Cover of Designing Complex Organizations

Designing Complex Organizations

Jay Galbraith

6 ideas

  1. Information Processing Drives Structural Choice

    The central determinant of organizational form is the amount of information that must be processed during execution, which rises with task uncertainty. The greater the uncertainty, the more information must be exchanged among decision-makers to coordinate, so structure exists to manage information load.

  2. Four Strategies for Managing Uncertainty

    Organizations facing rising uncertainty can choose among four design strategies: create slack resources, create self-contained tasks, invest in vertical information systems, or create lateral relations. The first two reduce the need to process information, while the last two increase the capacity to process it.

  3. Slack Resources As Coordination Cost

    An organization can reduce its information-processing burden by lowering performance targets — adding extra time, inventory, or budget so that exceptions don't require coordination. This works but imposes a measurable cost in the form of the slack itself, making it a deliberate trade-off rather than inefficiency.

  4. Self-Contained Units Cut Information Load

    Shifting from a functional structure to one organized around outputs (products, projects, geographies) reduces the information that must travel up and across the hierarchy, because each unit holds the resources to handle its own variations. The cost is lost economies of scale and reduced specialization depth.

  5. Ladder of Lateral Relations

    When more information capacity is needed, organizations escalate through increasingly costly horizontal mechanisms: direct contact, liaison roles, temporary task forces, permanent teams, integrating roles, and finally the full matrix. Each rung moves decision-making down to where the information exists rather than pushing information up the hierarchy.

  6. No Best Way To Organize

    Any organizational form is only effective relative to the task, environment, and strategy it must serve; designs are conditional bets, not universal solutions. The right question is not 'what is the best structure' but 'which design fits this level of uncertainty and these objectives.'

Save and mark ideas in the app