable company in a market dominated by powerhouses?

If you can't do either of those things, you can still create a disruptive new business without expanding the existing market, provided two conditions are met. One condition is that the existing technology has to be more than good enough for what customers in the least demanding part of the market need. Second, you've got to be able to profitably provide the product or service at a low price point, from a business model that isn't attractive to established companies. Steel minimills did that. They didn't create new markets for steel, but they produced rebars that were good enough for the lower end of the market at a 20% lower cost. Had the minimills tried to attack the market for high-quality sheet steel, maybe technologically they could have figured that out, but it would have taken billions to do it. And when they got there, they would have just occupied the piece of real estate that the established mills wanted to own. But by attacking rebar at the bottom, the lower-cost business model took away customers that the established steelmakers cared the least about. The test is whether you can create a lower-cost business model and earn attractive profits at those low price points.

In some of your more recent work, you examine when it's more profitable to be an integrated company doing all the design and manufacturing, and when you'd do better by outsourcing from suppliers and business partners all the other elements of the system that are not your "core competence." See if I understand the reasoning here. The right model depends on how evolved your product is. When you're trying to get an innovation up to snuff, the designers are still tinkering with it to make it better, so it's almost impossible to farm out any important pieces of the system. You need to manage development within one interdependent system. You're just not at the stage where you know enough to have somebody else supply parts. You're still fiddling around with it yourself. And predictable and reliable industry standards don't yet exist.

Yes, when a product isn't good enough for what customers want, with each generation of technology the engineers have to figure out how to put the system together in an even more efficient way. Standard interfaces of the sort that enable others to supply subsystems to you simply can't exist. That means you have to do everything in order to do anything. If you think back to the era of the mainframe computer, you couldn't have existed as an independent contract manufacturer of mainframes because the way they were made depended on the way they were designed. And the way they were going to be designed depended on the way you were going to make them. Similarly, you could not have existed as an independent supplier of operating systems, core memory, or logic circuitry -- because the design of each of those things depended upon the design of the others -- and the interdependence exists when products' functionality is not yet good enough. So at that stage you have to do all of the pieces yourself.

And then when does being an integrated company turn into a disadvantage?

When a company overshoots the market -- that is, when its product or service is even better than what customers want in a given tier of the market -- the way it competes has to change. No longer will making a better product get you traction with customers who are already overserved with functionality. When that happens, the innovations that count with customers are speed to market, customized responsiveness, and convenience. That's what Dell did in the '90s. The modular architecture of its products enabled it to outsource and deliver customized products rapidly. In that situation, being integrated is an albatross. It slows you down.

You started your own consulting company, Innosight, in January 2000. I'm assuming that you're not depending on any disruptive technologies to grow your business. But are you trying to become a disruptive innovator in management consulting?

This has really been fun. There's another construct to think about that's different from the tests we've been discussing. In the most demanding tiers of every market, problems have to be dealt with in an unstructured problem-solving mode. Take the health-care market. If you've got cancer, you go to the best oncologist money can buy, and she'll run a bunch of tests and analyze the data and develop hypotheses about what type it is and what type it isn't. She then embarks on a course of therapy, and the feedback from how you respond confirms or disproves her hypotheses. There are no pat answers to these kinds of unstructured problems. In the middle tiers of a market, many of the problems can be diagnosed and remedied in a pattern-recognition mode. Like type-one diabetes. If you're always thirsty, you urinate frequently, you're losing weight, and your eyesight is blurry, you have diabetes. The pattern is so clear it doesn't take nearly the skill to diagnose and treat conditions in the pattern-recognition area as it did up in the problem-solving mode. And then at the very simplest tiers of the market, things can be diagnosed and treated in a rules-based mode -- like, if the test strip turns blue, you're pregnant. That takes even less skill.

Over time scientific progress transforms issues that formerly needed to be dealt with in the problem-solving mode and pushes them down into pattern-recognition mode and from pattern-recognition mode down into a rules-based regime. That is the disruptive engine that enables people who didn't have the skill to play in the market before to do a better job than the unstructured-problem-solving experts historically could do. That's the role of technological progress, and that's the mechanism by which our lives get better.

And you think the same sort of scientific progress in problem solving can happen in many areas, including the consulting world?

Yes. The really hairy, vexing problems about strategy, human motivation, and organizational design historically have been handled up in the unstructured problem-solving mode. When you got one of those complicated dilemmas or challenges, you hired the folks at a full-service consulting firm and paid them millions of dollars, and they helped you. They're very good at what they do. But I think that our research has taken certain problems that historically could have been wrestled with only in an unstructured problem-solving mode, and has brought them down into the pattern-recognition mode. By understanding these patterns, Innosight can help companies deal with these types of challenges. Consultants in problem-solving mode, just like oncologists, have got to analyze the data, which are available only about the past. They then try to look into the future through the lenses of the past, and it's not very clear. Looking into the future through our models, which are really well grounded theories of cause and effect, helps you see the future a lot more clearly than you could using the tools of unstructured problem solving. This is what Innosight does. Several of my former students are running it.