Dominant Logic in the Car Industry: Why Companies Hold On to Broken Assumptions
The assumptions behind the car industry’s business model have been breaking for more than a decade, yet the German carmakers began to move only when the crisis arrived. This post asks why companies hold on to assumptions that have already broken. Its subject is the fourth element of every business model, the team and its values. To find out, I traced the education and careers of every management board member at nine German carmakers and suppliers since 2012, 195 people in total. The data points to one answer: the dominant logic of a company lives in the people at the top, and those people were selected by the very assumptions that are now breaking. Existing customers reinforce that loop until the day they leave. The post closes with five lessons for any leadership team whose assumptions are breaking right now.
Why the car industry holds on to broken assumptions: a puzzle about dominant logic
Up to August 2026, battery-electric cars took 21.7% of all new car registrations in the EU, up from 15.8% a year earlier. In August alone it was more than one car in four. In Germany, battery-electric registrations rose by 53.1% in eight months (ACEA, 2026). At the same time, the industry is cutting deep. In September 2026, Volkswagen‘s supervisory board approved a plan to reduce the workforce by around 50,000 further positions, end car production at four German plants and halve the model range, citing excess production capacity of 500,000 vehicles in Europe (AP, 2026). Six months earlier, the group had reported its lowest profit since 2016 (AFP, 2026).
None of this came as a surprise. The direction has been visible since 22 June 2012, when Tesla delivered the first Model S, the first premium sedan engineered from the ground up as an electric car (Tesla, 2012). From that day on, a premium electric car was something a customer could buy, not only something an engineer could imagine. That is why my data starts in 2012. In 2017 I called it a wake-up call for incumbents like Daimler. What puzzles me is the time it took. The assumptions broke years ago. The companies began to move only when the numbers forced them to.
Part of the answer is well known, and it is fair to the companies. Clayton Christensen showed that established firms rarely fail because they ignore their customers. They fail because they listen to them. Their best existing customers do not want the new thing at first, so investment flows to where those customers are (Christensen & Bower, 1996). The hybrid is still the most popular powertrain in the EU at 36.6% of registrations, which shows how long existing customers hold on as well.
I have written about this topic twice in recent weeks. The first post described how the assumptions behind a business model become invisible to the people running it. The second showed how Rumelt’s kernel of good strategy turns a broken assumption into a starting point for strategy, and why the fourth element, the team and its values, is where coherence most often fails. This post asks the question that remains after both: why does it take so long, even when the method is known? My answer is that the assumptions do not live in strategy documents. They live in people.
Dominant logic lives in people: the top team as the carrier of the business model
In 1986, C. K. Prahalad and Richard Bettis gave this a name: the dominant logic of a firm, the way its top managers conceptualise the business and decide where resources go (Prahalad & Bettis, 1986). The dominant logic is the business model as the leadership team sees it. It is learned through experience, it works as a filter, and it is reinforced every time the business succeeds. In 2011 I described the same phenomenon from the other side: a business model is a social construct of the dominant group of managers, and it limits the strategic options that group can even see.
Two years earlier than Prahalad and Bettis, Donald Hambrick and Phyllis Mason had argued that a company becomes a reflection of its top managers (Hambrick & Mason, 1984). No executive can take in everything a situation offers. Cognitive base and values narrow what managers look at, what they notice within that field of vision, and how they interpret what they notice. Because nobody can look directly into an executive’s head, Hambrick and Mason proposed observable traces: age, functional background, career experience, education, socioeconomic roots, financial position, and how similar the members of the top team are to one another. On education they are explicit. Someone trained in engineering can be expected to bring a different cognitive base from someone trained in history or law, and the choice of subject says something about a person’s values and cognitive preferences.
I use education in the same spirit. Every field of study comes with its own dominant logic. A mechanical engineer learns that value lies in a physical product that works reliably and improves step by step. A business graduate learns to read a company through its numbers. Both are valuable, and both become hard to unlearn after thirty years of success. Education is only a proxy for a worldview, and careers can change it. Hambrick and Mason say so themselves: people choose their subject young, with incomplete information, and some later transcend that choice. They also report that the few studies available at the time linked the level of education, not the type of degree, to receptiveness to innovation. My data does not test that question. It shows which logics were present at the top table, and with that caveat education is the most consistent trace available for every board member over fourteen years.
Team and values: where the dominant logic sits in the business model
In my business model framework, this is the fourth element, the company spirit: the team and its values, the element that decides how the other three are read. A board that rose through engines and plants defines the value proposition through the product, sees its core capabilities in powertrain and production, and judges every new investment against plants it has already written off. That interdependence makes a business model hard to copy and just as hard to change, which is why unlearning has to happen in all four elements at once.
The filter becomes visible in how German board members spoke about Tesla. In October 2017, a moderator in Passau remarked that Tesla fascinated its customers. Volkswagen’s chief executive Matthias Müller replied that there were “Ankündigungsweltmeister” in the world, world champions of announcements, without naming names. He set Volkswagen’s eleven million cars and annual profit of 13 to 14 billion euros against companies that sold 80,000 cars a year and, if he was correctly informed, destroyed a three-digit million sum every quarter (Heise, 2017; Fortune, 2017). In June 2019, BMW’s development chief Klaus Fröhlich told journalists in Munich: “There are no customer requests for BEVs. None.” European customers, he added, were not prepared to take the risk, because the infrastructure was not there and resale values were unknown (Forbes, 2019).
Both were experienced and successful managers, and both measured the new business with the yardsticks of the old one: volume, profit, the wishes of today’s customers. Not every board member saw it that way. In October 2019, Herbert Diess, by then Volkswagen’s chief executive, pushed back against the view of Tesla as a niche player. “Tesla is not niche,” he said, and called it a competitor Volkswagen took very seriously. (Automotive News, 2019).
What fourteen years of German car boards reveal about dominant logic
I recorded the field of study of every management board member at BMW, Mercedes-Benz, Volkswagen, Porsche and Audi, year by year from 2012 to September 2026, and did the same for the suppliers Bosch, Continental, ZF and Schaeffler. The sources are annual reports, company CVs, press releases and supervisory board CVs. For 194 of the 195 people, the field of study is documented; for nine of them the sources say only that they studied engineering, without naming the discipline.
Three findings stand out for the five carmakers.
The share of mechanical engineers rose during the transformation. Mechanical, automotive, production and aerospace engineers held 11 of 39 board seats (28%) in 2012, 16 of 38 (42%) in 2022, and 14 of 37 (38%) today. When the uncertainty grew, the boards reached for the competence they trusted most.
Three logics have held the boards for fourteen years. Engineering, industrial engineering and business or economics together held 85% of the seats in 2012 and 84% in 2026. In no year did they fall below 74%.
The disciplines of the new business model stayed at the edge. Electrical engineering, computer science and chemistry together never held more than 5 of 38 seats. Today they hold 3 of 37. Since October 2025, none of the five carmakers has an electrical engineer on its board, and nobody in the whole dataset studied battery chemistry or electrochemistry. Where the new disciplines do appear, they sit outside development: a business informatics graduate runs finance at Audi, and the two most recent board members with degrees in electrical engineering and chemistry ran procurement. The development boards are held by engineers whose training lies in mechanics and materials.
What the CEOs did before they led
Education is the starting point of a worldview. Career is what follows. So I traced the career of every chief executive at the nine companies since 2012, 29 people, and sorted each station before the top job into three environments: optimising an existing business, restructuring one without a new business model, or building a new business.
Not one of the fifteen carmaker CEOs had built a new business before taking the top job. Most spent their entire careers inside one group. In the previous post I recalled Andy Grove’s question to Gordon Moore at Intel: what would a new CEO do if the board brought one in from outside? The question works because an outsider carries no history. In the German car industry, the new CEO almost always came from inside. Dieter Zetsche led the turnaround at Chrysler, which is restructuring rather than a new business model. The closest case is Herbert Diess, who ran development at BMW when the electric i3 went into series production. At BMW, every chief executive since 2000 except one had previously run production (Börsen-Zeitung). Hambrick and Mason anticipated what such careers mean. Production belongs to what they called the throughput functions, which work at making the existing transformation process more efficient. A top team that has risen entirely inside one organisation, they argued, meets a radical technological shift with a restricted knowledge base, while the same team serves the company well in stable times. They even asked whether a firm led by executives who rose primarily through operations would “tend to be sluggish in responding to a new product initiative” (p. 194).
The exceptions sit with the suppliers. Volkmar Denner, a physicist, was instrumental in establishing Bosch’s MEMS sensor business (Bosch via Automotive World). Christian Fischer, Bosch’s chief executive since July 2026, led a young RFID company and before that the restructuring of a construction group (Bosch). Mathias Miedreich, ZF’s chief executive since October 2025, built Faurecia’s e-mobility business and ran the materials group Umicore (ZF). And among board members below the chief executive, Michael Bolle co-founded Systemonic, a Dresden chip start-up for software-defined wireless chips that Philips bought in 2003, before he became Bosch’s chief technology officer.
The Bosch board also looks different in its education. Physicists, electrical engineers and, since 2023, a chemist sit at its table. That does not make it free of a dominant logic; it has a different one. A Bosch manager once summed up the company’s self-image to me in a conversation: Bosch makes high-tech products in volumes of millions. That belief has served Bosch extraordinarily well, and it is still a belief about hardware and scale. I described the power of such mental models at Bosch Power Tools as early as 2009, in Diversity and Mental Models.
The people at the top of the carmakers were optimisers, and very good ones. That is no reproach. They rose because they were excellent at the business that paid for everything. It also means that the experience of finding a new business model was almost entirely absent from the rooms where the business model was decided.
Two loops that hold each other: why dominant logic reproduces itself
If the composition had changed slowly over fourteen years, it would merely be inertia. It barely changed at all, and the engineering share even rose. That calls for a mechanism. The economist W. Brian Arthur described how self-reinforcing feedback locks a technology onto a path (Arthur, 1994). Jörg Sydow, Georg Schreyögg and Jochen Koch carried the idea into organisations (Sydow, Schreyögg & Koch, 2009). In their model a company first has room to choose, then self-reinforcing mechanisms narrow that room step by step, until it ends in a lock-in. Two of the mechanisms they name matter most here: learning effects, because every year in the old business makes people better at it, and adaptive expectations, because everyone expects the path to continue and acts accordingly. In innovation research, the end point of such a path has a name: a dominant design, a term coined by Abernathy and Utterback for products. In my dissertation I argued that dominant designs emerge not only among products but among competing business models (Stähler, 2001). A board’s dominant logic is the mental mirror of that design.
Here the path is held by two loops at once.
The first loop runs through the board itself. Boards choose their successors, and careers run through the core functions of the existing business. The BMW pattern of production chiefs becoming chief executives is the clearest case. The loop is often defended from outside the boardroom as well. When Volkswagen looked for a successor to Martin Winterkorn, the head of its works council made clear that he wanted an engineer with a product background, and that a pure business graduate was out of the question (Auto Bild). Each selection of this kind is reasonable. Together they make sure the next generation thinks like the last. Hambrick and Mason saw this circularity in 1984: new chief executives of large firms are mostly promoted from within and often groomed by their predecessors, and executives are frequently chosen precisely because they have the “right” background for the strategy already in place.
The second loop runs through the existing customers. Customers are on a path of their own. Their idea of a good car was formed over decades by the very companies that sell it to them. Their filling stations, workshops, company-car rules and expected resale values all assume the old product. Fröhlich named exactly these reasons in 2019, when he explained why European customers hesitated. As long as they keep buying, every sale confirms the board in its assumptions, and the board invests in making the old product even better. Christensen and Bower found exactly this pattern: established firms led in developing new technologies whenever those technologies served their existing customers, and fell behind when they did not (Christensen & Bower, 1996).
The two loops stabilise each other, and that is why companies with the right data still wait. In the terms of Sydow and his co-authors, this is the step from path formation to lock-in. Then the new loop takes off. It has leaders of its own, people who carry no history in the old business model, and every doubling confirms their assumptions as surely as every sale confirms the incumbents’. As I argued in the powertrain post, a market tips when adoption on the demand side and the learning curve on the supply side accelerate together. At that moment an asymmetry opens up. The customer only has to switch provider. The traditional carmaker carries its path in concrete: engine plants, supplier contracts, the skills of its workforce. Christensen, Kaufman and Shih showed how financial logic deepens the trap, because incumbents compare every new investment with the marginal cost of assets they have already written off, while a new entrant only ever pays the full cost of the new (Christensen, Kaufman & Shih, 2008). The assumptions break over years. The customers leave within months.
Seeing is not enough: unlearning, and the question of who owns the dominant logic
Suppose a board does see that its assumptions are breaking. Two obstacles remain.
The first is unlearning. In 2009 I wrote that unlearning is the hard part of change, much harder than learning something new. Seven years later I went one step further and asked who is in charge of business model innovation. My answer then was that the chief executive has to unlearn the most, because the track record that made him chief executive is the very thing that has to be questioned. I also wrote that business models are the social construct of the dominant group of managers, and that only diversity can loosen its grip. Hambrick and Mason had put the same idea into research propositions: homogeneous top teams decide faster and do well in stable environments, while in turbulent and discontinuous environments heterogeneous teams should perform better. The data from 2012 to 2026 now shows what that looked like in one industry: a dominant group that renewed itself from within, and a top team in which three logics held about four out of five seats every year.
The second obstacle is ownership. A management board is cut along functions: production, development, purchasing, sales, finance, human resources, and in some years a region or a brand. Each member is accountable for making a function better. Nobody is accountable for the business model as a whole. There have been attempts. BMW gave Peter Schwarzenbauer responsibility for customer engagement and digital business innovation in March 2017 (CV); BMW redistributed these tasks to other board members in April 2019 (Autohaus), and Audi gave Geoffrey Bouquot a board seat for innovation and the software-defined vehicle in 2024 (Audi). Both remained exceptions, and neither lasted long in that form.
In my 2026 update of that post I argued that the workforce often understands the shift better than the managers who grew up in the old model. The car industry shows that this needs one qualification. Understanding the shift is one thing. When it came to choosing the person at the top, even a works council chose the familiar path. The same path runs through the whole ecosystem, down to the mechanic in the local garage who has spent a working life learning the combustion engine. The dominant logic is held by more people than the board, and it is defended wherever the old business model created security.
Five Dominant Logic Lessons Every Leadership Team Can Learn
The car industry is the visible case. The pattern is general, and the lessons apply to any company whose assumptions are breaking, whether through energy prices, geopolitics or artificial intelligence.
1. Map the logics in your leadership team. Write down what each member of your top team studied and where they made their career. Then ask which logics are represented and which are missing. If the business model you will need in ten years requires a logic that nobody at the table brings, you already know where the blind spot is.
2. Treat succession as strategy. Every appointment to the top team decides which logic will dominate the next decade. A company that always promotes the head of its strongest function will reproduce the business model that made that function strong. Ask before each appointment which assumption the candidate embodies, and whether that assumption is still true.
3. Listen to the customers you do not have yet. Your best existing customers keep you on your path, and they are right to do so for longer than you would like. Give someone the explicit task of understanding the customers who will buy the new business model, long before they show up in your own sales figures.
4. Plan unlearning like a project. Unlearning does not happen as a by-product of strategy workshops. It needs time, protected budgets and people who are allowed to question what the company is proudest of. The more successful the old business model, the more deliberate this has to be. The last of the six questions for broken assumptions asks which of your actions changes the team and its values. The data in this post shows why that question so often gets the answer none: the team was selected by the model it is now supposed to change.
5. Give the business model an owner at the top. A board organised by functions optimises functions. Someone at the top table needs the mandate, the resources and the time to work on the business model as a whole, and to keep doing so after the first setback.
Most of the people in my dataset did exactly what their organisations asked of them, and did it superbly. That is the uncomfortable part. Excellence was pointed at the wrong target, the pattern I described as the efficiency trap in The Firebugs in the Attic, and the system that selected them made sure it stayed that way. Transformation begins when a leadership team becomes aware of its own tacit assumptions and starts to work on them deliberately.
Understand. Imagine bigger. Act.
Method note
The dataset starts in 2012, the year Tesla began delivering the Model S, and covers every member of the management board (for Bosch: the board of management of Robert Bosch GmbH) of BMW, Mercedes-Benz Group (formerly Daimler), Volkswagen Group, Porsche AG, Audi, Bosch, Continental, ZF Friedrichshafen and Schaeffler from 2012 to 30 September 2026, 195 people in total. Board compositions were reconstructed for each year-end from annual reports, company announcements and CVs. Fields of study come from company CVs, annual reports, press releases and supervisory board CVs. For one person no field of study could be documented; for nine, the sources name an engineering degree without the discipline. The chart counts board seats at each reference date, so a person sitting on two boards (for example Gernot Döllner at Volkswagen and Audi) is counted twice. Education is used as a proxy for a worldview, following Hambrick and Mason; later career experience can change that worldview.
Annotated References
ACEA. (2026). New car registrations: +5.3% in August 2026 year-to-date; battery-electric 21.7% market share. European Automobile Manufacturers’ Association. The official EU registration statistics by powertrain. Source for the battery-electric share of 21.7%, the hybrid share of 36.6% and the 53.1% growth of battery-electric registrations in Germany in the first eight months of 2026. Link
Arthur, W. B. (1994). Increasing Returns and Path Dependence in the Economy. University of Michigan Press. The economics of self-reinforcing feedback: when each adoption makes the next more likely, systems lock in on a path that owes more to history than to inherent superiority. Used here to explain why board composition and customer behaviour reproduce the old business model. Overview
Christensen, C. M., & Bower, J. L. (1996). Customer power, strategic investment, and the failure of leading firms. Strategic Management Journal, 17(3), 197–218. Based on the disk drive industry, the article shows that well-managed incumbents led in developing technologies of every kind, even radical ones, as long as those technologies served their existing customers. Customers shape the allocation of resources. The academic core of the argument that existing customers hold a company on its path. Link
Christensen, C. M., Kaufman, S. P., & Shih, W. C. (2008). Innovation killers: How financial tools destroy your capacity to do new things. Harvard Business Review, January 2008. Shows how incumbents judge new investments against the marginal cost of depreciated assets, while new entrants face only the full cost of the new. Used here for the asymmetry between customers, who only switch providers, and carmakers, whose path is built into their plants. Link
Hambrick, D. C., & Mason, P. A. (1984). Upper echelons: The organization as a reflection of its top managers. Academy of Management Review, 9(2), 193–206. The founding article of upper echelons theory. Organisational outcomes are partly predictable from the background characteristics of the top management team, because cognitive base and values filter what executives see and how they interpret it. Observable traits such as age, functional track, career experience, education and team homogeneity serve as traces of how executives think. The article explicitly treats an engineering degree as a sign of a different cognitive base, expects long inside careers to restrict the search for answers in times of radical technological change, and expects heterogeneous teams to do better in discontinuous environments. It also names the limits used in this post: education is chosen young and can be transcended later, and executives are often selected for the background the existing strategy calls for. JSTOR; DOI
Prahalad, C. K., & Bettis, R. A. (1986). The dominant logic: A new linkage between diversity and performance. Strategic Management Journal, 7(6), 485–501. Introduced the concept of dominant logic: the mental map through which a firm’s top managers conceptualise the business and allocate resources. Formed by past success, it filters what management sees. The central concept of this post. DOI; free PDF, University of Michigan
Stähler, P. (2001). Geschäftsmodelle in der digitalen Ökonomie: Merkmale, Strategien und Auswirkungen. Josef Eul Verlag. The dissertation, that treated the business model as a unit of analysis and design, argued that competition among business models produces dominant designs, carried by network effects on the demand side and economies of scale on the supply side. ResearchGate
Sydow, J., Schreyögg, G., & Koch, J. (2009). Organizational path dependence: Opening the black box. Academy of Management Review, 34(4), 689–709. Transfers path dependence from technologies to organisations. A path emerges in three phases: preformation with open choices, path formation driven by self-reinforcing mechanisms (coordination, complementarity, learning and adaptive expectations), and lock-in. The article also discusses how organisations can break out of a path, and received the journal’s Decade Award in 2019. Used here for the mechanism behind the two loops. Link
