Entrepreneurship and innovation · Module 4

Scaling, innovation management, failure and recovery

What breaks when a venture grows, how established firms manage innovation that threatens their own business, and what the evidence says about failure and recovery.

Beyond the IB course UndergraduateMBAProfessional About 25 minutes · 2,781 words · checked 16 November 2026

From the IB course to this module

The IB course covers economies and diseconomies of scale and internal and external growth (1.5), research and development, incremental and disruptive innovation and intellectual property (5.8), insolvency and bankruptcy at HL (3.6) and cash flow (3.7). The Lab briefing Scaling, failure and exit sets out the patterns. This module adds the stages of growth and the crises between them, the operations of scaling (ramp-up, yield and learning curves), exploration against exploitation and the ambidextrous organization, stage-gate processes and innovation portfolios, the mechanism and limits of disruption, how failure is measured, insolvency as a recovery tool, and the evidence on learning from failure.

Section 01: Conceptual foundation

Scaling means growing output faster than costs and complexity. It fails when commitments, such as factories, hires and contracts, run ahead of the capabilities needed to honour them: a process that cannot yet make good product reliably, or an organization still run as if it had twenty people.

Innovation management means organizing to create new value while running the existing business well. The two compete for the same money, people and attention, and the existing business usually wins, because its returns are nearer and more certain.

Failure is common, often mismeasured and not always the end. A closed business may have been a success for its owner; a bankrupt one may leave factories, people and knowledge that others put back to work. The useful questions are how to fail cheaply, how to learn from it and how to recover value when it happens.

Section 02: Advanced explanation

What breaks when a venture scales

Greiner (1972) described organizations growing through phases of steady evolution, each ending in a revolution. Growth by creativity ends in a crisis of leadership, when founders cannot manage everything; growth by direction ends in a crisis of autonomy, as managers chafe at central control; delegation ends in a crisis of control; coordination ends in a crisis of red tape; and collaboration in a crisis he could not yet name. Scale-ups often meet the first two crises within a few years, because headcount grows faster than management systems.

Hoffman and Yeh (2018) argued for blitzscaling: in markets where the largest player takes most of the value, prioritize speed over efficiency even at high cost. Critics reply that the cases are drawn from survivors, and that the strategy burns capital and fragility into businesses whose markets were never winner-take-most.

In physical businesses, scaling is an operations problem. Good output equals capacity × yield × uptime, so a new plant running at 50% yield wastes half its materials, and raising yield to 75% increases good output by half with no extra inputs. Costs fall with experience: Wright (1936) observed that the labour needed per aircraft fell by a constant percentage with each doubling of cumulative output, so on an 80% curve the 1st, 2nd, 4th and 8th units cost 100, 80, 64 and 51. Premature scaling commits the fixed costs of the later stages before the early ones have been learned.

Managing innovation: exploration, exploitation and portfolios

March (1991) distinguished exploration (search, variation, experimentation) from exploitation (refinement, efficiency, execution). Organizations drift towards exploitation because its returns are surer and sooner; the result can be a competency trap, in which the firm becomes ever better at something that matters less.

O'Reilly and Tushman (2004) studied established firms launching breakthrough innovations and found the most successful were organized ambidextrously: they kept exploratory units separate, with their own processes, structures and cultures, while integrating them through a senior team that shared one strategy and allocated resources between old and new. The alternatives are to spin new ventures out or to buy them.

Process tools help but carry their own risks. Stage-gate systems (Cooper, 1990) divide development into stages separated by go or kill gates with explicit criteria; they discipline spending but can kill radical projects whose value cannot yet be shown in a forecast, which is why many firms now blend them with shorter, experimental cycles. Nagji and Tuff (2012) adapted the Ansoff matrix into a portfolio of core, adjacent and transformational initiatives, and reported that high-performing firms in their sample allocated roughly 70%, 20% and 10% of innovation resources to each, with the right balance varying by industry.

Disruption: the mechanism and its limits

Christensen (1997) explained why well-managed firms lose to entrants. Incumbents listen to their best customers and invest in sustaining innovations that those customers value and that earn higher margins. Disruptive innovations start worse on the attributes mainstream customers care about but are cheaper or simpler, and win a foothold at the low end or among new customers. Because they improve faster than mainstream customers' needs grow, they eventually become good enough, and incumbents whose resource allocation favoured their core find they have waited too long. His prescription was to give disruptive ventures separate organizations with cost structures suited to small, low-margin markets.

The theory is powerful and much abused. It is often applied after the fact to any successful entrant; critics argue that it predicts poorly in advance, that many apparently disruptive entrants fail, and that incumbents sometimes win by acquiring or imitating the entrant. Not every new technology is disruptive in Christensen's sense: one that is better and more expensive from the start is a sustaining innovation, however novel.

The managerial core is resource allocation. An incumbent that launches a business likely to cannibalise its own must fund losses that its normal criteria would reject, often for years, which is why such ventures are frequently shared with partners or run at arm's length.

Failure, its measurement and recovery

Failure is harder to measure than it looks. Many closures are not failures: Headd (2003) showed that a substantial share of owners of closed businesses regarded them as successful when they closed, because they sold, retired or moved on. Rates differ by whether one counts establishments or firms, closures or losses, and whether venture-backed "failure" means losing all capital or simply not returning it. Post-mortem lists of failure causes are self-reported, and founders tend to blame external causes.

Recovery takes several forms. A pivot changes the model while keeping the team. Restructuring reduces debt or scope. Insolvency law offers routes that preserve value: under Chapter 11 in the United States, management usually stays in control while the business keeps trading, with new debtor-in-possession financing ranking ahead of old debts; in many other systems, including Swedish bankruptcy, a court-appointed trustee takes over and sells the business or its assets. Even when the company dies, its factories, intellectual property and people can be redeployed by a buyer.

What do founders learn? Gompers, Kovner, Lerner and Scharfstein (2010) found that entrepreneurs with a previous success were more likely to succeed again than first-time founders, while those whose earlier ventures failed did about as well as first-timers: evidence of skill or reputation, not of learning from failure. Shepherd (2003) argued that learning depends on how founders recover from the grief of failure. McGrath (1999) framed intelligent failure: small, staged bets that end quickly and cheaply are how uncertain opportunities are explored, so the aim is to fail at low cost, not to avoid failing.

Section 03: Models and frameworks

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

Greiner (1972)

Phases of organizational growth

Organizations grow through phases of evolution separated by crises of leadership, autonomy, control and red tape; each phase's solution becomes the next phase's problem.

How to use it

Locate the venture's phase from its size, age and symptoms, anticipate the next crisis and change structure, delegation and systems before it arrives rather than after.

Where it fails

The sequence is a stylised description, not a law; fast-growing firms can meet several crises at once, and digital tools have changed how coordination scales.

Larry E. Greiner (1972). Evolution and revolution as organizations grow. Harvard Business Review.

March (1991); O'Reilly and Tushman (2004)

Exploration, exploitation and ambidexterity

Organizations must balance exploring new possibilities with exploiting existing ones. Ambidextrous firms separate exploratory units while integrating them through a senior team.

How to use it

Give exploratory ventures their own metrics, budgets and processes; protect them from the core's resource allocation; and integrate at the top where they need the core's assets.

Where it fails

Separation can starve new units of the core's capabilities and customers, and senior integration depends on leaders willing to fund losses against the core's interests.

James G. March (1991). Exploration and exploitation in organizational learning. Organization Science. Charles A. O'Reilly III and Michael L. Tushman (2004). The ambidextrous organization. Harvard Business Review 82(4).

Christensen (1997)

Disruptive innovation

Entrants with cheaper, simpler products win footholds that incumbents rationally ignore, then improve until they serve mainstream customers, displacing the incumbents.

How to use it

Test whether an entrant starts worse on mainstream attributes but improves faster than customers' needs; if so, consider a separate unit with a suitable cost structure, or acquisition.

Where it fails

It is easier to apply in hindsight than in advance; many cases fit only loosely, and incumbents often respond successfully.

Clayton M. Christensen (1997). The Innovator's Dilemma: When New Technologies Cause Great Firms to Fail. Harvard Business School Press. Robert G. Cooper (1990). Stage-gate systems: a new tool for managing new products. Business Horizons.

McGrath (1999)

Real options reasoning and intelligent failure

Treat uncertain ventures as options: invest small amounts to learn, expand when evidence is good and stop quickly and cheaply when it is not. Failure is an expected cost of exploring.

How to use it

Stage commitments, set the evidence that would trigger expansion or exit before each stage, and judge managers on the quality of their decisions rather than only on outcomes.

Where it fails

Options are valuable only if they are actually abandoned; organizations find stopping hard, and some opportunities cannot be staged (a factory cannot be built a little).

Rita Gunther McGrath (1999). Falling forward: real options reasoning and entrepreneurial failure. Academy of Management Review. Paul Gompers, Anna Kovner, Josh Lerner and David Scharfstein (2010). Performance persistence in entrepreneurship. Journal of Financial Economics.

Section 04: Real business applications

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

Real organization, 2025-08

Northvolt: when scaling outran capability

In September 2024 Northvolt, founded in 2016, announced the first outcomes of a strategic review: it put its cathode materials plant into care and maintenance, ended a second cathode project and sought partners for its battery systems business. Two weeks later it suspended a 30 GWh expansion of its Skellefteå gigafactory to focus on ramping up the first 16 GWh, and planned about 1,600 redundancies, some 20% of its global workforce; it said cell production there had tripled since the start of the year. In November it filed for Chapter 11 in the United States, gaining access to about $145 million of cash collateral and $100 million of new financing from a customer, and its co-founder stepped aside as chief executive. In March 2025 it filed for bankruptcy in Sweden, citing rising capital costs, geopolitical instability, supply chain disruption, shifts in demand and internal problems in its production ramp-up; its interim chairman said that output from its serial production lines had doubled and yield had improved by 50% since September, and that operational cash outflow had been cut by 55%. In August 2025 Lyten agreed to buy the remaining assets in Sweden and Germany and planned to restart production.

Interpretation. Northvolt built cells, cathode materials, recycling and battery systems at once, before the core process ran at acceptable yield. Its own figures show yield was improving when the money ran out. The assets survived the company, a reminder that recovery can mean redeployment rather than survival.

Real organization, 2025-06

MultiChoice and Showmax: an incumbent funds its own challenger

MultiChoice, whose pay-television business spans South Africa and the rest of Africa, relaunched its streaming service Showmax in February 2024, built on the sports streaming capabilities of Peacock and offering Premier League football on mobile among other plans. In the year to March 2025, Showmax's paying customer base grew by 44% (excluding discontinued packages), while its trading loss grew by 88% on the company's organic, constant-currency measure, to about ZAR 5.0 billion from ZAR 2.6 billion. The company attributed the losses to the start-up nature of the business, with a step-change in content costs and higher platform costs, and listed narrowing Showmax's losses among its priorities. Comcast has shared 30% of Showmax's losses since buying a stake in the business in April 2023.

Interpretation. This is ambidexterity in practice: a profitable incumbent funding a challenger that may cannibalise its own satellite business, sharing the losses with a partner that brings capabilities. Rising subscribers alongside rising losses is the J-curve of a new venture inside a large company, and the core's patience decides whether it is allowed to reach the upturn.

Illustrative

What a failure statistic counts

Of 1,000 new firms (illustrative), 520 have closed after five years: 190 were sold or merged, 110 closed with owners satisfied (retiring or moving on) and 220 closed with losses to owners or creditors. A "52% failure rate" counts every closure; a 22% rate counts only closures with losses. Both can be reported honestly; the reader must ask which is being counted.

Section 05: Case studies

Fictional case, simulated figures

Greenstack Farms: scaling before yield

Greenstack Farms (fictional) grows salad leaves in indoor vertical farms and sells them to supermarkets at an average of £3.40 a kg. After a successful pilot it built two larger farms, and the board now wants to commit to farms 4 and 5, citing £30 million of signed supply contracts. Staff numbers have grown from 40 to 210 in two years, and the founders still approve every hire and purchase over £5,000.

Greenstack Farms, output and unit cost by farm
Farm Design capacity (tonnes a year) Months operating Output as % of design Cost per kg (£)
Pilot, Leeds4003695%3.37
Farm 2, Bristol1,2001470%3.54
Farm 3, Glasgow1,200650%4.40

Each 1,200-tonne farm has fixed costs of £1.8 million a year and variable costs of £1.40 a kg; the pilot's fixed costs are £0.75 million a year. Simulated data.

Questions

  1. Calculate the output, as a share of design capacity, at which a 1,200-tonne farm breaks even at £3.40 a kg.
  2. Using the exhibit, explain why signed supply contracts do not remove the main risk in building farms 4 and 5 now.
  3. Which of Greiner's crises is the company approaching, and what would you change in how it is run?
  4. Design a staged plan with gates. What evidence should farm 3 provide before farm 4 is approved, and what would trigger stopping?

Section 06: Critical evaluation

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

Several findings are well established: organizations change their structures as they grow and struggle at the transitions; learning curves and yield improvement shape the economics of new factories; incumbents tend to underinvest in innovations that do not serve their existing customers; and venture outcomes are highly skewed.

Much popular thinking about failure is weaker than it sounds. Base rates depend on definitions, as the illustrative statistic shows, so figures quoted without a definition should be distrusted. Survivorship bias runs through the subject: books about scaling and stories of founders who failed before succeeding are drawn from the survivors, while thousands of similar attempts that ended badly go unrecorded. The fail fast ideology overstates the lesson: the evidence from Gompers and colleagues suggests that prior success predicts later success, while prior failure predicts little, so failure is not in itself an education.

Disruption theory remains contested. Its mechanism is plausible and has strong case support, but tests of its predictive power are mixed, and the label is applied too freely. Blitzscaling has even less systematic evidence; it may suit genuine network markets and harm everything else.

Recovery deserves more attention than it receives. Insolvency regimes that let a business keep trading while its debts are restructured, or that sell assets as going concerns, preserve value for employees, suppliers and the wider economy, as the sale of Northvolt's plants to a buyer planning to restart them illustrates. Whether a venture fails or not, the managerial question is how much value survives the attempt.

Section 07: Decision exercise

A fictional situation. Decide before you read the guidance.

Restructure, sell or wind down at Copperleaf

Copperleaf (fictional) makes software and sensors that cut energy use in commercial buildings. It has 140 staff, £1.1 million of cash and a monthly burn of £600,000, and its last round, at £80 million, was two years ago. Its software business (40% of revenue) is profitable; its hardware business loses money and needs £8 million to finish a new product. Existing investors will not fund it at the old valuation.

  1. A. Down round

    Raise £10 million from a new investor at a £20 million pre-money valuation, with a full ratchet for the new money in any further down round. Existing preferences and anti-dilution clauses would leave the founders with under 10%. The whole company continues.

  2. B. Sell through administration

    Appoint administrators and sell the business and assets to a competitor that will keep about 100 staff and honour customer contracts. Shareholders, including the founders, would probably receive nothing; some creditors would be paid in part.

  3. C. Shrink to the core

    Close the hardware business, cut staff to 60 and run the profitable software business, funded by a £2 million convertible bridge from existing investors. Smaller ambitions, but the company and its software customers survive.

Judge the options on

  • Value preserved for creditors, employees and customers
  • Probability that the business survives
  • Outcomes for founders and investors
  • Obligations to customers who depend on the products
  • Directors' duties as insolvency approaches
Read the guidance

With under two months of cash, the directors' first duty is to avoid worsening creditors' position; as insolvency approaches, their duties shift towards creditors, so every option must be judged partly on what creditors receive.

Option C is the strongest if the software business really stands alone: it preserves most of the value that exists, keeps customers supported and needs only a small bridge. Its cost is the hardware vision and 80 jobs, and it depends on existing investors agreeing quickly.

Option A keeps the whole company alive but hands nearly all of it to new money on terms that make any later difficulty catastrophic for everyone else; it is worth taking only if the hardware product has strong evidence of demand and the £8 million will reach a milestone that resets the valuation.

Option B is the right answer if neither A nor C can be completed before the cash runs out. It sacrifices the shareholders but preserves jobs and customer service through a buyer. The weakest course is to delay the choice while burning the last of the cash.

Section 08: Further reading

  • Larry E. Greiner (1972). Evolution and revolution as organizations grow. Harvard Business Review.

    The classic model of growth phases and the crises between them.

  • James G. March (1991). Exploration and exploitation in organizational learning. Organization Science.

    The foundational statement of the tension between exploration and exploitation.

  • Charles A. O'Reilly III and Michael L. Tushman (2004). The ambidextrous organization. Harvard Business Review 82(4).

    How ambidextrous organizations run new and established businesses side by side.

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

    The theory of disruptive innovation, still the starting point for the debate.

  • Rita Gunther McGrath (1999). Falling forward: real options reasoning and entrepreneurial failure. Academy of Management Review.

    Real options reasoning applied to entrepreneurial failure.

  • Scott A. Shane (2008). The Illusions of Entrepreneurship: The Costly Myths That Entrepreneurs, Investors, and Policy Makers Live By. Yale University Press.

    A sober, evidence-based corrective to popular myths about start-up success and failure.

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.