The Knowledge-Creating Company

The first prototype baked a crust around raw dough

In 1985, a Matsushita Electric team was trying to build a fully automatic home bread maker. Its first prototype could mix, ferment, and heat ingredients, but the result was scarcely bread: the crust overcooked while the center remained raw. Temperature, flour, yeast, motor speed, and even the shape of the dough case interacted. Technical decomposition had produced components without reproducing the skilled motion that joined them into a loaf. ¹

Software developer Ikuko Tanaka proposed leaving the laboratory. She apprenticed herself to the head baker at the Osaka International Hotel, observing, imitating, and repeatedly making bread. The authors' Matsushita case describes a crucial moment: Tanaka noticed that the baker stretched and twisted the dough. “Twisting stretch” was not a mechanical specification, but it gave the engineers an image they could work against. They also visited the bakery, handled dough, altered propeller strength and speed, and eventually added ribs inside the case so the turning dough would catch and stretch. ²

The product still had to survive commercialization. Marketing and manufacturing joined the original developers; the team had to stabilize quality and reduce cost without losing the bread it had learned to make. Managers and team members took prototypes home, where spouses and children supplied further feedback. The authors report that Home Bakery launched in Japan in February 1987 and sold 536,000 units in its first year. That outcome establishes commercial success as Matsushita and the authors recorded it. Tanaka is a named participant, but her experience and the unnamed baker's work reach the account through the authors' case reconstruction rather than an independent observation of the development process. The sales figure does not by itself prove the general theory later drawn from the episode.³

A movement became a phrase, a mechanism, and new practice

The Knowledge-Creating Company asks what kind of organization can make such a passage repeatable. The baker could produce a result that neither a manual nor an X-ray comparison had explained. Tanaka learned through shared bodily practice, then offered a phrase that preserved enough of the experience to orient joint experimentation. Engineers combined that image with mechanical knowledge. The new ribs and control system embodied a provisional answer; building, testing, and using the machine generated further know-how. ²

Ikujiro Nonaka and Hirotaka Takeuchi name four movements in that passage: socialization among practitioners, externalization through language and image, combination with formal knowledge, and internalization through renewed action. The initials—SECI—became more famous than the bread. In their account, these are not four storage bins or a one-way pipeline. Knowledge spirals across individual, team, and company as each representation returns to practice and changes what people can do.⁴

The case also shows why the coalition matters more than the acronym. Tanaka could recognize and name a difference but could not specify the machine alone. The baker supplied skilled practice; engineers translated movement into hardware; manufacturing and marketing changed the prototype's constraints; household testers judged use and taste. The company created overlap rather than handing a finished specification down a relay. This is the setting in which the book's “middle-up-down” managers, cross-functional teams, redundancy, and shared context become intelligible.⁵

The book generalizes from selected Japanese company stories

Nonaka first presented “the knowledge-creating company” to a broad management audience in a 1991 article. His 1994 theory article formalized organizational knowledge creation, and the 1995 book with Takeuchi expanded the argument through Matsushita, Honda, Canon, NEC, and other companies. It was written against an image of the company as a machine that processes objective information. Metaphor, intuition, commitment, prototype, and practice move to the center instead.⁶

That sequence is documented development within the authors' own work. The company cases are illustrative evidence selected and interpreted to develop the theory, not a representative comparison of successful and unsuccessful innovation. The Matsushita chapter uses one company to illustrate the whole process. The episode can be compared with SECI, but the design cannot show that the four modes were necessary, sufficient, or uniquely explanatory. ⁷

The national contrast needs the same care. The authors use Japanese firms to challenge management assumptions they associate with “Western” views of knowledge. That intervention opened an important question, but Japanese companies are not one harmonious knower. Differences in occupation, seniority, employment security, gender, supplier status, and access to managers can shape whose experience becomes a product concept. The case gives names and agency to some participants while leaving the head baker, factory workers, suppliers, and household testers much less visible than the focal development team. ⁸

Conversion can spread capability and relocate control

The bread maker did not contain the baker. Its ribs embodied one mechanically useful interpretation of a kneading motion. The baker's attention, adjustment, and judgment remained richer than the representation. Calling the passage “conversion” risks making tacit and explicit knowledge look like substances that can be poured between containers without loss.⁹

Stephen Gourlay's 2006 critique argues that only some proposed modes are plausible, that the case evidence admits simpler explanations, and that the framework omits knowledge that remains inherently tacit. He also objects that “knowledge” can become whatever managers validate as organizational knowledge. On that reading, the spiral describes not only creation but authorization: someone decides which experience deserves a concept, budget, prototype, and corporate owner.¹⁰

This is an analytical relation to professional autonomy, not a concept the authors apply to Matsushita. Articulation can let many people learn from one person's skill. It can also move credit, bargaining power, and control away from the practitioner whose experience made the representation possible. The Matsushita story shows Tanaka's central role. The published case does not report how the baker was credited or compensated, who owned the resulting know-how, or how factory workers could contest the way the product was made.¹¹

Later research kept both the model and the dispute alive

Nonaka and Georg von Krogh's 2009 response argues that tacit and explicit knowledge can be understood along a continuum and defends conversion while acknowledging the controversy. That is a refinement by a theory's author and collaborator, not independent confirmation.¹²

A 2019 effort to operationalize SECI first reviewed 108 publications and found an influential but fragmented empirical record: 55 were theoretical, 20 qualitative, and 33 quantitative, with heterogeneous measures and settings. The researchers then developed a questionnaire with 372 employees and replicated its factor structure with 466 health workers, finding an eight-dimensional measure with evidence of reliability and construct validity. That is independent empirical support for their self-report operationalization of SECI-related processes. It does not show that tacit knowledge literally converted into explicit knowledge in the Matsushita case.¹³

The bread-machine episode offers a durable sequence to investigate: a breakdown that formal analysis could not solve, apprenticeship that changed perception, an imperfect phrase that coordinated experimentation, a material prototype, and new practice around the artifact. The continuing dispute asks what that sequence actually warrants. Did knowledge convert, or did people create a useful new representation while embodied knowing remained? Who acquired capability, who acquired ownership, and what evidence would distinguish the two accounts?¹⁴

Read beside Organizational Learning II, the book is an editorial comparison rather than a direct extension. Argyris and Schön ask how inquiry changes governing practices; Nonaka and Takeuchi ask how experience becomes collective innovation. Matsushita makes their overlap inspectable: a company can learn to make something new while still leaving open who may define, authorize, and benefit from what the organization now knows. ¹⁵

Read Weber's Economy and Society as an editorial contrast between credentialed bureaucratic expertise and the distributed knowing described here. Read Barnard's Functions of the Executive as cooperation context for the formal and informal relationships that carry knowledge; neither dependency establishes direct influence. The links to learning quality and reliability, innovation, entrepreneurship, and renewal, and organizational intelligence are analytical routes through learning outcomes, renewal, and collective judgment. Benefit for all life asks who gains authority and value when craft becomes corporate capability; it supplies a normative lens, not a result reported by Nonaka and Takeuchi.¹⁶