Advanced strategy · Module 4

Disruption, digital transformation and business model innovation

Why do well-run incumbents lose to entrants whose products are worse at first, and what can they do about it?

Beyond the IB course UndergraduateMBAProfessional About 30 minutes · 3,179 words · checked 16 November 2026

From the IB course to this module

Subtopic 5.8 introduces incremental and disruptive innovation, and 5.9 the technologies behind digital change: data analytics, cloud computing, artificial intelligence and big data. The IB course defines the terms; this module explains the mechanisms. It sets out Christensen's theory of disruption and the evidence against it, why capable incumbents fail to respond, how business models create and capture value, and why digital transformation is an organizational problem before it is a technical one.

Section 01: Conceptual foundation

Three ideas are often confused. Disruptive innovation, in Christensen's precise sense, is a process in which an entrant starts with a product that is worse on the dimensions mainstream customers value, but cheaper, simpler or more convenient, and wins either the least demanding customers (a low-end foothold) or people who were not customers at all (a new-market foothold). Incumbents rationally ignore it, because those customers are unattractive and the margins thin; the entrant then improves until it is good enough for the mainstream. A better product aimed at the incumbent's best customers is a sustaining innovation, and incumbents usually win those battles.

A business model is the logic by which a firm creates value for customers and captures part of it: what it offers, to whom, through which activities and partners, and how it is paid. Business model innovation changes that logic, for example by selling a service instead of a product, or by giving the core product away and charging for something else.

Digital transformation is the use of data, software and connectivity to change how value is created and captured, not merely to make existing processes cheaper. Its distinctive economics, near-zero marginal cost, network effects and learning from data, reward scale and speed far more than the industrial economics around which most incumbents are organized.

Section 02: Advanced explanation

The mechanism of disruption

Christensen (1997) built the theory on the history of disk drives, in which each new, smaller generation was first adopted by makers of smaller computers and dismissed by the leading drive makers, whose best customers did not want it. The mechanism is a mismatch of trajectories: the performance customers can use rises slowly, while the performance technology delivers rises fast, so a product that is inadequate today can be good enough tomorrow.

The timing can be estimated. Suppose an entrant's product delivers half of what mainstream customers require, its performance improves by 25% a year, and their requirement grows by 5% a year. The gap closes by a factor of 1.25 divided by 1.05 each year, so the entrant reaches the requirement in ln 2 divided by ln (1.25 / 1.05), about four years. If its performance improves by 12% a year instead, the wait is about eleven years. Small differences in improvement rates decide whether disruption is urgent or distant.

Incumbents ignore the entrant for reasons that are rational in the short run. Resources follow the customers and margins that sustain the business: a proposal to sell a €300 product at a 20% margin loses an internal budget contest to one that improves a €1,000 product sold at 40%. Bower and Christensen (1995) located the failure in the resource allocation process, not in managers' intelligence.

Why capable incumbents struggle

Disruption is one route to incumbent failure among several. Henderson and Clark (1990) showed that architectural innovations, which change how known components fit together, defeat incumbents whose organization mirrors the old architecture. Tripsas and Gavetti (2000) found that Polaroid built strong digital imaging capabilities in the 1980s and 1990s but failed to exploit them, because managers believed profit came from consumables, as with instant film: the obstacle was cognition, not capability.

Gilbert (2005) separated two kinds of inertia in newspapers' response to online news. Under threat, firms overcame resource rigidity and invested heavily, but suffered routine rigidity, running the new venture with the processes of the print business. Separating the venture let it develop its own routines, consistent with O'Reilly and Tushman's (2004) case for ambidextrous organizations, with separate units for exploring and exploiting joined at the top.

Incumbents also have advantages the disruption story underplays. Teece (1986) showed that when an innovation is easy to imitate and depends on complementary assets (distribution, regulatory approval, manufacturing, brand), the owners of those assets often capture its value, and incumbents usually own them. The question for an incumbent is not only whether it can build the new product but whether its complementary assets will still matter.

Business models and value capture

Teece (2010) defined a business model as the logic by which a firm delivers value to customers and converts it into profit, and stressed that the same technology can be commercialized through different models with very different results. Three choices define a model: what is given away and what is charged for; which activities the firm performs and which partners perform; and how revenue is earned, per unit, by subscription, per transaction, by licence or through advertising.

Giving the core product away is business model innovation at its sharpest. It makes sense when the product would be copied or competed down to a zero price anyway, and when demand for a complement the firm controls rises with use of the free product. The arithmetic is unforgiving: if a million people use a free product and 2% convert to a €20 monthly plan, revenue is €4.8 million a year; if serving each user costs €0.50 a month, the free tier costs €6 million, and the model loses money until conversion passes 2.5% or costs fall.

Incumbents face a different arithmetic: cannibalisation. A pay-television customer paying €40 a month at a €15 margin who moves to the company's own €8 streaming service at a €2 margin costs it €13 a month. The move is still right if the customer would otherwise leave for a rival; the relevant comparison is not the old margin but what remains if the incumbent does nothing.

Digital transformation as organizational change

Digital technologies change industry economics in three ways: they cut the marginal cost of reproducing and distributing information towards zero, they create network effects that reward the largest platform, and they let products improve with use as data accumulate. Each tilts competition towards scale, speed and integration across functions.

Most incumbents are organized for different economics: business units with their own profit and loss, functional silos, annual planning and hierarchies built to control physical operations. Digital transformation therefore tends to fail as a technology project and succeed, when it does, as an organizational one: redrawing units around customers or data rather than products, moving decision rights to the teams that build and test, and changing how investment is allocated. Adding a digital team to an unchanged organization reproduces the routine rigidity Gilbert described.

The hardest problem is sequencing. The old business usually funds the new one, and cutting its costs to pay for the investment can hasten its decline. Leaders must judge how fast to move resources while the old model still pays the bills, a judgement for which there is no formula.

Section 03: Models and frameworks

Where each comes from, how to use it, and where it stops being useful.

Christensen (1997)

Disruptive innovation

Entrants whose products are worse on mainstream measures but cheaper, simpler or more convenient gain a low-end or new-market foothold that incumbents rationally ignore. The entrant then improves until it meets mainstream needs, and the incumbent's customers follow.

How to use it

Test a threat against the theory's conditions before calling it disruptive. Does the entrant start with customers the incumbent does not want, or with non-consumers? Is its performance improving faster than mainstream requirements? Is the incumbent's resource allocation steering it away? If all three hold, consider a separate unit with its own cost structure and measures of success.

Where it fails

The theory explains past cases better than it predicts new ones. Many claimed disruptions do not meet its conditions, and many entrants that do meet them still fail. It underplays incumbents' complementary assets and their ability to respond once the threat is clear.

Clayton M. Christensen (1997). The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail. Harvard Business School Press. Joseph L. Bower and Clayton M. Christensen (1995). Disruptive technologies: catching the wave. Harvard Business Review 73(1). Clayton M. Christensen, Michael E. Raynor and Rory McDonald (2015). What is disruptive innovation?. Harvard Business Review 93(12).

Teece (1986)

Profiting from innovation

Who captures the value of an innovation depends on how easily it can be imitated and on who owns the complementary assets needed to commercialize it, such as manufacturing, distribution, brand or regulatory approval.

How to use it

For any new technology, ask whether patents, secrecy or complexity protect it, and which complementary assets it needs. If protection is weak and the assets are specialized and held by others, partner with or license to the owners, or secure the assets yourself. If protection is strong, the innovator can capture value directly.

Where it fails

It explains who profits from an innovation, not how to generate one. The strength of protection and the scarcity of complementary assets change over time; digital distribution, for example, has made some assets far less scarce.

David J. Teece (1986). Profiting from technological innovation: implications for integration, collaboration, licensing and public policy. Research Policy 15(6).

Osterwalder and Pigneur (2010)

The business model canvas

The canvas describes a business model in nine blocks: customer segments, value propositions, channels, customer relationships, revenue streams, key resources, key activities, key partners and cost structure.

How to use it

Draw the incumbent's model and the challenger's side by side to see where the challenger creates or captures value differently. Then design alternatives by changing one block at a time, such as what is charged for or who performs an activity, and estimate the economics of each, including cannibalisation of the existing model.

Where it fails

The canvas describes a model but cannot say whether it will beat rivals' models; it has no place for competitors or for time, and it can encourage tidy boxes over the hard economics of conversion, cost and capture.

Alexander Osterwalder and Yves Pigneur (2010). Business Model Generation. John Wiley and Sons. David J. Teece (2010). Business models, business strategy and innovation. Long Range Planning 43(2-3).

O'Reilly and Tushman (2004)

Ambidexterity

An established firm can exploit its existing business and explore new ones at the same time by giving exploratory units their own processes, cultures and measures, integrating them at senior level and sharing selected resources.

How to use it

Separate the new venture where its routines must differ from the core's (pricing, speed, risk tolerance, measures of success), and decide which assets it may draw on, such as brand, customer data or distribution. Give it a sponsor with authority over both businesses, and agree in advance how far it may compete with the core.

Where it fails

Separation can starve the new unit of the core's assets or leave it a showcase without influence. Integration at the top depends on leaders who can hold two conflicting logics, and the evidence on when separation works comes mostly from case studies.

Charles A. O'Reilly III and Michael L. Tushman (2004). The ambidextrous organization. Harvard Business Review 82(4). Clark G. Gilbert (2005). Unbundling the structure of inertia: resource versus routine rigidity. Academy of Management Journal 48(5).

Section 04: Real business applications

Real organizations appear only with dated, cited facts; illustrative examples are labelled.

Real organization, 2025-12

MultiChoice: a pay-television incumbent under pressure from streaming

What the company reported. For the year to March 2025, MultiChoice reported 14.5 million linear pay-TV subscribers, down 8% after an 11% fall the year before, and revenue of ZAR 50.8 billion, down 9% as reported but up 1% organically, with subscription revenue down 11%. It blamed a weak consumer environment and changes in technology that were shifting consumer behaviour, including piracy, streaming services and social media. Price increases of 5.7% in South Africa and an average of 31% in local currency in the rest of Africa offset part of the loss. Newer businesses grew fast: DStv Stream revenue rose 48%, Showmax paying customers 44% (below its initial targets), and the payments business Moment processed $635 million, seven times the year before. The company cut the price of some packages and piloted weekly subscriptions in Uganda to match customers' cash flows. Canal+ completed its acquisition of the group in December 2025.

What it suggests. Streaming fits Christensen's pattern only in part: it began as a cheaper, more convenient option for viewers pay-TV served poorly, but the losses also reflect household budgets under strain. The response combined price rises on loyal customers, new business models such as weekly plans and an own streaming service, and diversification into payments and betting, and it ended in consolidation with a larger foreign rival.

Real organization, 2026-02

WPP: an advertising group redraws itself around a platform

What the company reported. For 2025 WPP reported revenue less pass-through costs of £10,176 million, down 5.4% like-for-like, with a fourth-quarter fall of 6.9%. Its headline operating margin fell from 15.0% to 13.0%; after a £641 million goodwill impairment, reported operating profit was £382 million, and the dividend fell from 39.4p to 15.0p. Its chief executive, in post for six months, said clients were navigating AI disruption, but attributed WPP's underperformance to "excessive organizational complexity, a lack of an integrated operating model and inconsistent strategic execution". The new plan, Elevate28, turns the holding company into a single company with four operating units unified by its WPP Open platform, with £500 million of gross annual savings by 2028 at a cash cost of about £400 million.

What it suggests. The company's own diagnosis is organizational rather than technological: AI raises the stakes, but the problem named is a federation of agencies that cannot offer clients an integrated, data-driven service. That is the pattern this module describes, in which transformation means redrawing the organization around a platform. Whether AI will shrink demand for agency services however WPP organizes is a separate question the plan cannot settle.

Real organization, 2025-12

Mistral AI: giving the model away and charging for the rest

What the company reported. In September 2023 Mistral AI released its first model under the Apache 2.0 licence, usable without restriction, arguing that open models would outperform proprietary ones for most uses, and promised optimized models, hosted services and dedicated deployments for enterprises. In June 2025 it launched Mistral Compute, an infrastructure offering from bare-metal servers to managed platforms, aimed at customers who want an alternative to US or China-based providers. In September 2025 it raised €1.7 billion at a post-money valuation of €11.7 billion, in a round led by the lithography equipment maker ASML, which described the investment as a strategic partnership. In December 2025 it released the Mistral 3 family under the same licence, distributed through its own platform and many third-party clouds, with custom training offered to enterprises.

What it suggests. Giving model weights away follows Teece's logic: where a product can be copied or matched, value is captured through complementary assets, here deployment, customization, infrastructure and customers' control over where their data is processed. The open releases build adoption and trust, and the paid offerings capture value. The risk is that the complements are contestable too, since cloud providers that distribute the models also sell infrastructure.

Section 05: Case studies

Fictional case, simulated figures

Clearwater Learning and the AI tutor

Clearwater Learning is a fictional publisher of university textbooks and digital courseware. Lecturers choose its titles and students pay. Print textbooks earn a 45% gross margin and digital courseware sold to universities 60%; an AI tutor launched in 2023 has a gross margin of minus 20% at its present scale.

A fictional start-up, Socrat, sells students an AI tutor directly for €6 a month, without involving lecturers, and claims 400,000 subscribers. Its answers are less reliable than Clearwater's on advanced topics, but students rate it as more convenient. Clearwater's board must decide how to respond before the next budget.

Clearwater Learning: revenue by product line (€m) and operating margin
Year Print textbooks Digital courseware AI tutor Total Operating margin
202212030015018%
202310436214215%
20248840613412%
20257042141269%

Simulated data. Gross margins by line, which have not changed over the period, are given in the case.

Questions

  1. Calculate gross profit by product line for 2022 and 2025. How much of the decline comes from print, and how much do the newer lines offset?
  2. Is Socrat a disruptive entrant in Christensen's sense? Test it against the theory's conditions of foothold, trajectory and the incumbent's incentives.
  3. Draw Clearwater's and Socrat's business models on the canvas. Which blocks differ, and which of Clearwater's complementary assets still matter?
  4. Should the AI tutor be run inside the courseware business or as a separate unit? Use the evidence on ambidexterity and routine rigidity.

Section 06: Critical evaluation

What the evidence supports, where the debate is, and what the models leave out.

How well does disruption theory hold up? Its core mechanism, that incumbents underinvest in innovations their best customers do not want, is well documented in industry histories. Its predictive power is weaker. King and Baatartogtokh (2015), using experts' assessments of 77 cases cited as disruptions, found that only a small minority met all of the theory's conditions. Lepore (2014) criticized the choice of cases and the theory's use as a general account of innovation. Christensen, Raynor and McDonald (2015) narrowed the definition and acknowledged that the theory had been applied too loosely. Gans (2016) separated demand-side disruption, where incumbents neglect new customers, from supply-side disruption, where they cannot absorb a new architecture, and argued that incumbents can often respond, for example by acquiring entrants.

Incumbents often survive. Teece's account of complementary assets is as well supported as the disruption cases. The practical lesson is to analyse which assets a new technology makes obsolete, not to assume that incumbents are doomed.

Evidence on transformation. Much of what is written about the success and failure rates of digital transformations comes from consultancy surveys with unclear samples and definitions, and the academic evidence is mostly case-based. Statements such as "most transformations fail" are interpretation, not established fact.

This module's cases. The real applications support a cautious reading: the incumbents faced several pressures at once, macroeconomic, technological and organizational, and their own documents attribute weakness to a mix of them. Attributing their difficulties to disruption alone would select the evidence to fit the theory.

Section 07: Decision exercise

A fictional situation. Decide before you read the guidance.

Harbourline: answering an embedded-insurance challenger

Harbourline is a fictional insurance broker with revenue of €400 million, most of it from commissions on home and travel insurance sold through its call centres and website. Travel and property booking platforms have begun offering insurance at the point of sale through Plume, a fictional insurtech that prices policies instantly from the platforms' data. Plume's policies cover less but cost 30% less and take one click to buy. Harbourline's travel commissions fell 12% last year, and its board has €50 million to invest.

  1. Build a separate digital venture

    Create a new brand outside the existing business, with its own technology, pricing and targets, to sell simple policies through platforms. This avoids the core's routines, but may cannibalise it and struggle for access to its insurer relationships.

  2. Transform the core

    Rebuild Harbourline's systems for instant pricing and integrate with platforms under its own brand. This keeps one organization and its relationships, but is slower and constrained by existing processes and commission structures.

  3. Acquire Plume

    Buy the challenger at a high multiple of revenue and run it separately. This is the fastest route to the capability, but expensive, and Plume's platform partners may not stay.

  4. Partner and supply

    Become a white-label provider of claims handling and insurer access to platforms and insurtechs, including Plume. This uses Harbourline's complementary assets, but accepts a lower-margin role without the customer relationship.

Judge the options on

  • Whether the threat meets the conditions for disruption
  • Which of Harbourline's assets remain valuable
  • Cannibalisation and its net effect on profit
  • Speed and organizational feasibility
  • Cash at risk and the options preserved
Read the guidance

Test the threat first. Plume starts with simpler, cheaper policies for customers who value convenience at the moment of purchase, a classic low-end or new-market foothold, and its pricing improves with platform data. If platforms become the main point of sale, Harbourline's call centres and website lose relevance, so the threat is serious even though Plume's products are worse today.

Then ask which assets still matter. Insurer relationships, claims handling and regulatory permissions are complementary assets that platforms and Plume may need; the website and call centres may not be. That makes partnering more attractive than it first appears, because it sells what remains scarce. A separate venture fits the evidence on routine rigidity but needs access to those assets. Transforming the core is the most comfortable option and the most exposed to the incumbent's habits, and acquisition buys speed at a price that assumes Plume's platform relationships will last.

A defensible answer combines a separate venture or a partnership with a clear limit on the cannibalisation Harbourline will accept, and names the signal that would trigger a larger commitment, such as the share of travel policies sold at the point of booking.

Section 08: Further reading

  • Clayton M. Christensen (1997). The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail. Harvard Business School Press.

    The original statement of the theory, built on the disk drive industry. Read it for the mechanism, then read its critics.

  • Clayton M. Christensen, Michael E. Raynor and Rory McDonald (2015). What is disruptive innovation?. Harvard Business Review 93(12).

    The authors' own correction of how the term is misused.

  • Andrew A. King and Baljir Baatartogtokh (2015). How useful is the theory of disruptive innovation?. MIT Sloan Management Review 57(1).

    A careful test of the theory against its own conditions.

  • David J. Teece (1986). Profiting from technological innovation: implications for integration, collaboration, licensing and public policy. Research Policy 15(6).

    Why innovators often fail to profit from their innovations, and why complementary assets matter.

  • David J. Teece (2010). Business models, business strategy and innovation. Long Range Planning 43(2-3).

    A clear account of what a business model is and how it differs from strategy.

  • Clark G. Gilbert (2005). Unbundling the structure of inertia: resource versus routine rigidity. Academy of Management Journal 48(5).

    Resource rigidity and routine rigidity, from newspapers' response to the internet.

  • Joshua Gans (2016). The Disruption Dilemma. MIT Press.

    An economist's balanced assessment of disruption theory and of incumbents' options.

Every work cited here is listed in the platform's sources register. Real organizations appear only with facts from the sources shown; everything else is labelled illustrative or fictional.