Organization Design

Galbraith's Star Model

Galbraith's star model aligns five choices in organization design: strategy, structure, processes, rewards, and people.

The questionIf this is our strategy, what structure, processes, rewards, and people policies does it call for, and do ours agree with each other?

Model · Jay Galbraith, 1970s onward

StrategyDirection and advantage
StructureWhere authority sits
ProcessesVertical and lateral flows
RewardsPay, promotion, recognition
PeopleHR policies

Schematic of the five design points, each of which has to agree with the others.

When to use it

A new strategy has been announced and the org chart has not changed, or the chart has changed and nothing else did.
The client says "we need a restructure" and you suspect structure is one of several misaligned choices.
A merger, a new business unit, or a move toward a matrix needs a design sequence and a way to compare options.
Incentives reward behavior that the strategy says it does not want.

When to avoid it

The strategy is still unsettled. The model has nothing to fit the organization to, and the design will move every time the strategy does.
You need to diagnose why an existing organization is underperforming, with no design decision in view. A diagnostic model such as Nadler-Tushman or Burke-Litwin fits that better.
The difficulty is how individuals are experiencing a change, not what the policies are. A transition model will serve you better.
The organization is small and informal enough that a full design exercise would cost more than it finds.

Overview

The Star Model treats organization design as five connected sets of choices: strategy, structure, processes, rewards, and people. Strategy comes first and supplies the criteria for choosing among structures. The other four are then set so they support one another. It supports a design decision, such as how to organize a new business unit, where a structure-only answer would miss the policies that actually drive behavior.

Its main claim is fit. A structure that expects cooperation across units fails if pay rewards only unit results. The model is a practitioner framework, widely taught and plausible, but it does not tell you which fit works best in a given case. It also does not diagnose an existing problem. For that, start with a diagnostic model.

The five points

Strategy

Direction. The products or services, the markets, the value offered, and the sources of advantage. In this model strategy is the first point addressed, because it supplies the criteria for choosing among structures. Each organizational form makes some activities easy at the expense of others, and strategy says which ones matter most.

Structure

Where power and authority sit. This covers how jobs are specialized, how many people report to each manager and how many levels result, how decision rights are split between the center and the units, and how departments are formed, whether by function, product, market, customer, or geography. Matrix structures are included.

Processes

The flow of information and decisions across the structure. Vertical processes, such as planning and budgeting, allocate funds and talent. Lateral processes follow the work across units, such as product development or order fulfillment, and are the usual way to coordinate across structural lines.

Rewards

Pay, promotion, bonuses, recognition, and other policies that align what individuals want with what the organization needs. The point of the model is that the reward system must be consistent with the structure and the processes, or it will pull behavior the other way.

People

Human resource policies. Recruiting, selection, rotation, training, and development produce the skills and mind-sets the design requires. A design that depends on cross-unit cooperation needs people who can cooperate, and these policies take the longest to change.

What it is not

Five independent topics. The model's claim is consistency among them. A choice made in one point without checking the others tends to cause damage in another.
A structure-first method. The model says structure is one policy among five, and that design efforts commonly spend too much time on the org chart and too little on processes and rewards.
A model of culture. Culture is treated as an outcome that the five policies influence. It has no point of its own to design.

The steps

Turn the strategy into design criteria

State where the firm competes, how it wins, and what the organization has to be good at. Translate that into four to six criteria you can score options against, such as speed of local decisions, or coordination across product lines. If the strategy cannot produce criteria, it is not yet specific enough to design from. What comes out is a criteria list the executive team accepts.
Describe current policies on all five points

Collect the actual documents, not the intended ones. That means org charts and decision rights, the planning calendar, cross-unit processes, compensation plans and promotion criteria, and HR policies. Interview people two levels down about how each works in practice. What comes out is a one-page description of the current state.
Compare structure options against the criteria

Generate several forms, such as functional, product, market, hybrid, and matrix. Score each against the criteria and write down the trade-offs openly, since every form has costs. Choose the form whose costs the other policies can best offset. What comes out is a scored comparison and a chosen structure with its costs stated.
Design the other four points to fit

Set the processes, rewards, and people policies so they support the chosen structure and each other. Check pairs, such as whether the compensation plan pays for the lateral behavior the structure assumes. What comes out is a design for each point, and a list of changes from the current state.
Test for fit and sequence the changes

List every remaining mismatch with an owner and a fix. Order the changes by what most alters behavior, and allow for lead times. Hiring and development take years, and announcing a structure takes a day. What comes out is an implementation sequence that says what changes first and why.

In practice

The client had just split its professional services arm from its product business, and the CEO's first question was where the reporting lines should go. The consultant started somewhere else and asked what the strategy needed the organization to be good at. The answer was selling integrated solutions to a few hundred large accounts, which meant coordinating product and services teams at the account level, and building real services expertise. Those became the design criteria.

Against those criteria, the current policies disagreed with each other. Structure kept products and services in separate divisions. The planning process ran in two silos and had no joint account plan. The sales compensation plan paid on product bookings only, so no salesperson had a reason to bring in services. People policies moved services staff into product roles as a standard career step, which drained the expertise the strategy needed. The CEO had expected a structure decision. The recommendation was to change the compensation plan and add a lateral account-planning process first, then settle the structure.

What you leave with

Design criteria derived from the strategy, written so alternatives can be scored against them.
A one-page description of current policies on all five points.
A scored comparison of structure options, with the chosen form and its trade-offs stated.
A fit check listing each mismatch between points, with a fix and an owner.
An implementation sequence that says which policy changes first and why.

Failure points

An org-chart-only redesign

Boxes move, the compensation plan and planning process stay put, and behavior does not change. People follow the measurement and pay systems they live with, not the chart.

A strategy too vague to design from

A statement such as "be more customer-centric" produces no criteria, so every structure scores about the same and the choice reverts to preference or politics. Push the strategy until it excludes something.

Fit treated as a checkbox

The team marks all five points as aligned without examining the pairs. The conflicts live in the pairs, such as individual rewards against a lateral process that needs shared credit.

Ignoring lead times for people policies

The structure is announced on day one, and the skills it needs are years away. For a long period the organization has the new design and the old capabilities.

Designing in the executive room only

The people who run the processes know where policies collide. If they are not consulted, those frictions surface after implementation.

The failures above are usually Stakeholder Algebra: while the analysis was precise, the people who decide were not handled correctly.

Origin

Jay Galbraith was an American organization design scholar and consultant. His firm says he developed the Star Model in the 1960s, and he wrote about organization design from the early 1970s, beginning with Designing Complex Organizations (1973). His later book Designing Organizations, which sets out the model for executives, went through several editions. In Designing Dynamic Organizations (2002), written with Diane Downey and Amy Kates, the authors give tools for turning a strategy into design criteria. Galbraith himself credits Alfred Chandler's Strategy and Structure (1962) as the start of strategic organization design.

The five points are strategy, structure, processes (including the lateral ones that cut across units), rewards, and people policies, meaning human resource practices. Galbraith's firm marks Star Model with a trademark symbol and holds copyright on its summary document. This note describes the idea in its own words and does not reproduce the official diagram.