Summary of The Innovator's Dilemma

This book is full of insight. Here's a summary, in case you want to get a good feel of what's in the book or want to refresh your memory without re-reading the whole book.

(Here, the term "technology" refers to not just engineering and manufacturing, but also marketing, investment and managerial processes -- anything that transforms the inputs of production, like land, labour and people to products and services of higher value.)

(Some of the examples are partly my own conclusions.)

Introduction

- Why is it so hard to *sustain* success?
- Is success so unpredictable as it sounds?
- This book is not about poorly-run firms. It's about why well-run firms often lose. It addresses the first question above and partly the second.

Chapter 1

- Innovations can be sustaining (better than the previous product on the same terms) or disruptive (worse than the previous product when judged by the traditional criteria, but better in different ways).
- Sustaining vs disruptive is different from incremental vs radical. The latter comparison is about the degree of change; the former about the dimensions on which the judgment is made. For example, moving from a 1 GB to a 2 GB hard disc would be an incremental change, and moving to a 1 TB hard disc would be a radical change (1000x capacity), but both are sustaining changes, since the things that matter are the same (capacity, cost per GB, etc). Whereas moving to an SSD is a disruptive change, since it performs worse than a hard disc on the traditional criteria (capacity, cost per GB, number of write cycles) but is better in terms of new criteria (speed, size, power consumption, withstands shocks, etc). So, forget about incremental vs. radical: it doesn't matter here.
- Incumbents beat newcomers at sustaining innovations, but lose with disruptive innovations.
- Disruptive innovations are often technically simpler than the previous generation. Example: PCs vs. mainframes. So the incumbent usually has prototypes of the disruptive technology in their labs -- they just choose not to bring it to market or to be committed about it. The failure is a business one, not a technical one.
- The essence of a disruptive attack is to force the incumbent into a position where it has to destroy its strengths in order to compete.
- A disruptive firm takes its technology as a given and asks, "who can I sell this to?" whereas the incumbent takes its market as a given and asks, "what technology can I develop for this market?"
- When a firm tries to determine what's right, it does so in the context of its value network. For example, better for an enterprise hard disc maker might mean higher reliability or built-in encryption, while for a laptop hard disc maker, better may mean smaller size or lower power consumption.

- A firm may think it's listening to the market, but it's usually listening to its market -- a subset of the entire market.
- Markets that don't exist can't be measured, so it's harder to find data in support of them.
- Firms that are well-managed and do the right thing (like listening to customers, investing resources to build higher-performance products that yield higher profits, going by data, etc) often can't handle disruptive innovations, and lose. If anything, the "right" thing may be counter-productive. This is why it's an innovator's dilemma.

Chapter 2

- Technology improves in S-curves. Improvements are slow initially when a technology is new, then rapid as it's understood, then again slow as it matures. This repeats for each generation of technology.
- Disruptive innovations are worse initially in traditional terms (3.5-inch hard discs had lower capacity than 5.25-inchers), so they struggle to find a market and find a marginal one, but technology often advances faster than the market needs, so they become good enough along that dimension (capacity, in this example), and their other advantages come in to the picture (portability, price, reliability, etc), so they take over from incumbents.
- At some point, they may also beat the traditional technology on its own terms (capacity), because they are on a different S-curve, i.e., a different generation of technology, that eventually beats the older one. But this happens in only some cases. Even if the original technology retains its advantages over the disruptor for ever, the market doesn't care, since both technologies are over-serving market needs along that dimension. The other dimensions where the disruptive technology is better become competitive advantages. (It doesn't matter whether 3.5 or 5.25 inch discs have more space, if both have more than you need. The small size of 3.5-inch hard discs was more important, since it let manufacturers build smaller computers.)
- The position that a firm occupies in the market dictates the firm's organizational structure, cost structure and margins. A mainframe component supplier may have a high cost structure, with an expensive sales force, because that's needed in the mainframe market. When a mainframe component supplier evaluates the PC business, the low margins will cause their profits to go down, below their cost structure, leading to loss, so the rational thing for them to do (from their point of view) is to not enter the PC market. Besides, their org structure and values will come in the way.
- A downmarket company looks upmarket and finds higher margins, so is happy to go upmarket. The upstream guys, however, as described above, do not find the low-end market appealing, so it's a one-sided fight, with a disastrous outcome for the incumbents.
- This is worth repeating: companies find it tempting to move upmarket but hard to move downmarket.
- By the time the disruptive product reaches the mainstream market, it's too late for the incumbent -- the upstart has the experience and volumes, etc. Besides, the incumbents have little time to react -- they enjoy a lot of success and profits till the day the disruptive technology becomes good enough, at which point the market switches over, and they die quickly.

Chapter 3

- Setting up a separate division in the incumbent company generally doesn't work. The division isn't given enough resources, or the cost structure or the management structure or the parent organization's values come in the way.
- One effective strategy for an incumbent is to take a stake in a disruptive company so that it can eat its cake and have it too.

Chapter 4

- Sound resource allocation causes companies to move upstream (margins are higher) but not downstream.
- There are three theories on who makes decisions and allocates capital at big companies. One says that it's senior management. The other says that middle managers first filter out ideas that lead to low margins, or are incompatible with the firm's strategy, or don't have a ready market, etc, and pass on only the "good" ideas to senior management. Senior management merely thinks they are making important decisions, but the actual decisions have already been made by middle managers. The third theory ("resource dependence") says that the customers really drive the direction of the company, resource allocation, etc -- they have more of a say in the company's decisions than the people in the company.
- Even when a senior manager decides to purse a disruptive strategy, the people in the organization are likely to ignore it or, at best, cooperate reluctantly if it doesn't fit *their* model of what it takes to succeed as an organization or as individuals within an organization. Well-run companies are not populated by yes-people.
- Another reason it's harder for a company to move downstream is that their customers can themselves be moving upmarket. The company may not notice it's moving upmarket.
- There are five reasons why it's hard to respond to a disruptive attack, but if you understand them, you can use them to your advantage.
1. Customers effectively control resource allocation in well-run companies.
2. Small markets don't provide sufficient growth for big companies.
3. Applications for disruptive technology is unknowable in advance, and you need to fail your way to success.
4. Organizations have capabilities, processes and values independent of the people within them. These strengths within their current business model are disabilities when dealing with a new market.
5. Technology supply may not equal market demand.

Chapters 5 and 6

- It's critical for a company to develop disruptive innovation early, but not so important whether it's early or late with sustaining technology.
- Companies are able to make huge, multi-year, bet-the-company investments on sustaining technology, but can't make small, technologically-straightforward investments on disruptive technology.
- Developing a new market is considered risky, but it's actually less so, and more promising, than to enter an existing market dominated by entrenched companies.
- The size of the organization should match the size of the market. A big company can't attack a small market. Instead, create a spinoff or acquire a small company (and run it as a separate business).
- Small markets cannot generate the revenue that a large company needs, to grow at a good rate. Being big is a very real disability for an organization.
- Building new markets requires iterations -- repeated failures till you succeed.
- A big company can't usually speed up the growth of the market to make it attractive to it.
- Asking users what they want may not help, either -- they don't know.

Chapter 7

- Emerging markets are discovered together by suppliers and customers -- they are unknowable ahead of time by either party.
- Strategies and plans for dealing with disruptive innovation should therefore be plans for learning and discovery rather than plans for execution.
- People are trained (both at school and on the job) in managing existing markets, but not in discovering new ones.
- Applying management and investment strategies for sustaining innovation to disruptive innovation will be disastrous. For example, these processes demand quantified data on the size of the market, revenues, etc.
- Act as if your forecasts for the market are wrong, rather than that they are right.
- More important than coming up the right strategy right out of the gate is to make sure you don't run out of resources or credibility until you iterate to find the strategy that works.
- A big company penalizes managers for failing, making it hard to iterate.

Chapter 8

- Managers think about whether people in the team have the capability for the job, but they should also think about whether the organization itself has the capability for the job.
- An organization's capabilities derive from resources (people, cash, equipment, brands, etc), processes (how to do a particular task; processes can be formal or informal) and values (what's important?). Looking at these three separately helps us better understand what an organization can or cannot do.
- Clear, consistent and broadly understood values are important as a company becomes bigger, since they allow employees to act autonomously and consistently, but they also define what an organization cannot do.
- Companies have processes to bring sustaining technologies to market, but not disruptive technologies. This is partly because disruptive technologies show up rarely.
- In a startup, people (resources) account for much of what gets done. Over time, processes and then values become more important. As this happens, it becomes harder for the company to change to new problems. Processes are less flexible than people, and values even less so.
- This is because the very raison d'etre for processes is to do the same thing consistently, over and over again -- processes are meant to *not* change. The other reason is that processes result in the creation of organizational boundaries, which in turn resist a change in the process.
- Some big companies, like McKinsey, have taken this to an extreme -- hundreds of MBAs join, and leave, every year, but they churn out solid work year after year, because the organization's strength is in its processes and values, not people.
- Small companies are actually more capable of disruptive innovation than big ones, despite having far less resources, because their processes and values aid rather than come in the way of disruptive innovation.
- Processes can impede disruptive innovation because of their task-specific nature -- you can't ask one process to do two fundamentally different things.
- One way of looking at a potential acquisition is to ask: am I paying for resources, processes or values? To what extent is each responsible for the company's success?
- If an acquired company's processes or values were the real driver of its success, the last thing you want to do is integrate it into the parent organization, because you're destroying the very thing that let it succeed.
- Financial analysts often have a better intuition for the value of resources than for processes.
- For example, Cisco has a very good sense of what it's doing. When it acquires a small company, it throws away the startup's processes and values, because that's not what Cisco paid for, and integrates the people into Cisco. But when it acquired StrataCom, a bigger, more mature company, it did not integrate it into Cisco. It let StrataCom stand alone, and infused its substantial resources into StrataCom, enabling it to grow at a faster rate.
- Similarly, Johnson & Johnson acquired small companies building disruptive technologies like disposable contact lenses, endoscopic surgery and diabetes blood glucose meters, let them stand alone, infused them with resources, and grew each to a billion-dollar business.
- Setting up a division within a big company to work on disruptive technology is hard, because while the division may have a lot of resources, the parent's processes and values hold it back, for the same reason a small company is more capable of building disruptive technology.
- Toyota upended the world automobile industry by better process (for development, manufacturing, supply chains, etc), and GM responded by buying fancy equipment for $60 billion (resources), which helped little when plugged into an antiquated process. You can't fight process with resources. Again, most of a big company's value is in process and values, not resources.
- Spinning out a company requires it to not have to compete with the parent company for resources; otherwise the parent company's values will starve the spinout. Only the CEO can ensure that this doesn't happen. Whether the spinout is physically separate is less important.
- The less the fit between the parent's and the child's processes, the more heavyweight a team the child needs to be. When there's no mismatch, you don't even need a separate team; a product manager can co-ordinate across existing functional groups (like software engineers and user experience designers) in the parent organization. When there's a huge mismatch, you need a separate team that uses its own processes to get work done.
- The less the fit between the parent's and the child's values, the more autonomous the child's organization needs to be. If there's a strong, sustaining fit, you don't need a spinout or a skunkworks -- the energy and resources of the parent organization will coalesce around it. If not, you need a spinout; otherwise the team will be starved of resources.

Chapter 9

- When technology grows faster than the market demands, the disruptive technology becomes good enough, and customers choose products based on other attributes. This leads to a new phase in the product life cycle. People initially bought computers based on their performance, and laptops were seen as underpowered (and expensive) machines. Once laptops became powerful enough, people decided that the performance advantage of a desktop isn't as important, and they'd rather have portability, and switched to laptops.
- Once a switch happens, it increases demand for the new generation of technology, and the economies of scale cause a reduction in price, so the price premium for the newer generation reduces, sometimes to zero.
- Disruption can happen several times in an industry. People first bought hard discs based on capacity till both 3.5 and 5.25 inch discs had enough capacity, then bought discs based on the size, till disks became small enough, then bought discs based on reliability, until most discs approached a million hours of mean time between failure, and are now buying disks based on price.
- Commoditization (competing solely on price) is the last phase of evolution of a product, when all aspects of the product have evolved to be good enough.
- The "buying hierarchy" model is a generalization of this, and describes the following four phases: users buy based on functionality, then reliability, then convenience, and finally price.
- We can look at the same progression from the point of view of the user rather than the product: early adopters buy based on functionality. As the required functionality is built, the market grows and late adopters enter the market and buy based on reliability, leading to another round of growth in the market, and when all products are reliable enough, the late majority come in and buy based on price. (This description leaves out the stragglers, who don't want change of any kind, and buy touch-tone phones only because you can no longer buy rotary dial phones.)
- The attributes that make disruptive products worthless in mainstream markets are typically their strongest selling points in emerging markets.
- So you're better off looking for a new market for your technology rather than trying to improve the technology to please the existing market. That is, look at it as a marketing challenge rather than a technical one.
- Disruptive products are usually simpler, cheaper, more reliable and more convenient than established products.
- Don't spend a ton of resources building a better-performing product than the market demands. A disruptor will eat your lunch.
- Since technology oversupply is dangerous to incumbents, they can try to prevent this situation by making sure that demand continues to rise. Microsoft and the computer hardware companies did this, probably unintentionally, with ever-increasing hardware demands for newer versions of Windows, causing customers to continue to demand beefier computers. But this can't go on forever, since people are finding the previous version of Windows to be good enough and are not upgrading. The resulting technology overshoot led to the popularity of laptops and now the iPad.
- Any of these three strategies to deal with technology oversupply (stick with the present technology and go upmarket - dangerous, move with customers's needs and adopt the disruptive technology, or prop up demand to postpone oversupply) requires predicting the two curves (technology supply and market demand) to be successful.

Chapter 10 (case study on electric cars)

- The best way (that the author knows) of identifying disruption is to draw a graph between performance improvement demanded in the market and performance improvement delivered by technology. If the two lines look like they'll meet in the future, that's the time where the industry is ripe for disruption. Of course, that requires you to identify the y-axis -- what do customers care about? There may be several of them that need to be addressed.
- The attributes that matter in a car are the ability to drive 125-150 miles without refueling, to accelerate to 60mph in 10 seconds (to safely merge with freeway traffic) and a wide array of options.
- Electric cars can be disruptive only if they manage to improve along these dimensions faster than market demands increase. There's a 2-4% p.a. improvement in technology along the above attributes, and only a 1% improvement in market demands, so they could be disruptive in the future.
- Electric cars are disruptive, so by definition the incumbents don't see them as a threat.
- Saying that electric cars will never perform as well as gasoline cars is missing the point. The question is: will they be good enough for the market?
- Experts are bad at predicting the future of disruptive technology and the market sizes, so we should be skeptical of them (just as we should be skeptical of our own analysis).
- If you're an electric car manufacturer, how would you create a (unsubsidized) market?
- First of all, the market we're looking for is not the current market, by definition, since it's a disruptive technology.
- Most car manufacturers will focus precisely and myopically on the current market, because of resource dependence (customers decide where a company invests) and small markets not providing sufficient growth for big companies.
- So, look for a market that's okay with a slow car that doesn't travel very far.
- Second, no one can learn ahead of time what the early markets for electric cars will be, whether by market research, asking customers, etc. The only option is to iterate in the market by selling real cars to real people who pay real money.
- Government incentives or mandates are likely to distort the search for a market, rather than aid it, so don't take those into account.
- Third, the business plan must be a plan for learning, not for executing a preconceived strategy. Don't spend all your resources on the first iteration, like Apple did with the Newton.
- Speculation on potential markets: cars for teenagers, whose parents don't want them to go fast or far, or taxis and other vehicles for use within congested Asian cities, since electric vehicles can be more efficient in stop-and-go traffic.
- Chrysler is trying to improve the technology for the market (building a car with almost 1 ton of batteries, which costs $100K) rather than trying to find a market for the cars as they exist. Ford's doing the same thing. Wrong strategy.
- Performance oversupply seems to have occurred in cars -- people don't buy cars based on which one accelerates to 60mph the fastest, and there's overchoice in options.
- Which means that the basis for competition will change. The first electric car will be simple and convenient.
- Since we need to iterate several times, design so that features, function and styling can be changed quickly and cheaply. Get the initial design out quickly and cheaply.
- Design for a low price. Disruptive technologies typically have a lower sticker price, though the running costs may be higher. Smaller disks were cheaper, but more expensive per MB. Hydraulic excavators were cheaper, though they cost more to move a given quantity of earth in a given time. Similarly, an electric car must be cheap, even if it costs more per km to run.
- Don't hope for technological breakthroughs. A technological breakthrough is needed only if you choose to view electric cars as a sustaining technology, rather than choosing to harness the laws of disruptive innovation.
- Maybe electric cars will eventually have a range of 150 miles, but those will be built by companies that create a new market and then develop the sustaining technologies need to take them upward into more attractive markets. Remember that companies are upwardly mobile and downwardly immobile.
- Remember that we'll need to create a new distribution channel, as has happened historically with disruptive innovation, because existing dealers won't accept it. The electric car won't fit the dealers' model for making money any more than it fits the incumbent car manufacturers' models for making money.
- Create an autonomous business unit or a separate company whose stock is owned by the parent. This makes resource dependence work for us rather than against us, since our customers will be electric car users rather than gasoline car users. A separate BU also solves the problem of a small market not providing sufficient growth for big companies. Rather than fight with company management as to whether the business is viable, let customers decide.