Jawn Lam

Distributed (all-remote) organization

Organizational StructuresContemporary and emerging forms13 min read

Members work across places and time zones, often with no shared office, and coordinate mainly through written, asynchronous standards.

What it is

A distributed organization has no shared site. Its members work across places and time zones and coordinate mainly through written, asynchronous standards such as handbooks and shared repositories and through mutual adjustment over digital tools. Dispersion is the defining layer over any grouping, and authority stays in one employer's hierarchy.

The distributed organization keeps its hierarchy and gives up the shared site

Select a unit or a letter to read it; the dashed band is a layer the form has only at the moderate end of its range.

Reports toAdjust to each other directlyMMutual adjustmentWStandardization of work processes

Over digital toolsTop managementUnits(Moderate endonly)Members, apartTopmanagementUnit byfunctionUnit byproductMember in onetime zoneMember inanother zoneBy function or by productWM
Over digital toolsTop managementUnits (Moderate end only)Members, apartTopFunctionProductOne zoneFar zoneBy function or by productWM
Select a unit or a letter to read it. M, mutual adjustment: People coordinate their own work by informal communication with each other. W, standardization of work processes: The content of the work is specified or programmed in advance, usually by analysts.Top management. Layer: Top management. Reports to no one. Authority stays in one employer's hierarchy. Real authority follows information (Aghion and Tirole 1997, a formal model), and dispersion leaves much of it with people far from the center, our judgment.Unit by function. Layer: Units. Reports to Top management. By function or by product. Dispersion is laid over whatever structure the firm already has; the form sets no grouping basis of its own, and grouping by function is one of the two bases recorded, as judgment. The profile spans flat and moderate, our judgment since nothing in the sources read sets the depth, so this middle line of units is drawn as found only at the moderate end.Unit by product. Layer: Units. Reports to Top management. By function or by product. The other basis recorded, as judgment. Written, asynchronous standards carry much of the coordination; in Choudhury and colleagues' (2020) case of GitLab, a public handbook documents the company's formal structure and processes.Member in one time zone. Layer: Members, apart. Reports to Unit by function. Works apart from the others, often with no shared office. O'Leary and Cummings (2007) describe dispersion as spatial, temporal, and configurational.Member in another zone. Layer: Members, apart. Reports to Unit by product. Works in another place and time zone. Geographic boundaries are crossed routinely, the form's defining move, while the firm's own edge need not open (judgment).W, standardization of work processes. The content of the work is specified or programmed in advance, usually by analysts.M, mutual adjustment. People coordinate their own work by informal communication with each other.
The structure in words
  1. Top management
    • Top management, reports to no one. Authority stays in one employer's hierarchy. Real authority follows information (Aghion and Tirole 1997, a formal model), and dispersion leaves much of it with people far from the center, our judgment.
  2. Units (Moderate end only)
    • Unit by function, reports to Top management, grouped by function or by product. Dispersion is laid over whatever structure the firm already has; the form sets no grouping basis of its own, and grouping by function is one of the two bases recorded, as judgment. The profile spans flat and moderate, our judgment since nothing in the sources read sets the depth, so this middle line of units is drawn as found only at the moderate end.
    • Unit by product, reports to Top management, grouped by function or by product. The other basis recorded, as judgment. Written, asynchronous standards carry much of the coordination; in Choudhury and colleagues' (2020) case of GitLab, a public handbook documents the company's formal structure and processes.
  3. Members, apart
    • Member in one time zone, reports to Unit by function. Works apart from the others, often with no shared office. O'Leary and Cummings (2007) describe dispersion as spatial, temporal, and configurational.
    • Member in another zone, reports to Unit by product. Works in another place and time zone. Geographic boundaries are crossed routinely, the form's defining move, while the firm's own edge need not open (judgment).
  • Member in one time zone and Member in another zone: adjust to each other directly (Over digital tools), coordinated by mutual adjustment.
  • By function or by product: coordinated by standardization of work processes.

Members work apart across places and time zones, inside one employer's hierarchy and over whatever grouping it has. Written standards such as handbooks carry much of the coordination, and members adjust to one another over digital tools.

Source: Hinds and Kiesler 2002; O'Leary and Cummings 2007; Aghion and Tirole 1997; Choudhury and colleagues 2020, one case study with scholarly commentary.

Each profile level is our judgment:

  • Grouping: spans function and product, since the form sets no basis of its own.
  • Specialization: medium. Formalization: spans medium and high, written standards carrying much of the coordination.
  • Centralization: mixed. Telecommuting had small, mainly beneficial effects on perceived autonomy (Gajendran and Harrison 2007, a meta-analysis of 46 studies of 12,883 employees), but autonomy over one's own work is not where decisions sit.
  • Layers and span: spans flat and moderate.
  • Decision rights: selective vertical and horizontal decentralization. Real authority follows information (Aghion and Tirole 1997, a formal model), and dispersion leaves much of it far from the center.
  • Coordinating mechanism: spans mutual adjustment and standardization of work processes.
  • Lateral mechanisms: direct contact and liaison roles, by electronic coordination, which Galbraith (2002) places between voluntary processes and formal groups.
  • Differentiation: moderate, with some integrating roles. Mode of exchange: bureaucracy.
  • Boundary permeability: spans selective and permeable; geographic boundaries are crossed routinely, while the firm's own edge need not open.

Where the distributed (all-remote) organization sits on the eleven dimensions

The filled segments are the form's levels; select a row for its reason and sources.

Select a dimension to read why the form sits there, and on what basis.Basis of grouping: Function or process and product or purpose. Basis: Judgment. The form sets no grouping basis of its own and lays its defining layer over whatever structure it finds; the two bases of a single-line hierarchy are recorded as judgment so a reader's profile can still match. Sources: Hinds and Kiesler 2002; O'Leary and Cummings 2007.Specialization: Medium. Basis: Judgment. Dispersion does not set the band; judgment. Sources: Hinds and Kiesler 2002.Standardization and formalization: Medium to high. Basis: Judgment. Written, asynchronous standards such as documentation and handbooks carry much of the coordination; judgment for the band. Sources: Hinds and Kiesler 2002; Choudhury and colleagues 2021.Centralization: Mixed. Basis: Judgment. Dispersion raised perceived autonomy in a meta-analysis of telecommuting, but autonomy over one's own work is not the location of decisions; judgment. Sources: Gajendran and Harrison 2007; O'Leary and Cummings 2007.Layers and span of control: Flat to moderate. Basis: Judgment. Nothing in the sources read sets the depth; judgment. Sources: Hinds and Kiesler 2002.Decision rights and the locus of authority: Selective vertical and horizontal decentralization. Basis: Judgment. Real authority follows information, and dispersion leaves much of it with people far from the center; judgment for the type. Sources: Aghion and Tirole 1997; O'Leary and Cummings 2007.Prime coordinating mechanism: Mutual adjustment and standardization of work processes. Basis: Judgment. Work done apart leans on written standardization and on mutual adjustment through digital tools; no source read sets the prime mechanism, so this is judgment. Sources: Hinds and Kiesler 2002.Lateral mechanisms: Direct contact and liaison roles. Basis: Judgment. Electronic coordination through intranets and similar tools, which Galbraith places between voluntary processes and formal groups on his ladder of lateral processes; judgment for the form. Sources: Galbraith 2002a; Mesmer-Magnus and colleagues 2011.Differentiation against integration: Moderate differentiation, some integrating roles. Basis: Judgment. Dispersed members differ in place and time and are joined by written standards; judgment. Sources: O'Leary and Cummings 2007; Lawrence and Lorsch 1967a.Mode of exchange and control: Bureaucracy. Basis: Judgment. One employer's hierarchy, worked from many places; judgment. Sources: Choudhury and colleagues 2021; Ouchi 1980.Boundary permeability: Selective to permeable. Basis: Judgment. Geographic boundaries are crossed routinely, the form's defining move, while the firm's edge need not open; judgment for the span. Sources: O'Leary and Cummings 2007; Ashkenas and colleagues 1995.

All eleven levels are our judgment, and no row is contested. The form's defining move shows in two spans: formalization medium to high, as written standards carry much of the coordination, and boundaries selective to permeable, as geographic ones are crossed routinely.

Source: Hinds and Kiesler 2002; O'Leary and Cummings 2007; Choudhury and colleagues 2021, a natural experiment in one agency; Gajendran and Harrison 2007, a meta-analysis of 46 studies of 12,883 employees; Aghion and Tirole 1997; Galbraith 2002a; Mesmer-Magnus and colleagues 2011, a meta-analysis of 94 studies; Lawrence and Lorsch 1967a, a multi-firm comparative field study; Ouchi 1980, a conceptual article; Ashkenas and colleagues 1995.

Where it came from

Research on work at a distance named the form; no single founder did. Pamela Hinds and Sara Kiesler edited Distributed Work (2002). Michael O'Leary and Jonathon Cummings (2007), in a research essay, set out dispersion's three dimensions: spatial, temporal, and configurational. Prithwiraj Choudhury, Cirrus Foroughi, and Barbara Larson (2021) separated geographic from temporal flexibility. At issue is how an organization coordinates when its members are dispersed along all three.

The schools of organization design and the forms' origins, 1900 to now

Each row is a school's era; each square under them is a form at its first year. Select a row or a square to read it; on a form's page, its own square and span are marked, with the schools that treat it.

1900192019401960198020002020Bureaucracy and its criticsClassical administrationHuman relations and System 4Transaction cost economicsCarnegie schoolSocio-technical systemsLabor-managed firmModularity and mirroringStrategy and structureContingency theoryFrench bureaucratic schoolJapanese-firm literatureScandinavian and DutchConfiguration theoryInformation processingLabor processStrategic choiceGerman codeterminationOrganizational economicsInstitutional theoryPopulation ecologyResource dependenceBusiness groupsScandinavian institutionalSocietal effectNetwork perspectiveComputational designNew-forms debateAlgorithmic managementForms by first yearDistributed (all-remote) organization, 2002 to 2021Now

The forms by first year

1900192019401960198020002020Distributed (all-remote) organization, 2002 to 2021Now

Schools that treat this form: Modularity and the mirroring hypothesis; the new-forms debate.

Select a school's row or a form's square to read it.Select a form's square to read it; open a school below.Bureaucracy and its critics1900s to 1950s, revived from the 1990sWhat it claimsWeber held that rational-legal authority exercised through an office hierarchy, written rules, trained officials, and promotion by qualification is technically superior in precision, speed, and reliability. Critics studied its unintended consequences, Michels's iron law sharpest.What it has shownMerton (1940) argued, with no test, that rules can turn from means into ends; case studies followed. The Aston group concluded that the type's dimensions vary independently, and Wagner's meta-analysis (2021) found that many of their interrelationships have weakened.Who disputes itHuman relations and System 4 sets the costs of impersonal rules to motivation and cooperation against the technical superiority Weber stresses.Bureaucracy and its critics in The Schools of Organization DesignClassical administration1910s to 1940s, and in German to the 1970sWhat it claimsThere are general principles of administration, among them division of work, unity of command, a scalar chain of authority, and a limited span of control, and a structure built to them will be efficient in most settings. Taylor added that planners should set work methods.What it has shownThe school offered reasoning and experience, not tests. Woodward (1965) found that the spans and levels of the more successful firms in her sample differed by production system, and Simon (1946) showed that the principles come in pairs that point in opposite directions.Who disputes itStructural contingency theory found that the structure of successful firms varies with technology and environment, where classical writers treat their principles as general.Classical administration in The Schools of Organization DesignHuman relations and System 41920s to 1960sWhat it claimsOutput depends on the social relations of work, the norms of informal groups, and the style of supervision as much as on formal design. Likert's System 4, overlapping work groups joined by linking-pin managers under participative management, outperforms authoritative systems.What it has shownThe founding evidence is the Hawthorne studies, which later reanalyses dispute (Carey 1967; Levitt and List 2011). Likert's support came largely from his own institute's surveys. Wagner (1994) found participation's effects on performance and satisfaction significant but small.Who disputes itLabor process and critical theory reads human relations as a technique of managerial control rather than a change in the distribution of power.Human relations and System 4 in The Schools of Organization DesignTransaction cost economics and the M-form hypothesis1937, then the 1970s to the presentWhat it claimsFirms exist, and are organized as they are, to economize on transaction costs. Hierarchy replaces the market when assets are specific, uncertainty is high, and exchange is frequent, and the multidivisional form should outperform the functional form in large, diversified firms.What it has shownArmour and Teece (1978) found, in United States petroleum firms from 1955 to 1973, a positive relationship between the multidivisional form and profitability while the form was spreading, but not once an equilibrium in organizational form was reached.Who disputes itThe network perspective on organizations: Granovetter argued that the theory undersocializes economic action, which is embedded in networks of personal relations that also produce trust and order.Transaction cost economics and the M-form hypothesis in The Schools of Organization DesignThe Carnegie school and the attention-based view1940s to the presentWhat it claimsPeople in organizations are boundedly rational, so structure matters because it sets the premises of their decisions and channels their attention: who sees which problems, which goals they pursue, and which routines they follow. Ocasio makes structure distribute attention.What it has shownMarch and Simon (1958) drew propositions from the existing literature, and Cyert and March (1963) added process models and simulations. Two studies show hierarchy shaping what is sent up as well as what is chosen (Reitzig and Maciejovsky 2015; Keum and See 2017).Who disputes itOrganizational economics: decision rights, authority, and complementarities models a principal and agents with defined objectives; the Carnegie school, firms as coalitions with unresolved goal conflict.The Carnegie school and the attention-based view in The Schools of Organization DesignSocio-technical systems1950s to 1980sWhat it claimsAn organization is a social system and a technical system that must be designed together. The same technology admits more than one way of organizing work, and groups that own a whole task and control its variances at the source perform better and are more satisfied.What it has shownTrist and Bamforth (1951) found that longwall mechanization broke up self-regulating work groups. A review of 134 experiments found them extremely successful overall, and Stewart (2006) linked team autonomy to performance; studies of socialist brigades and Volkswagen qualify it.Who disputes itLabor process and critical theory: Kelly argued that labor intensification, not joint optimization, accounts for what socio-technical redesign achieved, and that the part of pay incentives was seriously underestimated.Socio-technical systems in The Schools of Organization DesignThe economics of the labor-managed firm and of enterprise ownership1958 to the presentWhat it claimsA firm run by its workers behaves differently from one run for investors: Ward modeled the Yugoslav-style firm as maximizing income per worker. Hansmann argued that an ownership form's success depends on the balance between the costs of market contracting and of ownership.What it has shownBurdín (2014) found a 29 percent lower hazard of dissolution for Uruguayan worker-managed firms than for conventional ones, and Fakhfakh, Pérotin, and Gago (2012) found French labor-managed firms as productive or more so in every industry studied.Who disputes itOrganizational economics (Jensen and Meckling) contrasts the labor-managed firm, in which tradable residual claims are legally prohibited, with cooperatives and partnerships that arise out of voluntary contracting.The economics of the labor-managed firm and of enterprise ownership in The Schools of Organization DesignModularity and the mirroring hypothesis1960s to the presentWhat it claimsComplex systems are built of nearly decomposable parts, and an organization's structure tends to mirror the architecture of what it makes: modular products allow loosely coupled organizations, and tightly coupled products need tightly coupled teams. Conway's law runs in reverse.What it has shownColfer and Baldwin (2016) review the empirical studies and report both support and principled exceptions. Direct tests point both ways: MacCormack, Baldwin, and Rusnak (2012) found open source products more modular than their commercial pairs; dynamic components break the mirror.Who disputes itNo school in the record disputes it head on; it answers information processing, which meets the coordination load inside the organization; mirroring writers locate it in the product's architecture.Modularity and the mirroring hypothesis in The Schools of Organization DesignStrategy and structure1960s to 1980sWhat it claimsStructure follows strategy: firms that diversify by product or geography outgrow the centralized functional structure and, after a lag and usually a crisis, adopt divisions. Chandler drew this from the histories of DuPont, General Motors, Standard Oil of New Jersey, and Sears.What it has shownChandler (1962) is comparative business history of four firms set against a wider survey. Rumelt (1974) related diversification and structure to performance, and Egelhoff (1988), with 34 multinationals, found some of Stopford and Wells's relationships supported and others not.Who disputes itInstitutional theory: in Fligstein's model divisions spread with a product-related or unrelated strategy, presidents from sales or finance, and other firms in the industry changing; strategy was one of three.Strategy and structure in The Schools of Organization DesignStructural contingency theory1960s to the presentWhat it claimsThere is no one best structure. Performance depends on fit between structure and contingencies such as environment, technology, size, and strategy, on an open-system view of the organization. Drazin and Van de Ven separate fit as selection, interaction, and systems.What it has shownBurns and Stalker, Woodward, and the Aston group related structure to context; Drazin and Van de Ven (1985) supported selection and systems fit only. Pennings (1975) found that fit did not explain effectiveness; Schoonhoven (1981) found traditional hypotheses unsupported.Who disputes itStrategic choice: Child argued that contingency theory explains structure by the constraints of context and leaves out the political process by which power-holders choose strategic action, structural forms included.Structural contingency theory in The Schools of Organization DesignThe French bureaucratic school (strategic analysis)1960s to 1980sWhat it claimsA bureaucracy is a system of power relations organized around zones of uncertainty. Impersonal rules and centralization protect members from depending on one another, those who control uncertainties hold power, and, unable to correct its errors, it changes only through crisis.What it has shownLe phénomène bureaucratique (1963; English 1964) rests on two case studies, a clerical agency and a state industrial monopoly. L'acteur et le système (1977; English 1980) is theoretical.Who disputes itSocietal effect and comparative business systems: D'Iribarne argues that national logics of conduct, such as a French logic of honor, explain differences that Crozier attributed to power games.The French bureaucratic school (strategic analysis) in The Schools of Organization DesignThe Japanese-firm literature1960s to the presentWhat it claimsThe large Japanese firm coordinates operating decisions through horizontal information sharing among semi-autonomous shops and uses its ranking hierarchy mainly as an incentive scheme (Aoki). It creates knowledge through middle-up-down management (Nonaka and Takeuchi).What it has shownAoki (1986) is a formal model. Lincoln, Hanada, and McBride (1986) compared 51 Japanese and 55 United States manufacturing plants, and Nonaka and Takeuchi (1995) rest on case studies, chiefly of Japanese firms. Fruin (1994) found the enterprise system defined at three levels.Who disputes itLabor process and critical theory reads shop-floor delegation as control: Sewell and Wilkinson argue that just-in-time and total quality regimes delegate tactical responsibility while centralizing strategic control.The Japanese-firm literature in The Schools of Organization DesignThe Scandinavian and Dutch socio-technical schools1960s to 2000sWhat it claimsThe Scandinavian program sought democracy at work through field experiments with semi-autonomous groups, then through democratic dialogue. The Dutch school's design theory turns complex organizations with simple jobs into simple organizations with complex jobs.What it has shownEmery and Thorsrud reported field experiments in Norwegian industry; Gustavsen (1992) records that they spread less than hoped, his reason for turning to dialogue. Berggren studied work organization in the Swedish auto industry. The Dutch work rests on design cases and argument.Who disputes itThe Japanese-firm literature: defenders of the Toyota system hold that standardized work with employee participation learns faster; Berggren reports Swedish long-cycle plants matching a line plant on productivity.The Scandinavian and Dutch socio-technical schools in The Schools of Organization DesignConfiguration theory1970s to the presentWhat it claimsStructures cluster into a small number of internally coherent configurations whose elements reinforce one another, so organizations move between a few types in large jumps, not small steps. Mintzberg's five configurations each rest on one coordinating mechanism.What it has shownDoty, Glick, and Huber (1993) found that fit to Miles and Snow's ideal types predicted effectiveness while fit to Mintzberg's did not. A meta-analysis of 40 tests (Ketchen and colleagues 1997) found that configuration partly explains performance, more so with broad definitions.Who disputes itStructural contingency theory: Donaldson counters that most organizations sit at intermediate points on a continuum of fits, not in a few configurations, and that studies over time find mostly incremental change.Configuration theory in The Schools of Organization DesignInformation processing1970s to the presentWhat it claimsThe more uncertain the task, the more information must be processed during its execution. An organization either reduces that need, through slack resources or self-contained tasks, or raises its capacity, through vertical information systems or lateral relations up to the matrix.What it has shownGalbraith (1973, 1974) built the framework from cases and from Simon's theory of bounded rationality, and Tushman and Nadler (1978) set it out as an integrating concept for design. Daft and Lengel (1986) argued that structure must reduce equivocality as well as supply data.Who disputes itThe Carnegie school and the attention-based view: Ocasio holds that firm behavior results from how the firm channels and distributes attention, so structure governs where attention goes, not only how much is processed.Information processing in The Schools of Organization DesignLabor process and critical theory1970s to the presentWhat it claimsStructure is above all a system of control over labor. Braverman argued that management deskills work by separating conception from execution; Edwards traced a sequence from simple control by the boss to technical control through machines to bureaucratic control through rules.What it has shownHistorical and theoretical (Braverman 1974; Edwards 1979). Delbridge (1995), observing a Japanese-owned plant in England, found that just-in-time and total quality regimes combined the control of labor with management's economic goals more completely than piece rates had.Who disputes itBureaucracy and its critics: Adler argues that labor process theory ignores the progressive socialization of the labor process, which drives long-term skill upgrading, and disputes deskilling as the trend.Labor process and critical theory in The Schools of Organization DesignStrategic choice1970s to the presentWhat it claimsStructure is the outcome of choices by those with power in the organization, who decide on strategic action, including the structural form, the manipulation of environmental features, and the performance standards to be met. Context constrains but does not determine structure.What it has shownLargely conceptual: Child (1972) built the argument from gaps in contingency findings, and the empirical case rests mostly on the modest share of structural variance contingencies explain. A 2026 review of 42 studies finds upper echelons theory their main theoretical anchor.Who disputes itStructural contingency theory: Donaldson replies that organizations mostly adopt the structures their contingencies dictate, to avoid the loss of performance misfit brings, so choice over structure is at most limited.Strategic choice in The Schools of Organization DesignThe German codetermination literature1970s to the presentWhat it claimsWorks councils and employee board seats form a structural layer through which coordination can become negotiation between management and employees. Defenders argue that it supports long-term employment, training, and cooperation; critics, that it weakens shareholder control.What it has shownJäger, Schoefer, and Heining (2021) found board-level codetermination left wages unchanged and, if anything, raised capital formation. Jirjahn (2011) reports studies pointing to positive works council effects; a meta-analysis found codetermination laws tied to lower productivity.Who disputes itOrganizational economics (Gorton and Schmid) found that firms with equal employee and shareholder representation did worse for their shareholders than firms with one-third employee representation.The German codetermination literature in The Schools of Organization DesignOrganizational economics: decision rights, authority, and complementarities1976 to the presentWhat it claimsDesign is the joint choice of who holds which decision rights, how performance is measured, and how it is rewarded. Rights should be placed where the specific knowledge lies, formal authority differs from real authority, and practices are complements that pay off in clusters.What it has shownIchniowski, Shaw, and Prennushi (1997) found, on 36 steel finishing lines, that lines using a set of innovative work practices achieved substantially higher productivity, consistent with complementarities. Bloom and colleagues found more decentralization where trust was higher.Who disputes itGhoshal argued that management research has shaped managers' worldview with ideologically inspired amoral theories, built on a scientific model that excludes human intentionality and so denies moral considerations.Organizational economics: decision rights, authority, and complementarities in The Schools of Organization DesignInstitutional theory1977 to the presentWhat it claimsFormal structures are often adopted because they are legitimate in the organization's field, as rationalized myths, and then decoupled from the work done. Organizations in a field grow alike through coercive, mimetic, and normative isomorphism, whether or not it is efficient.What it has shownTolbert and Zucker (1983) found that city characteristics predicted early but not later adoption of civil service reform, which they read as institutionalization. Palmer, Jennings, and Zhou (1993) found support for institutional and economic accounts of late M-form adoption.Who disputes itStructural contingency theory: Kraatz and Zajac found little support for neoinstitutional predictions in 631 liberal arts colleges, which grew less alike; the change they studied was curricular, not structural.Institutional theory in The Schools of Organization DesignPopulation ecology1977 to the presentWhat it claimsOrganizations face strong structural inertia, because selection favors the reliability and accountability of stable structures. Change in a population's forms comes mainly from founding and failure, not adaptation, and changing core features resets the liability of newness.What it has shownAmburgey, Kelly, and Barnett (1993) found that the effects of change on the hazard of failure vary by type of change and over time, and that rates of change depend on how often and how recently an organization has changed. Carroll and Hannan (2000) set out the program's methods.Who disputes itFor structural contingency theory, Donaldson argues that ecology understates managers' capacity to adapt structure.Population ecology in The Schools of Organization DesignResource dependence1978 to the presentWhat it claimsOrganizations depend on outside parties for critical resources, and structure follows the management of that dependence: units that cope with critical dependencies gain power, and boundaries are redrawn through cooptation, joint ventures, and mergers to absorb constraint.What it has shownPfeffer and Salancik (1978) assembled studies of boards, mergers, and joint ventures. A meta-analysis of 157 tests (Drees and Heugens 2013) corroborated the main prediction: organizations meet resource dependencies with interorganizational arrangements that raise their autonomy.Who disputes itTransaction cost economics and the M-form hypothesis explains boundaries by efficiency; resource dependence, by power and constraint.Resource dependence in The Schools of Organization DesignThe business-groups literature1978 to the presentWhat it claimsGroups of legally independent firms bound by ownership and social ties are a durable form of their own, neither one firm nor a market. Where institutions function poorly, groups can benefit member firms and can destroy value; they are sometimes paragons, sometimes parasites.What it has shownKhanna and Palepu (2000) found that affiliates of the most diversified Indian groups outperformed unaffiliated firms. A meta-analysis of 141 studies (Carney and colleagues 2011) found that affiliation lowers performance in general, less so where institutions are underdeveloped.Who disputes itOrganizational economics: agency writers read pyramids and cross-holdings as devices that separate control from ownership; Almeida and Wolfenzon offer a financing rationale for pyramids instead.The business-groups literature in The Schools of Organization DesignScandinavian institutionalism1980s to the presentWhat it claimsIdeas of organizing do not diffuse unchanged; they travel and are translated as they are adopted, so a form taken up in one place is not the same form elsewhere. Organizations can also be built of organizations (meta-organizations), which behave differently.What it has shownA discussion paper by Czarniawska and Joerges (1995) contrasts the diffusion of ideas with their translation, drawing on examples such as Swedish municipal reform; the work on meta-organizations (Ahrne and Brunsson 2008) is conceptual.Who disputes itInstitutional theory can read the same processes, seen from a translocal view, as institutional isomorphism, though translation holds that each adopter modifies an idea, so an institution re-created elsewhere is new.Scandinavian institutionalism in The Schools of Organization DesignSocietal effect and comparative business systems1980s to the presentWhat it claimsFirms matched on size, technology, and industry differ systematically by country, because how a society educates, trains, recruits, and promotes its workforce, and its industrial relations, shape how work is divided and supervised, so structure is partly a national product.What it has shownMaurice, Sorge, and Warner (1980) compared closely matched manufacturing units in France, West Germany, and Great Britain and found more pervasive differences in their organization than usually recognized. Whitley (1999) set out a comparative account of business systems.Who disputes itStructural contingency theory stresses culture-free elements of structure; matched comparisons find nationally different shapes of organization.Societal effect and comparative business systems in The Schools of Organization DesignThe network perspective on organizations1985 to the presentWhat it claimsOrganizations and their members are nodes joined by ties, and the pattern of ties shapes what they can do. Economic action is embedded in networks of personal relations (Granovetter 1985), and network forms are a distinct form, not a hybrid of market and hierarchy (Powell 1990).What it has shownPodolny and Page (1998) review the research and find that network forms foster learning, confer status and legitimacy, bring economic benefits, help manage resource dependencies, and give employees considerable autonomy, and they ask for attention to networks' dysfunctions.Who disputes itTransaction cost economics: Williamson treats the hybrid as a third generic form between market and hierarchy; for Powell, network forms are a distinct form, not a mix of the two.The network perspective on organizations in The Schools of Organization DesignComputational organization design and search1990s to the presentWhat it claimsDesign choices interact, so design space is a rugged landscape with many local peaks, and structure affects how well an organization searches it. Decentralizing interdependent decisions can trap units on poor peaks; temporary decentralization, then reintegration, can help.What it has shownMostly simulation, with one field test: Csaszar (2012), with more than 150,000 stock-picking decisions by 609 mutual funds, found that decision-making structure affects the number of initiatives pursued and the omission and commission errors made.Who disputes itNo school in the record disputes it head on; it disputes structural contingency theory, whose bivariate fit, its simulations show, cannot capture how the best design depends on interactions among choices.Computational organization design and search in The Schools of Organization DesignThe new-forms debate1990s to the presentWhat it claimsOrganizations must solve four universal problems (dividing tasks, allocating them, providing rewards, and providing information), and a form is new when it solves them in new ways. Self-managing organizations radically decentralize authority; skeptics hold hierarchy reappears.What it has shownFoss (2003) traced the rise and partial reversal of Oticon's spaghetti organization to the problem of selective intervention. A meta-analysis of 54 studies of 13,914 teams (Greer and colleagues 2018) found that hierarchy within teams on net slightly lowers team performance.Who disputes itBureaucracy and its critics: removing formal structure leaves informal elites whose power is unaccountable (Freeman), and democratic and network organizations are much less hierarchy-free than their labels suggest.The new-forms debate in The Schools of Organization DesignThe algorithmic-management literature2015 to the presentWhat it claimsSoftware does much of the directing, evaluating, and disciplining that managers and rules once did. Kellogg, Valentine, and Christin (2020) call algorithmic control a new contested terrain; Puranam (2021) treats dividing decisions between people and machines as a design problem.What it has shownMostly qualitative field studies of gig work. Wood and colleagues (2019), with 107 interviews in six countries and 679 survey respondents, found that algorithmic management tends to give remote gig workers flexibility and autonomy, but also low pay, isolation, and overwork.Who disputes itLabor process and critical theory reads control at work as contested terrain; Kellogg, Valentine, and Christin take up that view but argue that algorithmic control is distinct from earlier forms.The algorithmic-management literature in The Schools of Organization DesignFunctional structureOrigin: 1911 to 1937.Family: Classical hierarchies.Line and staffOrigin: 1916 to 1943.Family: Classical hierarchies.Multidivisional (M-form) by productOrigin: 1920s practice; named and analyzed 1962 to 1975.Family: Divisionalized forms.Machine bureaucracyOrigin: 1922 (Weber); 1979 (Mintzberg).Family: Classical hierarchies.Divisional by customer or market segmentOrigin: 1937; 2005.Family: Divisionalized forms.Matrix organizationOrigin: 1950s to 1960s practice; analyzed 1971 to 1977.Family: Lateral and multidimensional forms.Co-determined organizationOrigin: 1951 to 1976 (the laws); 1984 to 2022 (the main studies).Family: Forms from other traditions.Socio-technical autonomous work groupsOrigin: 1951 to 1965.Family: Team-based and self-managing forms.Stratified hierarchy (requisite organization)Origin: 1951 to 1989.Family: Classical hierarchies.Cooperative organizationOrigin: 1958 to 1999.Family: Forms from other traditions.Linking-pin organization (System 4)Origin: 1961 to 1967.Family: Team-based and self-managing forms.Divisional by regionOrigin: 1962 to 1972.Family: Divisionalized forms.Adhocracy (innovative organization)Origin: 1968 to 1979.Family: Professional, project, and knowledge forms.J-form (the Japanese firm)Origin: 1968 to 1990.Family: Forms from other traditions.Professional bureaucracyOrigin: 1968 to 1979.Family: Professional, project, and knowledge forms.Holding company (H-form)Origin: 1972 to 1975; 1987.Family: Divisionalized forms.Ambidextrous organizationOrigin: 1976 to 2004.Family: Professional, project, and knowledge forms.Simple structureOrigin: 1979.Family: Classical hierarchies.Hybrid structureOrigin: 1980 to the present.Family: Lateral and multidimensional forms.Missionary (ideological) organizationOrigin: 1980 to 1989.Family: Ideological forms.Network, modular, and virtual organizationOrigin: 1986 to 1990.Family: Network and market-like forms.Transnational and multidimensional organizationOrigin: 1987 to 1989.Family: Lateral and multidimensional forms.Business group (keiretsu, chaebol, family house)Origin: 1988 to 2007.Family: Forms from other traditions.Shamrock organizationOrigin: 1988 to 1989.Family: Network and market-like forms.Federal organizationOrigin: 1989 to 1992.Family: Network and market-like forms.Self-managing organization (less-hierarchical forms)Origin: 1989 to 2017.Family: Team-based and self-managing forms.Boundaryless organizationOrigin: 1990 to 1995.Family: Network and market-like forms.Front-back organizationOrigin: 1990s to 2007.Family: Lateral and multidimensional forms.Professional partnership (P2 form)Origin: 1990.Family: Professional, project, and knowledge forms.The corporate center and shared servicesOrigin: 1991 to 2000.Family: Divisionalized forms.Internal market organizationOrigin: 1993 to 1994.Family: Network and market-like forms.Process (horizontal) organizationOrigin: 1993 to 1999.Family: Lateral and multidimensional forms.Cellular organizationOrigin: 1994 to 1997.Family: Network and market-like forms.Hypertext organizationOrigin: 1994 to 1995.Family: Forms from other traditions.Team-based organizationOrigin: 1995.Family: Team-based and self-managing forms.Project-based organizationOrigin: 2000 to 2006.Family: Professional, project, and knowledge forms.Agile at scaleOrigin: 2001 to 2016.Family: Contemporary and emerging forms.Distributed (all-remote) organizationOrigin: 2002 to 2021.Family: Contemporary and emerging forms.Open-source and community-based organizationOrigin: 2002 to 2007.Family: Contemporary and emerging forms.Algorithmically managed organizationOrigin: 2015 to 2020.Family: Contemporary and emerging forms.Team of teamsOrigin: 2015.Family: Team-based and self-managing forms.

The schools by first year

  1. Bureaucracy and its critics1900s to 1950s, revived from the 1990sWhat it claimsWeber held that rational-legal authority exercised through an office hierarchy, written rules, trained officials, and promotion by qualification is technically superior in precision, speed, and reliability. Critics studied its unintended consequences, Michels's iron law sharpest.What it has shownMerton (1940) argued, with no test, that rules can turn from means into ends; case studies followed. The Aston group concluded that the type's dimensions vary independently, and Wagner's meta-analysis (2021) found that many of their interrelationships have weakened.Who disputes itHuman relations and System 4 sets the costs of impersonal rules to motivation and cooperation against the technical superiority Weber stresses.
  2. Classical administration1910s to 1940s, and in German to the 1970sWhat it claimsThere are general principles of administration, among them division of work, unity of command, a scalar chain of authority, and a limited span of control, and a structure built to them will be efficient in most settings. Taylor added that planners should set work methods.What it has shownThe school offered reasoning and experience, not tests. Woodward (1965) found that the spans and levels of the more successful firms in her sample differed by production system, and Simon (1946) showed that the principles come in pairs that point in opposite directions.Who disputes itStructural contingency theory found that the structure of successful firms varies with technology and environment, where classical writers treat their principles as general.
  3. Human relations and System 41920s to 1960sWhat it claimsOutput depends on the social relations of work, the norms of informal groups, and the style of supervision as much as on formal design. Likert's System 4, overlapping work groups joined by linking-pin managers under participative management, outperforms authoritative systems.What it has shownThe founding evidence is the Hawthorne studies, which later reanalyses dispute (Carey 1967; Levitt and List 2011). Likert's support came largely from his own institute's surveys. Wagner (1994) found participation's effects on performance and satisfaction significant but small.Who disputes itLabor process and critical theory reads human relations as a technique of managerial control rather than a change in the distribution of power.
  4. Transaction cost economics and the M-form hypothesis1937, then the 1970s to the presentWhat it claimsFirms exist, and are organized as they are, to economize on transaction costs. Hierarchy replaces the market when assets are specific, uncertainty is high, and exchange is frequent, and the multidivisional form should outperform the functional form in large, diversified firms.What it has shownArmour and Teece (1978) found, in United States petroleum firms from 1955 to 1973, a positive relationship between the multidivisional form and profitability while the form was spreading, but not once an equilibrium in organizational form was reached.Who disputes itThe network perspective on organizations: Granovetter argued that the theory undersocializes economic action, which is embedded in networks of personal relations that also produce trust and order.
  5. The Carnegie school and the attention-based view1940s to the presentWhat it claimsPeople in organizations are boundedly rational, so structure matters because it sets the premises of their decisions and channels their attention: who sees which problems, which goals they pursue, and which routines they follow. Ocasio makes structure distribute attention.What it has shownMarch and Simon (1958) drew propositions from the existing literature, and Cyert and March (1963) added process models and simulations. Two studies show hierarchy shaping what is sent up as well as what is chosen (Reitzig and Maciejovsky 2015; Keum and See 2017).Who disputes itOrganizational economics: decision rights, authority, and complementarities models a principal and agents with defined objectives; the Carnegie school, firms as coalitions with unresolved goal conflict.
  6. Socio-technical systems1950s to 1980sWhat it claimsAn organization is a social system and a technical system that must be designed together. The same technology admits more than one way of organizing work, and groups that own a whole task and control its variances at the source perform better and are more satisfied.What it has shownTrist and Bamforth (1951) found that longwall mechanization broke up self-regulating work groups. A review of 134 experiments found them extremely successful overall, and Stewart (2006) linked team autonomy to performance; studies of socialist brigades and Volkswagen qualify it.Who disputes itLabor process and critical theory: Kelly argued that labor intensification, not joint optimization, accounts for what socio-technical redesign achieved, and that the part of pay incentives was seriously underestimated.
  7. The economics of the labor-managed firm and of enterprise ownership1958 to the presentWhat it claimsA firm run by its workers behaves differently from one run for investors: Ward modeled the Yugoslav-style firm as maximizing income per worker. Hansmann argued that an ownership form's success depends on the balance between the costs of market contracting and of ownership.What it has shownBurdín (2014) found a 29 percent lower hazard of dissolution for Uruguayan worker-managed firms than for conventional ones, and Fakhfakh, Pérotin, and Gago (2012) found French labor-managed firms as productive or more so in every industry studied.Who disputes itOrganizational economics (Jensen and Meckling) contrasts the labor-managed firm, in which tradable residual claims are legally prohibited, with cooperatives and partnerships that arise out of voluntary contracting.
  8. Modularity and the mirroring hypothesis1960s to the presentWhat it claimsComplex systems are built of nearly decomposable parts, and an organization's structure tends to mirror the architecture of what it makes: modular products allow loosely coupled organizations, and tightly coupled products need tightly coupled teams. Conway's law runs in reverse.What it has shownColfer and Baldwin (2016) review the empirical studies and report both support and principled exceptions. Direct tests point both ways: MacCormack, Baldwin, and Rusnak (2012) found open source products more modular than their commercial pairs; dynamic components break the mirror.Who disputes itNo school in the record disputes it head on; it answers information processing, which meets the coordination load inside the organization; mirroring writers locate it in the product's architecture.
  9. Strategy and structure1960s to 1980sWhat it claimsStructure follows strategy: firms that diversify by product or geography outgrow the centralized functional structure and, after a lag and usually a crisis, adopt divisions. Chandler drew this from the histories of DuPont, General Motors, Standard Oil of New Jersey, and Sears.What it has shownChandler (1962) is comparative business history of four firms set against a wider survey. Rumelt (1974) related diversification and structure to performance, and Egelhoff (1988), with 34 multinationals, found some of Stopford and Wells's relationships supported and others not.Who disputes itInstitutional theory: in Fligstein's model divisions spread with a product-related or unrelated strategy, presidents from sales or finance, and other firms in the industry changing; strategy was one of three.
  10. Structural contingency theory1960s to the presentWhat it claimsThere is no one best structure. Performance depends on fit between structure and contingencies such as environment, technology, size, and strategy, on an open-system view of the organization. Drazin and Van de Ven separate fit as selection, interaction, and systems.What it has shownBurns and Stalker, Woodward, and the Aston group related structure to context; Drazin and Van de Ven (1985) supported selection and systems fit only. Pennings (1975) found that fit did not explain effectiveness; Schoonhoven (1981) found traditional hypotheses unsupported.Who disputes itStrategic choice: Child argued that contingency theory explains structure by the constraints of context and leaves out the political process by which power-holders choose strategic action, structural forms included.
  11. The French bureaucratic school (strategic analysis)1960s to 1980sWhat it claimsA bureaucracy is a system of power relations organized around zones of uncertainty. Impersonal rules and centralization protect members from depending on one another, those who control uncertainties hold power, and, unable to correct its errors, it changes only through crisis.What it has shownLe phénomène bureaucratique (1963; English 1964) rests on two case studies, a clerical agency and a state industrial monopoly. L'acteur et le système (1977; English 1980) is theoretical.Who disputes itSocietal effect and comparative business systems: D'Iribarne argues that national logics of conduct, such as a French logic of honor, explain differences that Crozier attributed to power games.
  12. The Japanese-firm literature1960s to the presentWhat it claimsThe large Japanese firm coordinates operating decisions through horizontal information sharing among semi-autonomous shops and uses its ranking hierarchy mainly as an incentive scheme (Aoki). It creates knowledge through middle-up-down management (Nonaka and Takeuchi).What it has shownAoki (1986) is a formal model. Lincoln, Hanada, and McBride (1986) compared 51 Japanese and 55 United States manufacturing plants, and Nonaka and Takeuchi (1995) rest on case studies, chiefly of Japanese firms. Fruin (1994) found the enterprise system defined at three levels.Who disputes itLabor process and critical theory reads shop-floor delegation as control: Sewell and Wilkinson argue that just-in-time and total quality regimes delegate tactical responsibility while centralizing strategic control.
  13. The Scandinavian and Dutch socio-technical schools1960s to 2000sWhat it claimsThe Scandinavian program sought democracy at work through field experiments with semi-autonomous groups, then through democratic dialogue. The Dutch school's design theory turns complex organizations with simple jobs into simple organizations with complex jobs.What it has shownEmery and Thorsrud reported field experiments in Norwegian industry; Gustavsen (1992) records that they spread less than hoped, his reason for turning to dialogue. Berggren studied work organization in the Swedish auto industry. The Dutch work rests on design cases and argument.Who disputes itThe Japanese-firm literature: defenders of the Toyota system hold that standardized work with employee participation learns faster; Berggren reports Swedish long-cycle plants matching a line plant on productivity.
  14. Configuration theory1970s to the presentWhat it claimsStructures cluster into a small number of internally coherent configurations whose elements reinforce one another, so organizations move between a few types in large jumps, not small steps. Mintzberg's five configurations each rest on one coordinating mechanism.What it has shownDoty, Glick, and Huber (1993) found that fit to Miles and Snow's ideal types predicted effectiveness while fit to Mintzberg's did not. A meta-analysis of 40 tests (Ketchen and colleagues 1997) found that configuration partly explains performance, more so with broad definitions.Who disputes itStructural contingency theory: Donaldson counters that most organizations sit at intermediate points on a continuum of fits, not in a few configurations, and that studies over time find mostly incremental change.
  15. Information processing1970s to the presentWhat it claimsThe more uncertain the task, the more information must be processed during its execution. An organization either reduces that need, through slack resources or self-contained tasks, or raises its capacity, through vertical information systems or lateral relations up to the matrix.What it has shownGalbraith (1973, 1974) built the framework from cases and from Simon's theory of bounded rationality, and Tushman and Nadler (1978) set it out as an integrating concept for design. Daft and Lengel (1986) argued that structure must reduce equivocality as well as supply data.Who disputes itThe Carnegie school and the attention-based view: Ocasio holds that firm behavior results from how the firm channels and distributes attention, so structure governs where attention goes, not only how much is processed.
  16. Labor process and critical theory1970s to the presentWhat it claimsStructure is above all a system of control over labor. Braverman argued that management deskills work by separating conception from execution; Edwards traced a sequence from simple control by the boss to technical control through machines to bureaucratic control through rules.What it has shownHistorical and theoretical (Braverman 1974; Edwards 1979). Delbridge (1995), observing a Japanese-owned plant in England, found that just-in-time and total quality regimes combined the control of labor with management's economic goals more completely than piece rates had.Who disputes itBureaucracy and its critics: Adler argues that labor process theory ignores the progressive socialization of the labor process, which drives long-term skill upgrading, and disputes deskilling as the trend.
  17. Strategic choice1970s to the presentWhat it claimsStructure is the outcome of choices by those with power in the organization, who decide on strategic action, including the structural form, the manipulation of environmental features, and the performance standards to be met. Context constrains but does not determine structure.What it has shownLargely conceptual: Child (1972) built the argument from gaps in contingency findings, and the empirical case rests mostly on the modest share of structural variance contingencies explain. A 2026 review of 42 studies finds upper echelons theory their main theoretical anchor.Who disputes itStructural contingency theory: Donaldson replies that organizations mostly adopt the structures their contingencies dictate, to avoid the loss of performance misfit brings, so choice over structure is at most limited.
  18. The German codetermination literature1970s to the presentWhat it claimsWorks councils and employee board seats form a structural layer through which coordination can become negotiation between management and employees. Defenders argue that it supports long-term employment, training, and cooperation; critics, that it weakens shareholder control.What it has shownJäger, Schoefer, and Heining (2021) found board-level codetermination left wages unchanged and, if anything, raised capital formation. Jirjahn (2011) reports studies pointing to positive works council effects; a meta-analysis found codetermination laws tied to lower productivity.Who disputes itOrganizational economics (Gorton and Schmid) found that firms with equal employee and shareholder representation did worse for their shareholders than firms with one-third employee representation.
  19. Organizational economics: decision rights, authority, and complementarities1976 to the presentWhat it claimsDesign is the joint choice of who holds which decision rights, how performance is measured, and how it is rewarded. Rights should be placed where the specific knowledge lies, formal authority differs from real authority, and practices are complements that pay off in clusters.What it has shownIchniowski, Shaw, and Prennushi (1997) found, on 36 steel finishing lines, that lines using a set of innovative work practices achieved substantially higher productivity, consistent with complementarities. Bloom and colleagues found more decentralization where trust was higher.Who disputes itGhoshal argued that management research has shaped managers' worldview with ideologically inspired amoral theories, built on a scientific model that excludes human intentionality and so denies moral considerations.
  20. Institutional theory1977 to the presentWhat it claimsFormal structures are often adopted because they are legitimate in the organization's field, as rationalized myths, and then decoupled from the work done. Organizations in a field grow alike through coercive, mimetic, and normative isomorphism, whether or not it is efficient.What it has shownTolbert and Zucker (1983) found that city characteristics predicted early but not later adoption of civil service reform, which they read as institutionalization. Palmer, Jennings, and Zhou (1993) found support for institutional and economic accounts of late M-form adoption.Who disputes itStructural contingency theory: Kraatz and Zajac found little support for neoinstitutional predictions in 631 liberal arts colleges, which grew less alike; the change they studied was curricular, not structural.
  21. Population ecology1977 to the presentWhat it claimsOrganizations face strong structural inertia, because selection favors the reliability and accountability of stable structures. Change in a population's forms comes mainly from founding and failure, not adaptation, and changing core features resets the liability of newness.What it has shownAmburgey, Kelly, and Barnett (1993) found that the effects of change on the hazard of failure vary by type of change and over time, and that rates of change depend on how often and how recently an organization has changed. Carroll and Hannan (2000) set out the program's methods.Who disputes itFor structural contingency theory, Donaldson argues that ecology understates managers' capacity to adapt structure.
  22. Resource dependence1978 to the presentWhat it claimsOrganizations depend on outside parties for critical resources, and structure follows the management of that dependence: units that cope with critical dependencies gain power, and boundaries are redrawn through cooptation, joint ventures, and mergers to absorb constraint.What it has shownPfeffer and Salancik (1978) assembled studies of boards, mergers, and joint ventures. A meta-analysis of 157 tests (Drees and Heugens 2013) corroborated the main prediction: organizations meet resource dependencies with interorganizational arrangements that raise their autonomy.Who disputes itTransaction cost economics and the M-form hypothesis explains boundaries by efficiency; resource dependence, by power and constraint.
  23. The business-groups literature1978 to the presentWhat it claimsGroups of legally independent firms bound by ownership and social ties are a durable form of their own, neither one firm nor a market. Where institutions function poorly, groups can benefit member firms and can destroy value; they are sometimes paragons, sometimes parasites.What it has shownKhanna and Palepu (2000) found that affiliates of the most diversified Indian groups outperformed unaffiliated firms. A meta-analysis of 141 studies (Carney and colleagues 2011) found that affiliation lowers performance in general, less so where institutions are underdeveloped.Who disputes itOrganizational economics: agency writers read pyramids and cross-holdings as devices that separate control from ownership; Almeida and Wolfenzon offer a financing rationale for pyramids instead.
  24. Scandinavian institutionalism1980s to the presentWhat it claimsIdeas of organizing do not diffuse unchanged; they travel and are translated as they are adopted, so a form taken up in one place is not the same form elsewhere. Organizations can also be built of organizations (meta-organizations), which behave differently.What it has shownA discussion paper by Czarniawska and Joerges (1995) contrasts the diffusion of ideas with their translation, drawing on examples such as Swedish municipal reform; the work on meta-organizations (Ahrne and Brunsson 2008) is conceptual.Who disputes itInstitutional theory can read the same processes, seen from a translocal view, as institutional isomorphism, though translation holds that each adopter modifies an idea, so an institution re-created elsewhere is new.
  25. Societal effect and comparative business systems1980s to the presentWhat it claimsFirms matched on size, technology, and industry differ systematically by country, because how a society educates, trains, recruits, and promotes its workforce, and its industrial relations, shape how work is divided and supervised, so structure is partly a national product.What it has shownMaurice, Sorge, and Warner (1980) compared closely matched manufacturing units in France, West Germany, and Great Britain and found more pervasive differences in their organization than usually recognized. Whitley (1999) set out a comparative account of business systems.Who disputes itStructural contingency theory stresses culture-free elements of structure; matched comparisons find nationally different shapes of organization.
  26. The network perspective on organizations1985 to the presentWhat it claimsOrganizations and their members are nodes joined by ties, and the pattern of ties shapes what they can do. Economic action is embedded in networks of personal relations (Granovetter 1985), and network forms are a distinct form, not a hybrid of market and hierarchy (Powell 1990).What it has shownPodolny and Page (1998) review the research and find that network forms foster learning, confer status and legitimacy, bring economic benefits, help manage resource dependencies, and give employees considerable autonomy, and they ask for attention to networks' dysfunctions.Who disputes itTransaction cost economics: Williamson treats the hybrid as a third generic form between market and hierarchy; for Powell, network forms are a distinct form, not a mix of the two.
  27. Computational organization design and search1990s to the presentWhat it claimsDesign choices interact, so design space is a rugged landscape with many local peaks, and structure affects how well an organization searches it. Decentralizing interdependent decisions can trap units on poor peaks; temporary decentralization, then reintegration, can help.What it has shownMostly simulation, with one field test: Csaszar (2012), with more than 150,000 stock-picking decisions by 609 mutual funds, found that decision-making structure affects the number of initiatives pursued and the omission and commission errors made.Who disputes itNo school in the record disputes it head on; it disputes structural contingency theory, whose bivariate fit, its simulations show, cannot capture how the best design depends on interactions among choices.
  28. The new-forms debate1990s to the presentWhat it claimsOrganizations must solve four universal problems (dividing tasks, allocating them, providing rewards, and providing information), and a form is new when it solves them in new ways. Self-managing organizations radically decentralize authority; skeptics hold hierarchy reappears.What it has shownFoss (2003) traced the rise and partial reversal of Oticon's spaghetti organization to the problem of selective intervention. A meta-analysis of 54 studies of 13,914 teams (Greer and colleagues 2018) found that hierarchy within teams on net slightly lowers team performance.Who disputes itBureaucracy and its critics: removing formal structure leaves informal elites whose power is unaccountable (Freeman), and democratic and network organizations are much less hierarchy-free than their labels suggest.
  29. The algorithmic-management literature2015 to the presentWhat it claimsSoftware does much of the directing, evaluating, and disciplining that managers and rules once did. Kellogg, Valentine, and Christin (2020) call algorithmic control a new contested terrain; Puranam (2021) treats dividing decisions between people and machines as a design problem.What it has shownMostly qualitative field studies of gig work. Wood and colleagues (2019), with 107 interviews in six countries and 679 survey respondents, found that algorithmic management tends to give remote gig workers flexibility and autonomy, but also low pay, isolation, and overwork.Who disputes itLabor process and critical theory reads control at work as contested terrain; Kellogg, Valentine, and Christin take up that view but argue that algorithmic control is distinct from earlier forms.

The forms by first year

  1. Functional structure: 1911 to 1937
  2. Line and staff: 1916 to 1943
  3. Multidivisional (M-form) by product: 1920s practice; named and analyzed 1962 to 1975
  4. Machine bureaucracy: 1922 (Weber); 1979 (Mintzberg)
  5. Divisional by customer or market segment: 1937; 2005
  6. Matrix organization: 1950s to 1960s practice; analyzed 1971 to 1977
  7. Co-determined organization: 1951 to 1976 (the laws); 1984 to 2022 (the main studies)
  8. Socio-technical autonomous work groups: 1951 to 1965
  9. Stratified hierarchy (requisite organization): 1951 to 1989
  10. Cooperative organization: 1958 to 1999
  11. Linking-pin organization (System 4): 1961 to 1967
  12. Divisional by region: 1962 to 1972
  13. Adhocracy (innovative organization): 1968 to 1979
  14. J-form (the Japanese firm): 1968 to 1990
  15. Professional bureaucracy: 1968 to 1979
  16. Holding company (H-form): 1972 to 1975; 1987
  17. Ambidextrous organization: 1976 to 2004
  18. Simple structure: 1979
  19. Hybrid structure: 1980 to the present
  20. Missionary (ideological) organization: 1980 to 1989
  21. Network, modular, and virtual organization: 1986 to 1990
  22. Transnational and multidimensional organization: 1987 to 1989
  23. Business group (keiretsu, chaebol, family house): 1988 to 2007
  24. Shamrock organization: 1988 to 1989
  25. Federal organization: 1989 to 1992
  26. Self-managing organization (less-hierarchical forms): 1989 to 2017
  27. Boundaryless organization: 1990 to 1995
  28. Front-back organization: 1990s to 2007
  29. Professional partnership (P2 form): 1990
  30. The corporate center and shared services: 1991 to 2000
  31. Internal market organization: 1993 to 1994
  32. Process (horizontal) organization: 1993 to 1999
  33. Cellular organization: 1994 to 1997
  34. Hypertext organization: 1994 to 1995
  35. Team-based organization: 1995
  36. Project-based organization: 2000 to 2006
  37. Agile at scale: 2001 to 2016
  38. Distributed (all-remote) organization: 2002 to 2021
  39. Open-source and community-based organization: 2002 to 2007
  40. Algorithmically managed organization: 2015 to 2020
  41. Team of teams: 2015

Twenty-three of the 29 schools run to the present, and 16 of those began in the 1970s or later. The forms arrive in waves: 14 of the 41 first appear between 1985 and 1994.

Source: Fayol 1949; Weber 1978; Mayo 1933; Trist and Bamforth 1951, an interview-based field study in one industry; Burns and Stalker 1961, a field study of twenty firms; Mintzberg 1979; Child 1972a; Coase 1937; Meyer and Rowan 1977; Hannan and Freeman 1977; Galbraith 1973; Jensen and Meckling 1976; Crozier 1963; Emery and Thorsrud 1969; Yoshino 1968; Streeck 1984a; Kellogg and colleagues 2020, a review of earlier research through one theoretical lens, with no new data; Puranam and colleagues 2014, a conceptual argument; Chandler 1962, several independent case histories by a historian; Pfeffer and Salancik 1978; Simon 1946; Maurice and colleagues 1980, a comparison of matched factories in three countries; Levinthal 1997; Simon 1962; Braverman 1974; Hamilton and Biggart 1988, a comparison of three economies by scholars; Ward 1958; Czarniawska and Joerges 1995; Granovetter 1985.

Gary and Judith Olson's "Distance matters" (2000), a review of ten years of field and laboratory studies of synchronous collaboration before current tools, gave remote work a chance with high common ground, loosely coupled work, and readiness for collaboration and its technology.

Sector of origin: software and knowledge work (GitLab, in Choudhury and colleagues 2020, one case study).

Where it fits

Nine cells are our judgment, two rest on theory, and one on evidence; two are contested. No work tests the form as a whole against any factor. Burton and Obel's (2004) multi-contingency model, the most developed scholarly fit model, supplies the strategy and technology premises.

The distributed (all-remote) organization against the twelve contingency factors

Each factor at its three levels, Fits, Partial, or Strain; select a row for its reason, sources, and grade.

Select a factor to read its reason, sources, basis, and grade.Size. Written, asynchronous coordination works in small and mid-size groups; in one very large firm, firm-wide remote work made collaboration networks more siloed, with fewer bridges. Basis: Judgment. Sources: Yang and colleagues 2022; Herbsleb and Mockus 2003; Choudhury and colleagues 2020.Age. Organizations founded remote build the written habits the form needs from the start; older ones must replace co-located, informal routines. Basis: Judgment. Sources: Choudhury and colleagues 2020; Hannan and Freeman 1984.Environmental uncertainty (unpredictability, dynamism). Written standards coordinate well when work can be specified ahead; when new information must spread fast, dispersion makes it harder to acquire and share. Basis: Judgment. Sources: Yang and colleagues 2022; Olson and Olson 2000; Cramton 2001; Pennings 1975.Environmental complexity. Written, asynchronous coordination depends on common ground and loosely coupled work, not on how complex the environment is. Basis: Judgment. Sources: Olson and Olson 2000.Munificence and hostility. Working without shared offices saves cost under pressure, but nothing read ties the form to munificence. Basis: Judgment. Sources: Bloom and colleagues 2015.Diversity of products and markets. Dispersed, written coordination works under any business mix; nothing read ties it to diversity. Basis: Judgment. Sources: Olson and Olson 2000.Strategy type. Burton and Obel name the virtual network, spatially separated units communicating asynchronously, as a configuration a prospector may use. Basis: Theory. Grade: conceptual only, no empirical test. Sources: Burton and Obel 2004.Technology and routineness. Written, asynchronous standards carry work that can be written down; Burton and Obel say nonroutine work needs group meetings and rich media, which dispersion makes costly. Basis: Judgment. Sources: Burton and Obel 2004; Perrow 1967; Hickson and colleagues 1969; Mohr 1971.Interdependence. Remote collaboration has a chance with loosely coupled work and strains with tightly coupled work, and interdependence between offshored and onshore processes lowered performance, a harm that coordination mechanisms reduced. Basis: Theory. Grade: conceptual only, no empirical test. Sources: Olson and Olson 2000; Srikanth and Puranam 2011.Knowledge intensity and professionalization. Works disagree: remote work raised output for call-center staff in one randomized trial and for patent examiners given more flexibility, while other studies found remote call-center workers and skilled professionals less productive and remote networks more siloed; training level does not decide either way. Basis: Evidence. Grade: a single case or a practitioner's report. Sources: Bloom and colleagues 2015; Choudhury and colleagues 2021; Gajendran and Harrison 2007; Olson and Olson 2000; Yang and colleagues 2022; Emanuel and Harrington 2024; Gibbs and colleagues 2023. View 1: Low fits, Moderate fits, High fits. Working from home raised call-center output against office work in a randomized trial, moving patent examiners from work-from-home to work-from-anywhere raised their output further, and telecommuting had small but mainly beneficial effects on performance across 46 studies, so training level is not what decides; how tightly the work is coupled matters more. Sources: Bloom and colleagues 2015; Choudhury and colleagues 2021; Gajendran and Harrison 2007; Olson and Olson 2000. View 2: Low partial, Moderate partial, High partial. In a Fortune 500 firm's call centers remote workers answered 12 percent fewer calls per hour, much of the gap reflecting who chose remote jobs; over 10,000 skilled professionals at an Indian technology company worked more hours from home with slightly lower output, with more coordination time and fewer uninterrupted hours; and firm-wide remote work made information workers' networks more static and siloed. So remote work does not fit by default at any level of training. Sources: Yang and colleagues 2022; Emanuel and Harrington 2024; Gibbs and colleagues 2023.Global integration pressure. Distributed work does best when the work is loosely coupled; when activities in different places must be tightly coordinated, distributed work items took far longer and dispersed teams lost mutual knowledge, unless the firm invested heavily in coordination. Basis: Judgment. Sources: Herbsleb and Mockus 2003; Cramton 2001; Olson and Olson 2000; Srikanth and Puranam 2011. View 1: Low partial, Moderate partial, High strain. When activities in different places must be tightly coordinated, distributed work items took far longer and dispersed teams failed to keep mutual knowledge. Sources: Herbsleb and Mockus 2003; Cramton 2001; Olson and Olson 2000. View 2: Low partial, Moderate partial, High partial. Interdependence across sites lowered offshored process performance, but investing in coordination mechanisms reduced the harm, so tight coupling is a cost that can be designed against. Sources: Srikanth and Puranam 2011.Local responsiveness pressure. Distributed members may live in the countries served, but no work read sets the distributed form against local responsiveness. Basis: Judgment. Sources: Hinds and Kiesler 2002.

Counting the two contested cells, knowledge intensity and global integration pressure, by their first views, the 36 settings read Fits at 12, Partial at 21, and Strain at 3, the strains at nonroutine technology, reciprocal interdependence, and high global integration pressure. Nine of the twelve cells rest on judgment.

Source: Yang and colleagues 2022, an analysis of 61,182 employees in one firm; Herbsleb and Mockus 2003, an analysis of change data in two organizations; Choudhury and colleagues 2020, one case study with scholarly commentary; Hannan and Freeman 1984; Olson and Olson 2000, a review of ten years of field and laboratory studies; Cramton 2001, a study of thirteen dispersed teams; Pennings 1975, a quantitative field test of the structural-contingency model; Bloom and colleagues 2015, a randomized experiment in one firm; Burton and Obel 2004; Perrow 1967; Hickson and colleagues 1969, a multi-organization comparative survey; Mohr 1971, a quantitative field study of technology and structure; Srikanth and Puranam 2011, a survey of 126 offshored processes; Choudhury and colleagues 2021, a natural experiment in one agency; Gajendran and Harrison 2007, a meta-analysis of 46 studies of 12,883 employees; Emanuel and Harrington 2024, a quantitative study of workers in one Fortune 500 firm; Gibbs and colleagues 2023, a quantitative study of over 10,000 professionals at one Indian technology company; Hinds and Kiesler 2002.

Your own situation against the distributed (all-remote) organization

Set any factor to outline its cell in the strip above; the lines under the switches give the counts, the form's place among the 41, and the misfit rules that fire.

Size
Age
Environmental uncertainty (unpredictability, dynamism)
Environmental complexity
Munificence and hostility
Diversity of products and markets
Strategy type
Technology and routineness
Interdependence
Knowledge intensity and professionalization
Global integration pressure
Local responsiveness pressure

Set: 0 of 12 factors. Fits on 0, Partial on 0, Strain on 0, Contested on 0.

Its place in the explorer's rankingEvery factor is on Any: all 41 forms tie, and the explorer lists them alphabetically.1st of 411st of 41, tied with one other1st of 41, tied with 0 othersSunk by a misfit rule: 1st of 41Sunk by a misfit rule: 1st of 41, tied with one otherSunk by a misfit rule: 1st of 41, tied with 0 others

  • Flagged (its sources disagree): A prospector bound by heavy rules. Disputed: Burton and Obel 2004; against it, Sine and colleagues 2006; Adler and Borys 1996.
  • Flagged (its sources disagree): A small organization with a large one's rules. Disputed: Burton and Obel 2004; against it, Sine and colleagues 2006.
  • Flagged (its sources disagree): Nonroutine work bound by procedures. Disputed: Burton and Obel 2004; against it, Adler and colleagues 1999; Adler and Borys 1996; against it, Hickson and colleagues 1969; Child and Mansfield 1972.
  • Flagged (its sources disagree): Heavy rules in an unpredictable environment. Disputed: Burton and Obel 2004; against it, Sine and colleagues 2006; Bigley and Roberts 2001; Adler and Borys 1996.
  • Flagged (its sources disagree): Reactor: a defender's structure kept after the environment changed. Disputed: Miles and Snow 1978; Burton and Obel 2004; against it, Sine and colleagues 2006; Bigley and Roberts 2001; Adler and Borys 1996.
  • Flagged (it meets the form only through part of its profile): Reciprocal work coordinated by standards. Source: Thompson 1967; Van de Ven and colleagues 1976.

Ranking rule: by count of Fits on the factors the reader has set, then fewest Strains, then alphabetical; a fired misfit rule that no source disputes sinks the form to the bottom when every level of the form's profile meets it, a disputed one, or one that meets the form only through one end of a spanned level, one view of a contested profile cell, or one part of a two-part form, is flagged on the row (with both views where it is disputed), and a misfit between two of the reader's own settings is shown once above the list; a contested cell counts only where its views agree at the reader's level, and otherwise shows as Contested; factors on "Any" count for nothing; with every factor on "Any" the list is alphabetical and says so.

Open these settings in the explorer

The misfit rules that can fire for this form

  • A prospector bound by heavy rules fires when Strategy type at Prospector (exploration): flagged (its sources disagree). Disputed: Burton and Obel 2004; against it, Sine and colleagues 2006; Adler and Borys 1996.
  • A small organization with a large one's rules fires when Size at Small: flagged (its sources disagree). Disputed: Burton and Obel 2004; against it, Sine and colleagues 2006.
  • Nonroutine work bound by procedures fires when Technology and routineness at Nonroutine: flagged (its sources disagree). Disputed: Burton and Obel 2004; against it, Adler and colleagues 1999; Adler and Borys 1996; against it, Hickson and colleagues 1969; Child and Mansfield 1972.
  • Heavy rules in an unpredictable environment fires when Environmental uncertainty (unpredictability, dynamism) at Unpredictable: flagged (its sources disagree). Disputed: Burton and Obel 2004; against it, Sine and colleagues 2006; Bigley and Roberts 2001; Adler and Borys 1996.
  • Reactor: a defender's structure kept after the environment changed fires when Strategy type at Defender (exploitation) and Environmental uncertainty (unpredictability, dynamism) at Unpredictable: flagged (its sources disagree). Disputed: Miles and Snow 1978; Burton and Obel 2004; against it, Sine and colleagues 2006; Bigley and Roberts 2001; Adler and Borys 1996.
  • Reciprocal work coordinated by standards fires when Interdependence at Reciprocal: flagged (it meets the form only through part of its profile). Source: Thompson 1967; Van de Ven and colleagues 1976.

Six of the 39 contingency misfit rules can fire for this form, each meeting only one end of a span, so none would sink it. Five concern heavy rules and meet the high end of its formalization; the sixth, reciprocal work coordinated by standards, meets the standardization end of its coordinating mechanism.

Source: Burton and Obel 2004; Burton and colleagues 2020a.

  • Size. Judgment: fits small and medium, partial large. Remote work made networks more siloed (Yang and colleagues 2022, an analysis of 61,182 employees in one firm); no work compares sizes.
  • Age. Judgment: fits young and established, partial old. Firms founded remote build written habits from the start; older ones must replace informal, co-located routines (premise: Hannan and Freeman 1984, a theoretical argument, ungraded here).
  • Environmental uncertainty. Judgment: fits stable and shifting, partial unpredictable. Written standards suit work specified ahead; dispersion may make new information harder to acquire and share (Yang and colleagues), and mutual knowledge is a central problem of dispersed work (Cramton 2001, a study of thirteen dispersed teams).
  • Complexity, munificence, diversity, and local responsiveness. Judgment: partial throughout; no work read ties the form to them.
  • Strategy. Theory: partial defender and analyzer, fits prospector, for whom Burton and Obel name the virtual network a possible configuration.
  • Technology. Judgment: fits routine, partial mixed, strain nonroutine, since Burton and Obel give nonroutine work group meetings and rich media, which dispersion makes costly. Limits: Hickson, Pugh, and Pheysey (1969), a multi-organization comparative survey, found the technological imperative generally unsupported, and Mohr (1971), a quantitative field study of technology and structure, found no support for the hypothesis that effectiveness depends on matching technology and structure.
  • Interdependence. Theory: fits pooled, strain reciprocal; partial at sequential is our judgment, between the two. Remote work strains with tightly coupled work (Olson and Olson 2000); interdependence lowered offshored processes' performance, a harm that investing in coordination mechanisms reduced (Srikanth and Puranam 2011, a survey of 126 offshored processes). Neither studies this form.
  • Knowledge intensity. Evidence, contested at every level, so the ranking shows Contested and counts nothing. One view, fits throughout: working from home raised call-center performance 13 percent (Bloom and colleagues 2015, a randomized experiment in one firm); work from anywhere raised output 4.4 percent for patent examiners already working from home (Choudhury, Foroughi, and Larson 2021, a natural experiment in one agency), so not remote against office work; telecommuting had small, mainly beneficial effects on performance (Gajendran and Harrison). The other, partial throughout: remote call-center workers answered 12 percent fewer calls per hour, much of it selection (Emanuel and Harrington 2024, a quantitative study of workers in one Fortune 500 firm); skilled professionals were less productive at home (Gibbs, Mengel, and Siemroth 2023, a quantitative study of over 10,000 professionals at one Indian technology company); networks grew siloed (Yang and colleagues), whose information workers we read as the moderate level.
  • Global integration. Judgment, contested; reading the coupling of work across sites as this pressure is ours. One view: strain at high, since tightly coordinated work across places took far longer and lost mutual knowledge (Herbsleb and Mockus 2003, an analysis of change data in two organizations; Cramton; Olson and Olson). The other: partial at high, a cost coordination mechanisms can be designed against (Srikanth and Puranam). Both say partial at low and moderate, the only levels the ranking counts.

What it does well

Most of this evidence measures individuals and teams, not the all-remote organization. It can raise output in work measured one person at a time. Bloom and colleagues' (2015) call-center staff at Ctrip raised performance 13 percent at home, and attrition halved. The examiners raised output without more rework.

It keeps people: hybrid work, two days a week at home, cut quits by a third (Bloom, Han, and Liang 2024, a randomized trial in one firm).

Writing can carry dialogue: four mechanisms of writing supported it at a distance (Fayard and Metiu 2014, a qualitative study of two historical collaborations).

What it costs

Work crossing sites takes longer, and, in our reading, what an office carries by talk must be written down and kept current. Distributed work items took about two and a half times as long as collocated ones (Herbsleb and Mockus 2003); skilled professionals at home worked more hours for slightly lower output, so productivity fell 8 to 19 percent, with more coordination time and fewer uninterrupted hours (Gibbs and colleagues 2023).

Ties thin. Telecommuting more than 2.5 days a week harmed relationships with coworkers (Gajendran and Harrison 2007). In teams, virtuality improved the sharing of unique information but hindered open sharing, which matters more to virtual teams' performance; low virtuality improved sharing and high virtuality hindered it (Mesmer-Magnus and colleagues 2011, a meta-analysis of 94 studies).

Careers may pay: in the 2015 trial, promotion conditional on performance fell for those at home, though not under hybrid work in 2024.

How it fails

Each mode comes from studies of teams or firms working apart; applying it to the all-remote organization is our reading.

How the distributed (all-remote) organization fails, and the coordination failures behind it

The entry's failure modes at the left, each tied to the Coordination Failures patterns it names at the right; a pattern links its entry there.

Grounding failureCoordination overhead overrunSiloed knowledge
Grounding failureCoordination overhead overrunSiloed knowledge

How many ties the failure has. Highest: Each site works from a different pictureWork that crosses sites slows downThe network closes into silosThe team splits along its sitesGrounding failureCoordination overhead overrunSiloed knowledge

Select a failure to read its place in the network.Each site works from a different picture. Degree 1Work that crosses sites slows down. Degree 1The network closes into silos. Degree 1The team splits along its sites. Degree 1Grounding failure. Degree 1Coordination overhead overrun. Degree 1Siloed knowledge. Degree 2
FailureDegree
Each site works from a different picture1
Work that crosses sites slows down1
The network closes into silos1
The team splits along its sites1
Grounding failure1
Coordination overhead overrun1
Siloed knowledge2

Four failure modes name three patterns, and siloed knowledge is named twice: by the network closing into silos and by the team splitting along its sites. The modes come from studies of teams or firms working apart, so each tie to this form is our reading.

Source: Cramton 2001, a study of thirteen dispersed teams; Herbsleb and Mockus 2003, an analysis of change data in two organizations; Yang and colleagues 2022, an analysis of 61,182 employees in one firm; Hinds and Mortensen 2005, a field study of 43 teams in one multinational; O'Leary and Mortensen 2010, a quasi-experiment with 62 teams.

Each site works from a different picture. It is a grounding failure. Cramton (2001) found five kinds of failure to keep mutual knowledge and suggests that unrecognized differences in partners' situations make them more likely to blame each other's dispositions.

Work that crosses sites slows down. Herbsleb and Mockus's distributed items took longer, apparently because they involved more people: coordination overhead overrun.

The network closes into silos. Remote work left fewer bridges across one firm's network (Yang and colleagues 2022): siloed knowledge.

The team splits along its sites. Distributed teams had more conflict (Hinds and Mortensen 2005, a field study of 43 teams in one multinational), and geographic subgroups weakened identification and transactive memory and raised conflict, while teams of geographically isolated members, nearer this form in our reading, did better than balanced or imbalanced ones (O'Leary and Mortensen 2010, a quasi-experiment with 62 teams): siloed knowledge again.

Each misfit rule is either one of Burton and Obel's (2004) propositions or our judgment. Burton, Lauridsen, and Obel (2002), a quantitative study of 224 Danish small and medium-sized firms, tested an earlier version only as a set and found lower return on assets where misfits stood. Single pairings of factor and structure are the interaction form of fit, which Drazin and Van de Ven (1985), a quantitative study of 629 work units in two states' employment security agencies, found unsupported, where the selection form the explorer uses and the systems form were supported. Volberda and colleagues (2012), a survey of 3,259 respondents in 1,904 companies, found contingency and institutional fit complementary, firms in quasi fit gaining from improving either, though firms in perfect institutional fit slightly lost performance striving for contingency fit, so a firing rule alone does not show a move would help a firm its field treats as legitimate.

No rule names this form; six meet its profile, each at one end of a span. One, our judgment, is undisputed but only flags it, meeting the standardization end of its coordinating-mechanism span, not mutual adjustment: reciprocal work coordinated mainly by standard procedures, which Thompson (1967) pairs with pooled work, though its nearest test (Van de Ven, Delbecq, and Koenig 1976, a quantitative study of 197 work units in one agency) largely supported Thompson and found modes added as a mix.

Five are disputed, so flagged, not sunk. A defender's rule-bound structure in an unpredictable environment, our judgment, resembles Miles and Snow's (1978) third route to the reactor; the counter-views to heavy rules below dispute its rule-bound half. Four are Burton and Obel's, against heavy rules in a prospector, a small organization, nonroutine work, and an unpredictable environment; the first two land where the fit table says fits. Against them: Sine, Mitsuhashi, and Kirsch (2006), a quantitative study of new ventures in one sector, on formalized founding teams outperforming (small and unpredictable, and prospector by our reading, since they measured founding teams, not strategy), though in Lee (2022), a quantitative study of many start-ups in one industry, flatter hierarchy helped creative success and hurt commercial execution; Adler and Borys (1996), a conceptual argument with no new test, on enabling formalization (all but small); Bigley and Roberts (2001), one inductive case study, on a flexible incident command system (unpredictable); and, on nonroutine work, Adler, Goldoftas, and Levine (1999), one case study of a single plant, and the Aston reappraisal (Hickson, Pugh, and Pheysey; Child and Mansfield 1972, a multi-firm comparative study), in which size bore on structure far more than technology. Limits: Schoonhoven (1981), a quantitative study of hospital operating room suites, found traditional contingency hypotheses drawn from Galbraith unsupported, while more precise interactive and nonmonotonic hypotheses did better; Glisson (1978), a path analysis of 30 human service organizations, found structure shaping routinization, so written procedures may make work more routine; and Pennings (1975), a quantitative field test of the structural-contingency model, found the fit between environment and structure did not explain effectiveness. Donaldson (2001) traces Mohr's and Pennings's failures to method and reports that later studies of the job unpredictability Mohr studied supported contingency theory.

What the research found

No study tests the all-remote organization as a design; both meta-analyses sit below it.

The productivity findings disagree: gains in individually measured, relatively independent work (Bloom and colleagues; Choudhury, Foroughi, and Larson) against later losses in call centers, much of it selection (Emanuel and Harrington), and among skilled professionals (Gibbs and colleagues). Whether gains hold for interdependent work is open.

Bloom, Han, and Liang (2024), the most recent trial read, found no effect of hybrid work on performance grades or lines of code written. It is hybrid, not fully remote, work; later works the search found are unread.

Two schools read the form differently. The modularity school expects structure to mirror the architecture of what an organization makes; Conway's law (Conway 1968, an essay), the software version, runs the other way: systems copy their designers' communication structure. Tests point both ways: Colfer and Baldwin (2016), a review of 142 empirical studies, found mirroring prevalent but not universal and unsupported in open collaborative projects, while MacCormack, Baldwin, and Rusnak (2012), a matched comparison of five pairs of software products, found the loosely coupled community's product more modular in every pair. Against information processing, the modularity school holds that standardized interfaces embed coordination, so a modular product holds a loosely coupled organization together (Sanchez and Mahoney 1996, a conceptual argument); Srikanth and Puranam found tacit coordination distinct from modularizing work. In the new-forms debate, Puranam, Alexy, and Reitzig (2014), a conceptual argument, call a form new when it solves the universal problems of organizing in new ways; skeptics hold that hierarchy reappears (Foss and Klein 2022, a book-length argument; Diefenbach and Sillince 2011, a conceptual analysis).

The evidence ledger holds 17 studies filed under this form.

In practice: one published case

GitLab, as Choudhury and colleagues describe it in "GitLab: work where you want, when you want" (Journal of Organization Design, 2020), one case study.

GitLab, a software company of more than 1,000 employees in more than 60 countries, worked all remote. It had no physical office, and every step of employment, from hiring to firing, was done remotely except a yearly companywide gathering. It coordinated asynchronously through shared repositories, a public handbook edited like code, and team chat in place of internal email. It kept an org chart, managers, and a directly responsible individual for each task.

The piece reports growth from about 374 employees a year earlier to more than 1,000, and no measures of productivity, performance, or retention.

The invited scholars, in the source's conclusion, questioned how far the all-remote asynchronous model generalizes. It may not suit work that needs physical co-presence or confidential information; it may work for a smaller set of people than expected; even an all-remote company sometimes needs colocation; and its spread may be limited by the supply of workers suited to remote work and by the risks of transparency.

The account covers one firm, described from its public handbook and pages and from conversations with employees.

Source: Choudhury and colleagues 2020. GitLab: work where you want, when you want, Journal of Organization Design 9(1), article 23; one case study with scholarly commentary.

Variants and near relatives

All-remote company handbooks. Companies with no offices publish their own handbooks, such as GitLab's (Choudhury and colleagues 2020); the handbooks themselves are practitioner reports, not cited.

Hybrid work. Arrangements mixing office and remote days, studied in a randomized trial by Bloom, Han, and Liang (2024); distinct from the Hybrid structure, which puts units on different bases side by side.

The Boundaryless organization, the nearest neighbor, loosens the boundaries between levels, functions, partners, and places without making dispersion the way the work is done. This form removes the shared site itself and coordinates through written standards and asynchronous adjustment.

Divisional by region groups by territory; here place is no basis of grouping. The Network, modular, and virtual organization moves activities to other firms by contract; the Open-source and community-based organization lets anyone contribute and earn authority. This form keeps one employer and its hierarchy. The digital organization (Burton, Obel, and Håkonsson 2020) is a property any design can have, not a form; its dispersion of work is taught here.

The distributed organization beside its nearest neighbor, the boundaryless organization

Select a number to fill what it marks in both drawings; the list under them gives the dimensions on which the two profiles differ.

It removes the shared site itself: people work apart across places and time zones and coordinate mainly through written standards and asynchronous adjustment, where the boundaryless form loosens the boundaries between levels, functions, partners, and places without making dispersion the way the work is done.

Distributed (all-remote) organization

Over digital toolsTop managementUnits(Moderate endonly)Members, apartTopmanagementUnit byfunctionUnit byproductMember in onetime zoneMember inanother zoneBy function or by productWM1112
Over digital toolsTop managementUnits (Moderate end only)Members, apartTopFunctionProductOne zoneFar zoneBy function or by productWM1112

Reports toAdjust to each other directlyMMutual adjustmentWStandardization of work processes

Boundaryless organization

TopUnits(Moderate endonly)Front lineAcross theboundariesTopmanagementFunctionalunitProduct unitSupplier,customer, partnerFront-lineteamFront-lineteamCross-boundaryforumBy function or by productM12
TopUnits (Moderate end only)Front lineAcross the boundariesTopFunctionProductPartnerTeamTeamForumBy function or by productM12

Reports toAdjust to each other directlyMMutual adjustmentThe firm; partners stand outside it

  1. Members work apart across places and time zones and adjust to one another over digital tools, where the boundaryless organization brings people from many units together in a cross-boundary forum.
  2. Handbooks and other written standards join mutual adjustment in the distributed organization, formalized medium to high; the boundaryless form's one prime mechanism is mutual adjustment, formalized low to medium.

Differs on 5 of 11 dimensions

  • Standardization and formalization Medium to high; Boundaryless organization: Low to medium
  • Prime coordinating mechanism Mutual adjustment and standardization of work processes; Boundaryless organization: Mutual adjustment
  • Lateral mechanisms Direct contact and liaison roles; Boundaryless organization: Task forces and teams
  • Mode of exchange and control Bureaucracy; Boundaryless organization: Bureaucracy and clan
  • Boundary permeability Selective to permeable; Boundaryless organization: Permeable

The distributed organization, like the boundaryless one, keeps one employer's hierarchy over whatever grouping it finds. The boundaryless organization makes the boundaries between levels, units, partners, and places permeable; the distributed organization removes the shared site and coordinates mainly through written standards and adjustment over digital tools.

Source: Hinds and Kiesler 2002; O'Leary and Cummings 2007; Choudhury and colleagues 2020, one case study with scholarly commentary; Ashkenas and colleagues 1995; Ulrich and colleagues 2002.

Moving into or out of it

Most transition evidence comes from firms that kept their offices. Microsoft's forced, firm-wide shift of 2020 made networks more static and siloed and communication more asynchronous (Yang and colleagues 2022). A planned move gained: examiners already at home raised output when allowed to work from anywhere, in one agency.

What helps: spontaneous communication, which built shared identity and context (Hinds and Mortensen 2005); writing that supplies context (Fayard and Metiu 2014); and investment in coordination mechanisms (Srikanth and Puranam 2011).

An older firm must replace co-located routines (our judgment), and Hannan and Freeman (1984) argue that inertia rises with age and that attempts at reorganization raise death rates, though tests show effects vary by type of change.

No study traces a move out of the form. Change & Transformation takes up the running of a reorganization.

The thinking behind it

Also called

Remote-first organization; all-remote company. The sources also say distributed work (Hinds and Kiesler), geographically dispersed teams (O'Leary and Cummings), telecommuting (Gajendran and Harrison), work from anywhere (Choudhury and colleagues 2021), and virtual teams (Mesmer-Magnus and colleagues); Burton and Obel (2004) say virtual network. "Virtual organization" is used both ways, for dispersion and for the contracting form under the Network, modular, and virtual organization.

Works cited

On this page, “read” means read by this series: where a line says a work, or part of one, was not read, the series did not read it.

Back to the explorer