Innovators Dilemma Christensen

1. 1Map your allocation filter first. List the formal processes (stage gates, margin hurdles, market-size thresholds, forecast requirements) and informal ones (who gets heard, which customers' requests reach engineering) that decide what gets funded. The source says these processes, not managers' intentions, control the resource flow.
2. 2Score every proposal on the three tests the passages name: margin (does it beat current gross margins?), market size (is the forecast large enough to matter to your growth need?), and customer pull (do your most powerful current customers want it, and can they use it?). Record the scores in writing so the filter becomes visible.
3. 3Classify each proposal as sustaining or disruptive. Sustaining means current customers need it and market size, growth rate and trajectory are known. Disruptive means simpler, cheaper, lower margin, aimed at an emerging or insignificant market, and initially wanted by your least profitable customers. Expect a disruptive proposal to fail all three tests.
4. 4Treat a failed test as a diagnosis, not a verdict. When a proposal fails because it is low-margin, small, and unwanted by top customers, the filter is working as designed. Ask whether the technology could be performance-competitive in your market later, since customer needs often rise more slowly than technology improves.
5. 5Do the growth arithmetic. Compute the new revenue you need next year (the source's example: a $4 billion company growing 20% needs $800 million). Compare it with the forecast size of the emerging market. If the market cannot cover it, the main organization will never rank the project highly.
6. 6Do not wait for the market to become 'large enough to be interesting' and do not rely on forecasts of a market that does not exist yet. Plan to learn by entering early with small commitments.
7. 7Create an autonomous organization for any disruptive proposal worth pursuing. Size it to the target market, give it a cost structure that profits at low margins, and point it at a different set of customers who want the simpler product, free of the mainstream customers' power over its resources.
8. 8Keep the mainstream allocation process intact for sustaining work. Do not weaken it. The source says it is what keeps established firms leading every sustaining innovation, so the fix is a separate venue, not a looser filter.

Seagate, 1985–86. Its executives weighed shelving the 3.5-inch drive against moving upmarket. Looking downmarket, the 1987 forecast for 3.5-inch drives was small, margins were uncertain, and manufacturing predicted much higher cost per megabyte than 5.25-inch drives. Looking upmarket, 5.25-inch drives of 60 to 100 MB were forecast at $500 million by 1987, with gross margins of 35–40 percent, against 25–30 percent on Seagate's high-volume 20 MB drives. A competing proposal, the ST251 line, was also under evaluation. Seagate shelved the 3.5-inch project, and the passage calls the decision 'starkly rational.' The firm then accelerated sustaining work: new 5.25-inch models per year, as a percentage of the prior year's models on the market, rose to 57, 78 and 115 percent in 1985, 1986 and 1987, with thin-film disks, voice-coil actuators, RLL codes and embedded SCSI. Every number favored the choice, yet the logic is the same that lets disruptive technologies slip past leaders. Applying the method shows the proposal failing all three tests (margin, market size, customer pull), and shows that the fix was a separate organization for the small market.

• If a proposal is lower-margin, aimed at a small or emerging market, and unwanted by your best customers, then expect the main allocation process to reject it. Do not argue it through that process; house it in an autonomous unit.
• If your growth requirement dwarfs the forecast market (e.g., $800 million needed versus a small emerging segment), then assign the project to an organization whose size matches the market.
• If the project is sustaining (known customers, known needs, known market size), then run it through standard analysis and planning, because that is where those tools work.
• If the project targets a market that does not yet exist, then do not demand a reliable forecast or a margin case from it. Treat plans as a first guess to be revised by early market experience.
• If a customer-wanted sustaining project and a disruptive project compete for the same engineers and budget, then the sustaining project will win. Protect the disruptive one by separating the resources.

• Concluding the filter is broken and loosening it for everyone. The tell: sustaining projects start getting approved with weaker evidence while disruptive ones still stall.
• Waiting until the new market is 'large enough to be interesting.' The tell: your first serious review of the technology comes after your customers begin asking for it, by which point it is too late.
• Asking the mainstream organization to carry a low-margin project inside its high-end cost structure. The tell: the project can't reach profitability at the margins its market supports, and its people get pulled onto bigger-customer work.
• Judging a technology only by today's performance against current demand. The tell: statements that it 'underperforms' with no check of whether customer needs are rising more slowly than the technology improves.
• Treating a rational-looking decision as proof of a sound strategy. The tell: a clean spreadsheet, as in Seagate's case, that omits the downmarket technology's future trajectory.

• ▸Proposal name: one line describing the product or technology.
• ▸Type (sustaining / disruptive): sustaining if current customers need it and market, trajectory and needs are known; disruptive if simpler, cheaper, and attractive first to your least profitable customers.
• ▸Margin test: expected gross margin versus your current margin on comparable products (e.g., 25–30% vs. 35–40%).
• ▸Market-size test: forecast revenue of the target market and when; note whether it is emerging or insignificant today.
• ▸Customer-pull test: do your most powerful current customers want it, and can they use it now? Name them.
• ▸Growth gap: new revenue you need next year versus the forecast market size.
• ▸Future-performance check: will customer needs in your market rise more slowly than the technology improves? State the evidence.
• ▸Verdict from the main allocation process: fund, shelve, or reject, with the stated reason.
• ▸Venue decision: keep in mainstream, or create an autonomous organization (name its size, cost structure, and target customers).

1. 1Classify the innovation first. Ask whether it is a sustaining technology aimed at a known market with understood customer needs, or an enabling/disruptive one whose applications are unknown. Sustaining means a planned, researched, execution approach fits. Disruptive means a discovery approach.
2. 2If it is disruptive, state the premise in writing: the market applications are unknowable, not just unknown. Add the assumption that your forecasts are wrong and your chosen strategy may be wrong too. This is the starting position of discovery-based planning.
3. 3Drop the quantification gate. Do not require market-size and financial-return numbers before entry. When neither revenues nor costs can be known, demanding them produces paralysis or serious mistakes.
4. 4Rewrite the plan as a plan for learning. List what must be known (who uses it, why, which features matter, what the market is worth) and design the cheapest actions that would reveal each item.
5. 5Sell the product to whoever can use it, then watch where it goes. Intel had no explicit strategy for the microprocessor market and sold the chip to whoever seemed able to use it. Treat early sales as discovery data, not as a test of your forecast.
6. 6Budget for several iterations. Customers and manufacturers discover the market together, and that takes time. Apple's path to the desktop standard ran through the Apple I, II, II+, III, Lisa, and three Macintosh generations. Fund the learning cycle, not one big bet.
7. 7Do not ask customers what they want and then spend heavily to deliver it. In an emerging market they do not know. Do not push the technology beyond the state of the art or spend aggressively to convince people to buy a fixed design.
8. 8Match the organization to the market's size. Emerging markets are small by definition, so the unit running the effort must be able to turn a profit at small scale. Judge it against small-market yardsticks, not against the parent company's growth needs.

Honda entered the North American motorcycle market in 1959 with an execution-style target: 10 percent of a market estimated at 550,000 units a year, growing 5 percent annually. Its forecast was wrong about the market's size and also about its nature. By 1975 the market had grown 16 percent per year to 5,000,000 units a year, and the units came largely from an application Honda could not have foreseen. A discovery-based reading of this case: the planned numbers mattered less than Honda's ability to learn where demand actually was. Intel shows the same pattern. It had no explicit microprocessor strategy, sold the chip to whoever could use it, and went on to dominate the market. Apple's Newton is the counter-case. Apple assumed customers knew what they wanted, spent heavily to find out, and then spent heavily to sell the result. 140,000 units sold, and in the giant Apple of 1994 that read as failure, though it was three times the Apple II's early sales.

• If the technology improves performance in a market where customer needs are understood, then run a planned, researched execution approach with forecasts and financial gates.
• If the technology enables something previously deemed impossible and neither you nor customers know how it will be used, then plan for learning and treat every forecast as probably wrong.
• If your investment process demands market-size and return numbers before entry and none can exist, then change the process for this project instead of inventing the numbers.
• If an early product underperforms its plan, then ask what it revealed about who values what before judging it a failure. Apple's early machines failed several times before the third Macintosh found the standard.
• If the sponsoring organization needs large revenue to feel a win, then move the effort into a unit sized to the emerging market. Selling 140,000 units was a failure at 1994 Apple and would have been a triumph at 1979 Apple.

• Applying sustaining-innovation marketing habits to a disruptive product. The tell: market research is commissioned and its output is treated as the decision basis even though no data on the market exists.
• Treating forecasts as commitments. The tell: the plan fixes a market share and growth rate, as Honda's 10 percent of 550,000 units did, and the team defends those figures instead of testing them against what sells.
• Trying to shorten mutual discovery with money. The tell: heavy spending to learn what customers want, to push technology past the state of the art, and to persuade buyers, as with the Newton.
• Judging results by the parent company's scale. The tell: a product that outsells an earlier success three to one is labeled a failure because the absolute numbers look small.
• Assuming capable people guarantee a capable organization. The tell: the right people are assigned, but the processes and values of the host organization still demand quantified returns and big markets.

• ▸Innovation and target use: name it and describe the intended customer in one line.
• ▸Sustaining or disruptive: state which, and why. Does it improve performance in a known market, or enable something previously thought impossible?
• ▸Evidence of market knowledge: list the customer needs and market data that actually exist. Write 'none' if none do.
• ▸Posture chosen: execution or discovery, with a one-line justification from the two fields above.
• ▸Assumptions I treat as probably wrong: list the forecast figures and strategic choices you are holding loosely.
• ▸Learning questions: list what must be learned (users, uses, valued features, market size).
• ▸Cheapest tests: for each learning question, the lowest-cost action that would answer it, such as selling to whoever can use the product.
• ▸Iteration budget: how many product and market attempts you can fund before judging success or failure.
• ▸Scale test: what revenue makes this unit profitable, and does the host organization accept that size as a win?

1. 1Classify the innovation first. Ask whether it improves performance along the dimensions mainstream customers in major markets have historically valued. If yes, it is sustaining, even if it is radical or technically hard, and the sustaining playbook applies. If it underperforms established products on those dimensions in the near term but brings a different value proposition, treat it as disruptive. The passages say this distinction, not incremental versus radical, determines which rules apply.
2. 2Write down the near-term performance gap. List where the product is worse than the incumbent's on mainstream measures, and what it offers instead (a different value proposition). Disruptive technologies 'generally underperform established products in mainstream markets,' so a gap is expected. The test is whether some customers value the different attributes.
3. 3Stop using your mainstream customers as the validation panel. The source says there are times when it is right not to listen to customers, right to build lower-performance, lower-margin products, and right to pursue small markets. Record which of your usual approval filters (customer demand, margin targets, market-size thresholds) the project would fail, and decide explicitly to override them for this project.
4. 4Give responsibility to the organization whose customers need the technology (chapter 5). Assign the project to the unit, or a new one, whose customers value the new attributes. Do not assign it to the unit whose customers would reject it. Your resource-allocation process favors investments that promise the best returns from existing customers, so a project sitting there is steadily starved.
5. 5Match organization size to market size (chapter 6). Staff and fund the effort so the new market's early revenue can still matter to it. A large organization measuring itself against large markets will reject a small market that is the right one.
6. 6Discover the new market by probing rather than forecasting (chapter 7). Treat the market as emerging and unknown. Identify who values the different attributes and get the product in front of them. Revise the plan from what you learn instead of defending a forecast.
7. 7Appraise capabilities and disabilities honestly (chapter 8). Check whether your processes and values, which are tuned to your current business, are disabilities for this one. If they are, run the project outside them.
8. 8Track performance supplied versus performance demanded (chapter 9). The pace of technological progress can outstrip what markets need. Watch where the technology's trajectory crosses the performance that different market tiers require, because that is when a once-inferior approach becomes competitive. Measure success by share in the new market, not by profit in the old one.

Xerox long dominated plain-paper photocopiers for large, high-volume copying centers. Yet it missed the growth and profit opportunities in small tabletop photocopiers and became only a minor player there. Run the steps backward and the failure is clear. Tabletop copiers underperformed on the speed and volume that Xerox's mainstream customers valued, so they screened out as low-margin and small, and the customers Xerox listened to did not want them. Xerox was well managed, and that is the source's point: the decisions that led to failure were made when the leader was regarded as among the best in the world. The same pattern appears elsewhere in the passages. Steel minimills now hold 40 percent of the North American steel market, including nearly all of the bars, rods and structural steel markets, and no integrated steel company had built a minimill-technology plant by 1995. Of thirty cable-actuated power shovel makers, only four survived the twenty-five-year move to hydraulic excavation. In disk drives, the dominant firm kept its lead through only two of six new architectural technologies. Success means having been the one that took the new market, and these cases show how rarely incumbents do.

• If the innovation improves what mainstream customers already value, then treat it as sustaining and run normal customer-led, return-driven investment. If it is worse on those dimensions but offers a different value proposition, then treat it as disruptive and apply the disruptive rules.
• If the project fails your standard filters (customer demand, margin, market size) because of how disruptive technologies normally look, then do not read that failure as a verdict. Check whether the filters are measuring the wrong market.
• If your current customers would reject the product, then place it with an organization whose customers need it. Do not run it through the unit serving the mainstream.
• If the target market is small, then size the organization so a small market still matters to it. Do not wait for a market that fits a large organization's growth needs.
• If product performance already exceeds what a market tier needs, then expect a different technology to become competitive in that tier. Keep monitoring supply against demand.

• Misclassifying a disruptive technology as a hard sustaining one. The tell: you judge it by technical difficulty or radicalness, and the roadmap measures it against mainstream performance specs.
• Asking your best customers whether to proceed. The tell: the project dies on 'customers don't want it' even though the source says listening to them is exactly what leads leaders astray here.
• Applying the return-on-investment screen to a small emerging market. The tell: the business case is rejected for low margins and small size before any market learning has happened.
• Housing the project inside the mainstream organization. The tell: it competes for resources against sustaining projects with larger, more certain returns and keeps losing.
• Assuming good management is a defense. The tell: the justification is 'we are a well-run company with strong innovation capabilities,' the same standing Sears, Xerox and Digital had when they failed.

• ▸Innovation and technology: name it in one line, using the source's broad definition of technology (processes, not just engineering).
• ▸Sustaining or disruptive: does it improve performance on mainstream customers' valued dimensions? Answer yes or no and give the evidence.
• ▸Near-term performance gap: list where it underperforms the established product.
• ▸Different value proposition: state the attributes that make it valuable to some customers.
• ▸Standard filters it fails: list which of customer demand, margin or market size it fails and why.
• ▸Customers who need it: name the organization or unit whose customers value the new attributes.
• ▸Organization-to-market size match: state the market's early revenue and the organization's size, and check they are proportionate.
• ▸Market discovery plan: list the assumptions to test and the first customers to test them with.
• ▸Capability and disability check: name the processes and values that would obstruct the project and where it will be insulated from them.
• ▸New-market share target and tracking: define the market, your share goal, and the performance supplied versus demanded signal you will watch.

1. 1Classify the innovation first. Ask one question: does it improve the performance of an established product along the dimensions mainstream customers in major markets have historically valued? If yes, it is sustaining, however radical or technically difficult. Do not use the incremental-versus-radical test; the source says that is the wrong distinction.
2. 2Run the disruption screen before committing resources. Check three tells: the product is simpler and cheaper with lower margins, it is first commercialized in an emerging or insignificant market, and your most profitable customers neither want it nor can use it. If none of these hold, treat it as sustaining and use the standard playbook. If any hold, route it to a separate disruptive process.
3. 3Anchor the plan in what leading customers have already articulated. In sustaining contexts the market size and growth rates are generally known, the technology trajectory is established, and leading customers' needs are well stated. Write the target spec from those needs, not from your engineers' preferences.
4. 4Do the analysis the source says works here: sound market research, a full financial case, and execution to plan. Rank the investment against alternatives by expected return, and fund it through your normal resource allocation process. That process is built to favor exactly this kind of project.
5. 5Commit aggressively, even to radically difficult technology. The evidence says established firms led in every single instance of sustaining innovation in the disk drive industry. The risk to manage is execution, not whether the effort is organizationally legitimate.
6. 6Keep the project inside the mainstream organization and route it through your best customers. Resource dependence works in your favor: customers and investors want this investment, so money and talent follow it.
7. 7Measure success as market share held or improved in the existing mainstream market. Set the baseline share before launch and review against it after commercialization.
8. 8Do not over-invest in being first. The source notes that in many sustaining innovations technology followers do about as well as leaders, so size the urgency to the actual competitive stakes.

The source's disk drive history is the model case. Established firms stayed atop wave after wave of sustaining technologies, meaning technologies their customers needed, and led in every single instance of sustaining innovation in the industry. The reason is that market sizes and growth rates were known, technology trajectories were established, and leading customers' needs were well articulated. Research and planning therefore worked as designed. Even radically difficult sustaining technologies rarely caused leading firms to fail. Those same firms later stumbled over simpler disruptive technologies, because their best customers did not want them and they promised lower margins. Sears is the contrast: well run and highly regarded, it held its position through its core approach but missed discount retailing and home centers. The lesson is that sustaining success proves the management system works, and says nothing about readiness for disruption.

• If the change improves performance on the dimensions mainstream customers already value, then run it through the normal planning, research and funding process, even when it is technically radical.
• If your best customers are asking for it and the margin is higher, then fund it inside the mainstream organization and set a market share target.
• If the change is simpler, cheaper, lower margin, aimed at a small or emerging market, and unusable by your top customers, then do not judge it by sustaining criteria; send it to a separate, appropriately sized organization.
• If leadership in this particular sustaining innovation is not competitively important, then do not pay a premium to be first; a fast follower does about as well.
• If sustaining wins are piling up, then do not read them as proof against disruption risk; the same customer-driven system that wins here also kills the ideas your customers do not want.

• Treating technical difficulty as a sign of disruption. The tell: the team calls a hard, expensive improvement for current customers 'disruptive' and diverts it from the mainstream process, when radically difficult sustaining technologies rarely topple leaders.
• Misclassifying a disruptive technology as sustaining. The tell: the business case is built on mainstream customer demand, yet the product is simpler and cheaper, margins are lower, and top customers cannot use it.
• Applying sustaining discipline to emerging markets. The tell: you demand precise forecasts and a return case for a market that does not yet exist, when the only thing certain about such forecasts is that they are wrong.
• Overpaying for leadership. The tell: you spend to be first in a sustaining innovation where followers do about as well.
• Mistaking sustaining success for good health. The tell: share is stable and customers are happy, yet nobody is funding anything the best customers do not ask for.

• ▸Innovation description: state in one line what changes and for which product.
• ▸Performance dimension improved: name the dimension and confirm mainstream customers have historically valued it.
• ▸Disruption tell 1: simpler and cheaper with lower margins? (yes/no, evidence)
• ▸Disruption tell 2: first sold in an emerging or insignificant market? (yes/no, evidence)
• ▸Disruption tell 3: do top-margin customers reject or fail to use it? (yes/no, evidence)
• ▸Classification: sustaining if all three tells are 'no'; otherwise route to a separate organization.
• ▸Customer-articulated need: cite the named leading customers' stated requirement.
• ▸Baseline mainstream market share: record share before launch.
• ▸Target and post-launch share: set the maintain-or-improve target, then record the actual result.
• ▸Leadership value check: is being first competitively important here, or can a fast follower do as well?

1. 1Audit your own history first: list the architectural or technological transitions in your industry and mark whether the incumbent leader held its lead after each. In disk drives, the dominant firm held its lead in only two of six new architectural transitions, so assume your default odds are poor.
2. 2Define 'technology' broadly. Per the source, it covers the processes that turn labor, capital, materials, and information into products of greater value, including marketing, investment, and managerial processes. Scan for changes in any of them, such as Sears' retail model versus a discount warehouse model, not only engineering changes.
3. 3Run each emerging technology through the three-part rationality test the source gives. Check whether (1) the product is simpler and cheaper with lower margins, (2) it is first commercialized in an emerging or insignificant market, and (3) your most profitable customers don't want it and initially can't use it. Three yeses mark a disruptive candidate that your normal investment process will reject.
4. 4Identify who adopts it first. A disruptive technology is initially embraced by the least profitable customers in the market. If those are the only customers asking for it, do not treat their lack of interest from your best accounts as evidence against it.
5. 5Separate the decision from the standard process. Your usual disciplines (listening to best customers, funding the highest-return innovations, studying market trends) are what produce the 'no.' Set a rule that disruptive candidates are not judged by those disciplines alone.
6. 6Assign responsibility for the disruptive technology to an organization whose customers need it, and match the size of that organization to the size of the market (Part Two, chapters 5 and 6). A small market cannot move a large unit's growth numbers.
7. 7Treat the market for the disruptive product as one to discover, not forecast (chapter 7). Fund early probing of emerging markets instead of building a case from your current customers' stated needs.
8. 8Keep running the core business well while you do this. The source frames the task as doing what is right for near-term health and simultaneously focusing adequate resources on the technologies that could ultimately cause your downfall. Do not trade one for the other.

Digital Equipment Corporation shows the pattern. In 1986 it was described as 'a moving train' at $7.6 billion in size, and it featured prominently in the McKinsey study behind In Search of Excellence. DEC had created the minicomputer market, which IBM, the mainframe leader, had missed by years. Yet DEC then missed the desktop personal computer, and a few years later writers said it had 'squandered two years trying halfway measures' against low-margin PCs and workstations. Apply the three-part test to the PC in DEC's position: the PC was simpler and lower-margin than the minicomputer, it started in a small new market, and DEC's best minicomputer customers did not want it. Every standard DEC discipline returned 'don't invest.' Leadership across waves would have required the Part Two approach: a separate organization sized to the small PC market, with the core minicomputer business kept running. The source notes that the decisions behind DEC's decline were made while it was most praised as well managed.

• If a new technology is simpler, cheaper, lower-margin, and your best customers can't yet use it, then do not let your normal return-on-investment screen decide; route it to a separate evaluation.
• If your best customers say they don't want it, then treat that as an expected signal of a disruptive product, not as proof the market is absent.
• If the market for the new product is small, then house it in an organization small enough that the market's size matters to it.
• If you are being praised as one of the best-managed firms in your industry, then check which transition you are currently ignoring; Sears and DEC were praised while making their fatal decisions.
• If the proposed answer is to plan better, work harder, become more customer-driven, or take a longer-term view, then distrust it; the source says these exacerbate the problem.

• Mistaking failure for poor management: you explain rival successes by their luck or your own bad execution. The tell: your post-mortem lists arrogance or bad planning, while the decisions were all defensible under standard practice.
• Over-listening to your best customers: you build new-product roadmaps only from top-account requests. The tell: every funded project is a better version of the current product, and no project serves low-end or new-use customers.
• Demanding a business case before the market exists: you require size and margin forecasts for a market that has not formed. The tell: the proposal is rejected as too small or low-margin, and a newcomer later takes the market.
• Halfway measures: you adopt the new technology inside the old organization at reduced scale. The tell: a two-year effort that is underfunded relative to the core business and has no clear market, as with DEC's response to PCs.
• Defending the old technology against a lower-cost format by improving the core: you win near-term metrics while the new format grows. The tell: strong quarterly results alongside a rival's growth in a segment you describe as unprofitable, as Sears did with discount retailing and home centers.

• ▸Technology/change under review: name the new product, process, or business model, including marketing or managerial changes.
• ▸Industry transition history: list prior architectural shifts and whether the leader held its lead after each.
• ▸Simpler/cheaper, lower margin? (yes/no + evidence): does it promise lower margins than your current product?
• ▸First market: is it emerging or insignificant by your revenue standards? Name it.
• ▸Best-customer reaction: do your most profitable customers not want it or not yet able to use it? Cite what they said.
• ▸Early adopters: who are the least profitable customers embracing it first?
• ▸Standard-process verdict: what did your usual investment screen conclude, and why?
• ▸Owning organization: which unit's customers need this, and is its size matched to the market?
• ▸Core protection: what keeps the established business healthy while this proceeds?

1. 1Write down the performance dimensions your mainstream customers have historically valued in this product (the passages' test: 'dimensions of performance that mainstream customers in major markets have historically valued'). Every classification starts from this list.
2. 2Compare the new technology against that list. If it improves performance on those dimensions, it is sustaining. This holds even when it is discontinuous, radical, or makes your existing competencies and assets obsolete.
3. 3If it performs worse on the mainstream dimensions in the near term, check whether it offers a different value proposition: cheaper, simpler, smaller, or more convenient. Underperformance plus new attributes marks it as a disruptive candidate.
4. 4Identify who values those new attributes. The passages say 'a few fringe (and generally new) customers' do. If you can name no such customers, you may be looking at a failed product and not a disruptive one.
5. 5Ignore the incremental-versus-radical axis, the component-versus-architecture axis, the technical difficulty, and the risk magnitude. The disk drive history showed these had little relationship to who led and who followed. Classify on customer value, not engineering difficulty.
6. 6Ask whether your own customers want it. The passages' pattern: when customers needed an innovation, leaders mustered the resources to build it, even a hard one. When customers did not want it, leaders could not commercialize even a technologically simple one. A 'no' from your customers on a simple technology is the warning sign.
7. 7Record the classification and the reasoning. Then choose the management approach. For sustaining innovations, the standard practices (listen to customers, allocate capital to the best returns) work. For disruptive ones, the passages say those same practices can lead to failure, and they point to organizational approaches covered in later chapters (not in these passages).

Take the disk drive industry as the passages describe it. Magneto-resistive heads and thin-film (sputtered) disks replaced earlier technology. They allowed denser recording, which is exactly what mainstream customers valued. Under the test above they are sustaining, even though they were architectural and component changes that 'rendered prior competencies irrelevant and made massive investments in skills and assets obsolete.' Industry leaders adopted them anyway. The 8-inch drive was different. It was 'technologically straightforward,' but it underperformed established drives on what mainstream customers valued and brought other attributes. The leaders stumbled on it. The classification therefore tracks customer value and not technical difficulty. The passages' footnote adds a label: the head and disk innovations were 'positive discontinuities' in an established trajectory, while the disruptive changes were 'negative discontinuities.' Established firms generally led through the first and lost their lead on the second.

• If the innovation improves performance on dimensions mainstream customers already value, then classify it as sustaining, however radical or costly it is, and expect established firms to be able to handle it.
• If it performs worse on mainstream dimensions but is cheaper, simpler, smaller, or more convenient, then treat it as potentially disruptive and look for the fringe or new customers who value those attributes.
• If your best customers do not want it and your investment process screens it out, then do not read that as proof it is unimportant. The passages say it is right at times 'not to listen to customers' and to pursue small markets with lower-margin, lower-performance products.
• If you are tempted to judge by technical difficulty or by how radical it looks, then stop. These had little relationship to leadership patterns in the disk drive history.

• Classifying by technical novelty: calling a hard, radical technology 'disruptive' because it is difficult. Tell: your argument rests on how much engineering it demands, not on what customers value.
• Dismissing a simple technology as trivial because it is technically straightforward. Tell: you say 'we could build that easily' and still never commercialize it, as with the 8-inch drive.
• Letting the customer list decide. Tell: you test the innovation only against existing mainstream customers' demands, so it always fails and you never look for new or fringe customers.
• Applying standard good-management tools (better planning, more customer focus, working harder) to a disruptive case. Tell: the effort grows and the position still erodes, as at Sears, which was praised as best-managed while it ignored discount retailing.

• ▸Innovation name: one line describing the new technology or product.
• ▸Mainstream performance dimensions: list what your mainstream customers have historically valued.
• ▸Effect on those dimensions: better, same, or worse in the near term. Note the evidence.
• ▸New value attributes: tick any of cheaper, simpler, smaller, more convenient, and describe each.
• ▸Who values the new attributes: name the fringe or new customers, or write 'none identified'.
• ▸Do current customers want it? Yes or no, with the source of the answer.
• ▸Classification: sustaining or disruptive, with a one-sentence reason that does not rely on technical difficulty.
• ▸Competency impact (for context only): does it make existing skills obsolete? Note it, but do not use it to classify.
• ▸Management implication: standard allocation process or a different organizational approach, with the reason.

1. 1Classify the innovation before you classify the firm. Ask whether it improves performance along the dimensions mainstream customers already value (sustaining) or underperforms on those dimensions while offering something cheaper, simpler, smaller, or more convenient to fringe customers (disruptive). Firm status predicts nothing until this is settled.
2. 2Tag each firm in your sample as an established incumbent (already selling in the market before the innovation emerged) or an entrant (first product comes after). Use one fixed cutoff date per innovation and apply it to every firm.
3. 3If the innovation is sustaining, predict incumbent leadership. In the disk drive data, established firms led every sustaining innovation, whether radical or incremental, hardware or software, component or architecture, competence-enhancing or competence-destroying. Do not predict an entrant win because the technology is hard or expensive.
4. 4If the innovation is disruptive, predict incumbent lag and entrant leadership. Check whether the dominant firm kept its lead into the next generation. In disk drives it did so only twice in six architectural transitions.
5. 5Test the incumbent's lag against capability. Compare the performance of the incumbent's eventual new-architecture product with entrants' products. The 8-to-5.25 and 5.25-to-3.5 incumbent drives were fully performance-competitive, so the barrier was not engineering.
6. 6Trace the lag to the incumbent's customers and financial structure. Ask whether the product promises lower margins, whether it first sells in an emerging or insignificant market, and whether the most profitable customers can't use it yet. If all three hold, expect a rational refusal to invest aggressively.
7. 7Measure adoption timing, not just adoption. Record what share of incumbents had launched by a given date (e.g., 35 percent of 5.25-inch makers had introduced 3.5-inch drives by 1988) and whether they launched only after an attack on their home market.
8. 8Check whether product performance has outrun mainstream need. Track how mainstream customers actually use the product, because that signals when the basis of competition is about to shift and a low-end opening is forming.

Take thin-film heads versus the 3.5-inch drive. Thin-film heads were a sustaining technology that was extraordinarily hard to develop. Burroughs (1976) and IBM (1979), both established firms, first incorporated them. Between 1982 and 1986 about sixty firms entered the rigid disk drive industry, yet only four, all commercial failures, tried thin-film heads in their initial products. Even performance-oriented entrants such as Maxtor and Conner Peripherals learned first on conventional ferrite heads. Firm status therefore favored incumbents here despite the difficulty. The 3.5-inch drive is the contrasting case, a disruptive architectural shift. By 1988 only 35 percent of the manufacturers established in 5.25-inch drives for the desktop PC market had introduced 3.5-inch drives. Their eventual products matched entrant performance, so engineering was not the barrier. Their customers, notably IBM and its competitors and resellers, did not yet value portable computing. The same incumbent status that produced leadership in thin-film heads produced lag in 3.5-inch drives.

• If the innovation sustains the established trajectory, then back the incumbents to lead, regardless of how radical or costly the technology is.
• If the innovation is disruptive and the firm's best customers can't use it yet, then expect an incumbent to delay and an entrant or new-market firm to lead.
• If an incumbent's late product matches entrant performance, then diagnose the delay as customer and margin driven, not technical, and look at the customer base.
• If an incumbent plans to launch only after the new technology has matured in its new applications, then treat the delay as a self-fulfilling cannibalization risk, since the launch comes as a response to attack.
• If mainstream customers' needs are growing more slowly than product performance, then treat the incumbent as exposed to a low-end entrant, however strong its current position.

• Equating incumbent weakness with technical difficulty. The tell: you explain failure by 'the technology was too hard' when incumbents led thin-film heads and disks, the hardest sustaining changes.
• Using radical-versus-incremental as the sorting variable. The tell: your forecasts treat a radical sustaining change as a threat and a simple disruptive change as safe, when the disk drive and minicomputer cases show the opposite.
• Blaming incumbent failure on arrogance or risk aversion. The tell: you recommend cultural fixes, though the same firms led every sustaining innovation.
• Assuming an entrant label means entrants win. The tell: you cheer entrants that attempted hard sustaining technologies; the few that tried thin-film heads or disks mostly folded after entry.
• Counting one dominant firm's record as an industry pattern. The tell: you generalize from one lead firm without checking the six transitions, in only two of which the dominant firm kept its lead.

• ▸Innovation: name the technology or product change being assessed.
• ▸Type: sustaining or disruptive? Justify by whether it improves performance on dimensions mainstream customers already value.
• ▸Firm and status: list each firm and mark incumbent or entrant using a single cutoff date.
• ▸Margin test: does the product promise lower margins than the incumbent's current line? Yes/no and evidence.
• ▸Market test: is it first commercialized in an emerging or insignificant market? Yes/no and evidence.
• ▸Customer test: can the incumbent's most profitable customers use it today? Yes/no and evidence.
• ▸Capability check: is the incumbent's version performance-competitive with entrants'? Note the gap, if any.
• ▸Adoption timing: share of incumbents that have launched and by what date; did launches follow an attack on the home market?
• ▸Prediction: who leads (incumbent or entrant) and what evidence would prove this wrong.

1. 1Classify the innovation first. Ask whether it addresses the well-understood needs of known actors in your current value network. If yes, it is a sustaining innovation, and the mainstream organization handles it well, whatever its technical difficulty. If it is valuable mainly in an emerging network, treat it as disruptive and keep going.
2. 2Name the value network each side lives in. Write down the rank-ordering of performance attributes and the characteristic cost structure (gross margins) of the mainstream network and of the network the innovation serves. A different attribute ranking or margin structure is the trigger for separation.
3. 3Locate the capabilities in processes and values, not in people. List the processes that would run the project (design, budgeting, sales, planning). List the values, meaning the criteria managers use to prioritize. People are flexible; processes and values are not.
4. 4Run the fit test on each. Process: does the method that works for the mainstream product also fit this one? A minicomputer design process does not fit a desktop PC. Values: do mainstream prioritization criteria, such as favoring high-margin products, rank this project low? If either fails, the mainstream organization is incapable of this task however good its staff.
5. 5Where the fit test fails, create new capability instead of coaching the old organization. Put the project in an organization whose customers need the technology, with its own processes and its own prioritization criteria. The passages preview this as Chapter 8's framework, which this guide does not reproduce.
6. 6Cut the resource-allocation link. A technology valuable only in lower-margin networks will not be viewed as profitable inside the mainstream system and will not attract resources or managerial attention. Fund the unit by the standards of its own network's cost structure, not by the parent's margin hurdles.
7. 7Replace forecast-driven approval with discovery-based planning. Assume the market forecasts and the chosen strategy are wrong. Define what must be learned, set up cheap tests to learn it, and let the unit's plan evolve. Do not require market size and return figures before entry.
8. 8Tie the unit to the customers and channels of the new network. Check that its sales path and customer feedback come from buyers who need the product now, not from the mainstream customers who would reject it.

Hydraulic excavators in the late 1940s show what the logic of autonomy looks like. Early backhoes held only ¼ cubic yard, reached about six feet and rotated 180 degrees, against cable excavators that rotated 360 degrees. Mining, general excavation and sewer contractors wanted 1–4 cubic yard buckets, so the machines served none of the established customers' needs. The entrants, including HOPTO, did not try to win those customers. They built a new application: attachments for the back of small Ford, J. I. Case, John Deere, International Harvester and Massey Ferguson tractors. Small residential contractors used them to dig narrow trenches from street water and sewer lines to house foundations, work that big cable shovels had never justified and that had been done by hand. The machines were sold through tractor and implement dealerships used to small customers, and they did the job in under an hour per house. The entrants had their own customers, channel and success criteria, and no mainstream prioritization system to override them. Autonomy exists to give an innovation that same freedom. The passages say that this is a case of entrants, not of an incumbent's separate unit, so the transferable lesson is the separation of customers, values and resource logic.

• If the innovation addresses known needs of known actors in your value network, then keep it in the mainstream organization; autonomy adds cost without benefit.
• If it is valuable only in a network with lower gross margins than yours, then do not run it through the mainstream resource-allocation process, because it will not rank as profitable and will not attract resources.
• If your process or your prioritization criteria would rate the project low or handle it badly, then reassigning talented people will not fix it; change the organization that houses the work.
• If the approval process demands quantified market size and returns before entry, then the project is in the wrong organization or under the wrong planning method; use discovery-based planning.
• If the project's first customers are small, remote from your mainstream, and unable to use the product at mainstream performance levels, then give it its own customers and channel instead of forcing it to serve your current ones.

• Assuming good people are enough. The tell: a top team is staffed onto the project, yet it is still run by the parent's processes and ranked by the parent's margin criteria, and it stalls.
• Declaring separation but keeping the mainstream's resource allocation. The tell: the unit must justify itself against the parent's margin and market-size hurdles and loses budget to higher-margin products.
• Demanding forecasts for an unknowable market. The tell: approval hinges on market-size and financial projections that nobody can credibly make, so the project is paralyzed or the numbers are invented.
• Treating all innovations as needing autonomy. The tell: sustaining projects are carved out and starved of the mainstream's customers and know-how, though established firms lead in these.
• Pointing the unit at mainstream customers. The tell: the unit's product is judged by the customers who reject it, as cable-shovel buyers would have rejected early backhoes, so its real early market is never developed.

• ▸Innovation and the customers who need it: state the product and name the actual first users.
• ▸Sustaining or disruptive: does it serve known needs within our current value network? Yes or no, with evidence.
• ▸Mainstream vs. new value network: list each network's performance-attribute ranking and typical gross margin.
• ▸Process fit: name the processes that would run the project and say whether each fits the task.
• ▸Values fit: state where the project would rank under our current prioritization criteria, such as margin.
• ▸Resource allocation: describe how the project would be funded today and whether it would survive that process.
• ▸Planning method: say whether approval requires market forecasts, or whether discovery-based learning milestones are set.
• ▸Customer and channel link: name the customers and sales channel the unit will use and show they come from the new network.
• ▸Decision: autonomous unit or mainstream, with the single finding that drives it.

1. 1Size the opening market before you pick the organization. Estimate revenue in the first emerging application, not the eventual market. The passages show that Seagate, Conner and Quantum entered markets nobody could size: hobbyist PCs, and portable computers before Compaq knew their potential. Write the first-year number down and treat it as the benchmark for the organization.
2. 2Measure the parent's growth appetite as a number. Take the revenue a project must add to move the parent's growth rate. Compare it to the initial market. If a market of a few million dollars cannot move a company of billions, a mainstream business unit will correctly rank it below every larger initiative. That ranking is the misfit, not a sign of bad judgment.
3. 3Trace who actually allocates resources day to day. Executive approval does not settle the question. DEC's executives backed the PC move, but the people making daily resource decisions never saw the sense in funding low-margin products their customers did not want. List the managers who control engineers, budget and sales time, and ask what each would cut when the project competes with a high-margin line.
4. 4Compare the cost structures. Ask what overhead and margin the new market can carry, then what overhead the parent needs to stay competitive in its core. DEC could not cut enough overhead to compete in low-end PCs because it needed those costs for higher-performance products. If the two structures cannot coexist in one unit, the project needs a separate organization.
5. 5Set the new organization's size to the market, not the other way round. IBM built an autonomous PC organization in Florida, far from its New York headquarters. It could procure components from any source, sell through its own channels, and build a cost structure suited to personal computing. Give the unit its own procurement, channel and cost authority. Staff it so a small win moves its numbers.
6. 6Define success metrics that belong to the new market. IBM's Florida unit was free to succeed on metrics relevant to personal computing. Write down two or three: revenue per new customer, gross margin at the market's price level, and learning milestones such as applications discovered. Do not apply the parent's margin or size hurdle.
7. 7Expect trial and error. The entrants sold to whoever would buy, and the dominant applications emerged from that process. Plan the budget for several cheap experiments, not one big launch, and review them often enough that a small win is visible and celebrated.
8. 8Re-check the fit as the market grows, and keep the link loose. The passages note the argument that IBM's later decision to tie its PC division closely to the mainstream organization contributed to its difficulties. Set a trigger, such as market revenue reaching a stated fraction of the parent, before any integration is discussed.

DEC launched personal computers for consumers four times between 1983 and 1995 and withdrew four times. Each launch came from within the mainstream company. The PCs were technologically simpler than DEC's minicomputers, but they carried low margins and served customers DEC's core did not. Day-to-day resource decisions went to higher-performance projects that promised upscale margins, such as the Alpha microprocessor and the move into mainframes. DEC also could not hack away enough overhead to compete in low-end PCs, because the core needed that overhead. IBM took the opposite path. It built an autonomous organization in Florida, free to source components from anywhere, sell through its own channels, and build a cost structure to fit the market. That unit succeeded in the first five years of the PC industry, when other leading mainframe and minicomputer makers missed the wave. The contrast is the construct in practice: the same company could fund a small, low-margin market only through a unit sized to it.

• If the target market is too small to move the parent's growth rate, then do not run it inside the mainstream organization. Create a separate unit whose size matches the market.
• If the new market's margins and cost structure cannot coexist with the core's overhead, then separate the organization. Do not ask one cost structure to serve both.
• If executive sponsorship exists but daily resource allocators would rank the project below core work, then expect starvation and move resource control into the new unit.
• If the market is not yet defined, then fund small, fast experiments and judge the team on learning and small wins, not on size targets borrowed from the core.
• If the new market grows large enough to matter to the parent, then reconsider integration deliberately. Do not link the unit closely before that point, since IBM's later tight linkage is cited as a possible factor in its difficulties.

• Relying on executive sponsorship alone. The tell: top leaders approve the initiative, but engineers and budget flow to higher-margin projects, as at DEC.
• Judging the new unit by the parent's metrics. The tell: the project is called unprofitable because its margins and volumes look unattractive next to the core's, which is how established firms viewed emerging markets.
• Straddling two cost structures in one organization. The tell: the team cannot cut overhead to the new market's level because the core needs the same overhead.
• Reading the small first market as the permanent ceiling. The tell: the case is rejected on current size and margins, though the entrants used it as a base and moved upmarket at 50 percent annual improvement.
• Re-absorbing the unit too early. The tell: the autonomous unit starts sharing channels, sourcing and metrics with the mainstream business before the market is large.

• ▸Initiative and target market: name the product and the first application it serves.
• ▸First-year market revenue estimate: state the number and how uncertain it is.
• ▸Parent revenue and growth-rate effect: write the revenue the parent needs to move its growth, and the percentage this market would add.
• ▸Who allocates resources day to day: list the managers who control budget and engineers, and what each would fund first.
• ▸Margin and overhead gap: compare the market's affordable cost structure with the core's required overhead.
• ▸Proposed organization: state whether the unit is inside or separate, its location, and its size relative to the market.
• ▸Autonomy granted: record whether it controls procurement, channels and cost structure (yes or no each).
• ▸Success metrics: list two or three metrics suited to the new market, and what counts as a small win.
• ▸Integration trigger and review date: state the market size that would prompt reconsidering linkage, and when you recheck.

1. 1Map the real resource flow, not the org chart. List where the last 12 months of capital, engineering headcount, and management attention went, then trace each allocation to the customer or investor demand behind it. The theory says customers and investors dictate spending even when managers think they control it.
2. 2Inventory your idea-killing system. Write down each stage where a proposal is screened (margin thresholds, customer-validation requirements, sales-forecast demands) and note how many proposals it has killed for lacking a known customer. The highest-performing firms are the best at killing ideas their customers don't want, so a strong filter is a sign of dependence, not a defect.
3. 3Classify each proposed project as sustaining or disruptive by one test: do your current mainstream customers need it? Sustaining projects (needed by customers) get funded even when huge, long, and risky. Disruptive ones (lower margin, customers don't want them yet) stall, even when the technology is simple.
4. 4Check the demand-assurance asymmetry. For each stalled project, ask whether the stall came from technical difficulty or from the absence of an assured buyer. If the technology is straightforward and the project still languishes, resource dependence is the cause.
5. 5Test the margin fit. Compare the project's expected gross margin with your current cost structure. A cost structure tailored to high-end markets cannot profitably serve low-end or emerging ones, so the project will look unattractive however good it is.
6. 6Account for the individual-manager layer. Identify who would have to champion the project and what a failed bet costs that person's career. Because finding new markets inherently involves failure, managers who cannot afford failure will not back it, which compounds the pull of customers.
7. 7Decide the structure: set up an autonomous organization with its own cost structure, built to profit at low margins and charged with building an independent business around the disruptive technology. Embed it among a different set of customers, those who want the disruptive product, so their demand supplies the resource pull.
8. 8Give the new unit plans to learn, not plans to execute. Judge it on what it discovers about the market, since failure is intrinsic to finding applications, rather than against the forecast-and-execute standard used for sustaining technologies.

The disk drive industry is the source's test case. Established leaders stayed atop wave after wave of sustaining technologies, meaning those their customers needed, and placed huge, long, risky bets to develop them whenever demand was assured. When the technology was a simpler disruptive one with no assured demand, the same firms could not make even technologically straightforward bets. Their customers did not want the product, margins were lower, and no manager wanted to champion a project that might fail because the market was not there. The result: 65 percent of the companies entering the disk drive industry tried to do so in an established market rather than an emerging one. The source's remedy is structural. The only cases where mainstream firms built a timely position in a disruptive technology were those where managers created an autonomous organization with a different customer set. Toyota shows the cost-structure side. It entered the North American market with the low-priced Corona, then moved up-market for better margins, and in the process added costs that made the entry tier's margins unattractive.

• If a project is needed by your current mainstream customers, expect funding to flow even at large scale and high risk. If it is wanted only by customers you don't yet serve, expect it to stall in the mainstream organization.
• If the technology is simple but the project languishes, diagnose resource dependence (no assured buyer, margins that don't fit) before diagnosing technical or talent problems.
• If your cost structure is tailored to high-end markets, don't try to make low-margin disruptive opportunities work inside it. Create an independent organization with a cost structure built for those margins.
• If the project's demand is unknown, don't ask a mainstream manager to champion it under a plan-then-execute standard. Make it a plan-to-learn effort in a unit where failure is survivable.
• If you cannot name the customers who want the disruptive product, find them first. Aligning the innovation with the "right" customers is what raises its odds of getting resources.

• Treating the stall as a motivation or competence problem and exhorting the team harder. The tell: the technology is technologically straightforward, yet the project keeps losing in every budget cycle.
• Fighting the pressure by ordering the mainstream organization to fund the disruptive project freely. The tell: the project gets nominal funding, but people and processes keep reallocating critical financial and human resources to customer-backed work.
• Housing the disruptive effort inside the mainstream business and selling it through the existing sales force. The tell: the new product is developed for established customers and established markets, not for new applications (Burgelman found that new ventures' entry to customers usually came through the salespeople of established lines, which helped with established markets but not with finding new applications).
• Judging success by the strength of the idea-killing system. The tell: leaders celebrate the filter's rigor while every rejected idea shares one trait, that current customers didn't ask for it.
• Ignoring the career risk that individual managers carry. The tell: capable managers decline to sponsor projects whose market is unproven, and proposals arrive only when demand is already assured.

• ▸Project name: the proposal or technology under review.
• ▸Who wants it today: name the customers and investors asking for it, or write "none identified".
• ▸Sustaining or disruptive: sustaining if your mainstream customers need it, disruptive if they don't want it yet.
• ▸Expected gross margin vs. current cost structure: state the margin and whether your overhead can make it profitable.
• ▸Funding history: money, headcount, and attention received over the last 12 months, and the stated reason for any cuts or delays.
• ▸Stall diagnosis: technical barrier or missing assured demand? Cite the evidence.
• ▸Champion career risk: who would sponsor it and what a failed bet would cost them.
• ▸Structural decision: stay in mainstream unit, or create an autonomous organization (cost structure, leader, target customers).
• ▸Learning plan: the market questions the unit must answer and how it will judge progress without a forecast.