Galbraith Star Model Questionnaire

To run a star model design diagnostic, use this questionnaire. It assesses the five design policy areas and their alignment, from 1 (badly misaligned) to 5 (fully aligned).

Use it before a reorganization, not after. Its main practical purpose is to prevent a structure-only change — the most common and most expensive design failure.

See galbraith-star-model.

Evidence: Moderate. The underlying information-processing theory has support; the five-point framework is a design heuristic and has not been tested as a model.

Step 1: Strategy

1. Clarity

• Question: Is the strategy specific enough to serve as a design criterion — that is, could two people use it to choose between two structures and reach the same answer?
  • Rating (1–5): [ ]

2. Choices

• Question: Does the strategy say what we will not do, and what we are willing to be worse at?
  • Rating (1–5): [ ]

3. Design criteria

• Question: Have we written three to five explicit criteria that any design must satisfy?
  • Rating (1–5): [ ]

Step 2: Structure

1. Specialization

• Question: Do we have the right kinds of specialists for this strategy, at the right depth?
  • Rating (1–5): [ ]

2. Shape

• Question: Are spans of control and the number of layers appropriate — neither starving managers of attention nor inserting layers that only relay?
  • Rating (1–5): [ ]

3. Distribution of power

• Question: Are decisions made at the level with the relevant information, and is centralization vs. decentralization a deliberate choice rather than an accident?
  • Rating (1–5): [ ]

4. Departmentalization

• Question: Is it clear which dimension has primacy — function, product, geography, or customer — and does everyone give the same answer?
  • Rating (1–5): [ ]

Step 3: Processes

1. Vertical processes

• Question: Do planning, budgeting, and resource allocation actually direct resources toward the strategy, on a cadence that matches the work?
  • Rating (1–5): [ ]

2. Lateral processes

• Question: Do we have explicit mechanisms — integrator roles, standing forums, shared metrics, defined teams — for coordinating across units without escalating?
  • Rating (1–5): [ ]

3. Escalation load

• Question: How often do cross-unit conflicts reach a common boss for resolution? (Score 5 for rarely, 1 for constantly.)
  • Rating (1–5): [ ]

Step 4: Rewards

1. Goals and metrics

• Question: Do the metrics people are judged on match the behavior the strategy requires?
  • Rating (1–5): [ ]

2. Compensation and recognition

• Question: Does pay, bonus, and recognition reinforce the strategy, or reward the previous one?
  • Rating (1–5): [ ]

3. Promotion criteria

• Question: Do the people who get promoted here demonstrate the behavior the new design requires?
  • Rating (1–5): [ ]

Step 5: People

1. Capability

• Question: Do we have the skills the design assumes exist, particularly in the roles the design newly creates?
  • Rating (1–5): [ ]

2. Development

• Question: Do we have a route to build those capabilities on the timescale the design needs?
  • Rating (1–5): [ ]

3. Selection and rotation

• Question: Do recruitment and movement of people support the design, or does everyone stay in one silo for a career?
  • Rating (1–5): [ ]

Step 6: Analysis

• The lowest-scoring point is the binding constraint, regardless of how much effort has gone into the others. A star with four strong points and one weak point behaves like the weak point.

• Check Rewards specifically. Rewards is the point most commonly left untouched during a redesign, and it silently vetoes the other four. If Step 4 scores below 3 while Structure scores above 4, expect the reorganization to fail and to be blamed on culture.

• Low Step 3 lateral process scores with high Structure scores mean you have drawn boxes and not designed coordination. This is the cheapest thing to fix: a named integrator role or a standing cross-unit forum frequently resolves what looks like a structural problem, at a fraction of the cost.

• A low score on Step 2, item 4 — dimension primacy — predicts constant escalation. If the leadership team cannot agree which dimension wins, the organization has a matrix in which every conflict routes to the top.

• Low Step 1 scores mean stop. Design arguments without strategy criteria are unresolvable, and the loudest or most senior person will win by default.

• Name what will get worse. Every design accepts a tradeoff. Before proceeding, write down what this design is deliberately choosing to be worse at. Unnamed downsides become surprises, and surprises become evidence that the design failed.