Rummler and Brache's Nine Boxes

Most executives know their products, their customers and their competitors well. Far fewer understand, in real operational detail, how their organization actually develops, makes, sells and delivers what it sells. Asked to draw their business on paper, most people default to the traditional organization chart, a hierarchy of boxes connected by reporting lines.

That chart has genuine value as an administrative reference. But it says nothing about the business's customers, its products and services, or the workflows used to build and deliver them. It cannot explain what the organization does, whom it serves or how the work actually gets done. That gap rarely matters in a small or young organization, where everyone understands each other's roles well enough to compensate. As the organization grows and its environment and technology become more complex, the vertical view the org chart provides turns into a liability.

When managers see their organizations only vertically, they manage them that way too. Goals get set function by function in isolation, and subordinate managers start treating other departments as rivals rather than partners in a shared competitive effort. A silo culture takes hold, pulling managers into resolving turf disputes instead of focusing on customers and competitors. Each department can excel by its own local measures while the organization as a whole quietly underperforms.

Geary Rummler and Alan Brache introduced their Nine Boxes Model in 1990 as a corrective to exactly this problem, in their book Improving Performance: How to Manage the White Space on the Organization Chart, a work that later sold more than 150,000 copies and helped launch the modern process improvement movement. 1 They called the space between the boxes on an org chart the white space, and argued that the biggest performance gains usually sit precisely there, at the points where work or information passes from one function to another.

Seeing the organization horizontally

Understanding what an organization does, for whom and how, requires a horizontal or system-based view rather than a vertical one. This kind of view restores the three elements a traditional org chart omits: the customer, the product or service, and the flow of work that cuts across functional boundaries to produce it.

Organizations behave as adaptive systems. They take in resources, convert them into products and services that match market needs, and generate financial value for their owners, all while responding to competitors and to the broader social, economic and political environment around them. Because everything inside and outside an organization is connected in this way, the biggest opportunities for improvement sit at the functional interfaces, the exact points where a product, a document or a piece of information is handed from one department to the next.

The three levels of performance

Once an organization is viewed horizontally, its performance can be understood at three levels. These levels form the basic anatomy of performance: each is essential, each depends on the other two, and a failure at any one of them limits how well the whole organization can perform.

The organization level is the broadest lens. It covers the enterprise's relationship with its market and the basic skeleton of its major functions, including strategy, organization-wide goals and measures, structure and how resources are deployed. This is the level closest to what a traditional org chart shows, and it sets the direction and boundaries that the other two levels operate within.

The process level sits beneath it, and it is where an organization actually produces its output through work that cuts across functional lines to meet customer needs efficiently and effectively. An organization is only ever as good as its processes. The organization level can set direction and flag threats or opportunities, but substantive change usually has to happen here, at the process level, because even a sound strategy and capable people cannot compensate for a process that is fundamentally broken.

The job or performer level is the narrowest lens, focused on the individuals who perform and manage the work itself. This level covers hiring and promotion decisions, job responsibilities and standards, feedback, rewards and training. It rounds out the model, since even the best strategy and the best-designed process still depend on people who understand what is expected of them and have what they need to deliver it.

The three performance factors

The three levels form only one dimension of the model. The second dimension asks how well an organization's goals, structures and management practices work at each of those levels, since overall performance, meaning how well the organization meets customer expectations, is a product of all three factors operating together across all three levels.

Goals set the specific standards each level needs, reflecting what customers expect in terms of quality, quantity, timeliness and cost. Design covers the components and configuration needed at each level to let those goals be met efficiently. Management covers the ongoing practices that keep goals current and ensure they are actually being achieved. Crossing these three factors against the three levels produces nine variables, a comprehensive set of improvement levers available to managers at any point in the organization, arranged as a three-by-three matrix: the nine boxes.

This structure also explains why narrow improvement efforts tend to underdeliver. Training alone rarely fixes organization or process performance, because training addresses only the job or performer dimension and leaves the other two levels untouched. Automating a process rarely delivers its full potential either, unless that automation is tied back to organizational goals and to the people who will operate within it.

Managing the organization level

At the organization level, goals are strategic. A sound strategy identifies the organization's products and services, its target customer groups, its sources of competitive advantage and its priorities across products and markets. Strong organizational goals incorporate the organization's values, customer requirements, financial and nonfinancial expectations, targets by product family and market, and clear expectations for each competitive advantage the organization wants to build or protect. These goals should be grounded in the critical success factors of the industry, informed by competitive and environmental scanning such as benchmarking, and stated in quantifiable terms wherever possible.

Setting organization goals well means answering three questions: has the strategy been clearly articulated and communicated, does it hold up against external threats and opportunities and internal strengths and weaknesses, and have the required outputs and expected performance levels been determined and communicated. Goals also need regular reassessment, since treating them as fixed once achieved breeds the kind of complacency that blocks continuous improvement.

Organization design is the discipline of structuring the organization so those goals can actually be met. A relationship map, which traces the customer-supplier relationships among an organization's line and staff functions, is a useful diagnostic tool here. It shows the inputs and outputs flowing between functions and exposes the white space between the boxes of the org chart, helping leaders see how work really gets done, spot disconnects such as missing or misdirected inputs and outputs, and evaluate alternative ways of grouping people and defining reporting lines. Organization design starts by examining and improving these input-output relationships, guided by questions such as whether all the necessary functions exist, whether each is truly needed, and whether the current flow between functions actually supports the strategy.

Organization management spans four areas. Goal management means ensuring every function's subgoals reinforce organizational goals rather than optimizing locally at the expense of the whole, the way a product development team should chase market needs rather than technical novelty for its own sake. Performance management means gathering regular feedback, forming cross-functional teams to solve shared problems, and measuring results against goals so they can be reset as conditions change. Resource management means allocating people and budget in line with organizational goals rather than through blunt, ad hoc measures like across-the-board headcount cuts that can quietly undermine the very goals they are meant to protect. Interface management means paying close attention to the white space between functions, since a goal like a successful product launch depends entirely on how well marketing, product development and field operations work together across that boundary.

Managing the process level

Every process is a sequence of steps standing between an input and an output. Some processes stay entirely within one function, but most cut across several, spanning the same white space that shows up on an organization chart. Rummler and Brache distinguish three kinds of processes: primary processes that produce something an external customer receives, such as manufacturing, distribution or customer service; support processes that are invisible to the customer but essential to running the business, such as recruiting, budgeting or purchasing; and management processes, the actions managers take to guide the business, such as strategic planning and performance management.

A process behaves like a value chain, with each step adding to the value created by the one before it, and it consumes real resources of capital, people, time, equipment and materials along the way that need to be weighed against the value produced. Because an organization can never outperform its processes, no amount of individual skill can permanently offset a process that is poorly conceived.

Process goals are less intuitive than organization or job goals because processes, unlike functions or departments, rarely have their own explicit targets. A billing process, for instance, may have no goal of its own even though the billing department certainly does. Process goals typically come from three places: organizational goals, customer requirements and benchmarking data, with process benchmarking against a comparable process in a high-performing organization being especially useful. The central question is whether the goals for key processes genuinely align with what customers and the organization need.

Process design follows once those goals are set, and the tool of choice here is the process map, built by a cross-functional team to document, step by step, how a process actually converts inputs into outputs. Where a relationship map shows how functions relate to each other, a process map shows the sequence of steps within a specific process, and building one often reveals that no formal process exists at all, that the work simply happens through habit and improvisation. The key design question is whether the process, as it stands, is the most efficient and effective way to achieve its goals.

Process management requires four ongoing disciplines. Goal management means breaking overall process goals into subgoals at each stage, such as completing every order within 24 hours, and then deriving functional goals from those subgoals so that each function is measured by how well it serves the process rather than by its own isolated targets. Performance management means building systems to capture internal and external customer feedback on process outputs and appointing process owners accountable for the process end to end. Continuous improvement depends on tracking performance against goals and subgoals, feeding that information back to every function involved, building mechanisms to resolve process problems as they surface, and adjusting goals as customer needs shift. Resource management, at the process level, means calculating the people and budget a process needs to hit its goals first, then allocating each function's share based on its contribution, so that a function's budget becomes the sum of what each process it supports requires. Interface management means treating every customer-supplier handoff between functions within a process as a point that must be actively monitored and measured for quality and efficiency.

The essential questions for managing a process are whether appropriate subgoals have been set, whether process performance is actually being measured, whether sufficient resources have been allocated, and whether the interfaces between process steps are being managed rather than left to chance.

Managing the performer level

It would be a mistake to assume that poor individual performance is the root of most organizational problems, since people are only one part of a much larger performance engine. Rummler and Brache used the Human Performance System, a model built around six variables, to explain what actually drives performance at the individual level. The goals, design and management already in place at the organization and process levels form the backdrop for this system, which zooms in to give a micro-level picture of the individual and their immediate environment, following the same input-process-output-feedback logic found at the higher levels.

Inputs are the raw materials, requests, assignments and resources, including the systems and procedures inherited from the process level, that prompt someone to act. Performers are the individuals or teams who convert those inputs into outputs, and their traits, skills, knowledge and behavior all shape the result. Outputs are the products or services performers generate in service of organizational and process goals. Consequences are what performers experience after producing an output, whether positive, such as recognition or a bonus, or negative, such as a complaint or disciplinary action, and the same consequence can register differently depending on the performer's own perspective. Feedback is the information, whether from error reports, statistics, rejected work or performance reviews, that tells a performer what they are doing and how well they are doing it.

Overall output quality is a function of all four of these variables working together: input quality, performer capability, consequences and feedback. Managing the performer level well means systematically analyzing and improving all of them, because jobs that are not designed around process requirements, or environments that do not let people contribute their best, will keep organization and process goals out of reach no matter how talented the individuals are.

Job goals must reflect the contribution each role makes to a process, not just to a function, which is the core distinction between this approach and a traditional goal cascade: job goals should link to functional goals, but both should ultimately trace back to the processes they support. Goals need to specify both what a performer is expected to produce and how well, since those two elements define the output component of the Human Performance System. The key questions are whether job outputs and standards connect to process requirements, and whether those in turn connect to customer and organizational requirements.

Job design covers how responsibilities are divided among roles, how job activities are sequenced, what policies and procedures govern the work, and whether the physical or ergonomic environment supports it. The relevant questions are whether process requirements show up in the right jobs, whether job steps follow a logical sequence, whether supportive policies exist, and whether the job environment is ergonomically sound.

Job management addresses six factors that determine how well the Human Performance System functions for a given role. Performance specifications are the outputs and standards that make up job goals, and managers need to set these participatively, in a way that is clearly derived from process requirements. Task support means giving performers high-quality inputs, minimal interference and logical procedures, along with the resources they need, a factor closely tied to job design itself. Consequences must actively reinforce the achievement of job goals; a company that wants to sell more of a new product but still pays commissions tied to an older one is sending a consequence that works against its own stated goal, and consequences only motivate if they are meaningful to the performer and delivered promptly enough to matter. Feedback tells a performer whether to keep going or change course, and without it, good performance can drift and poor performance can persist unchecked. Skills and knowledge, including both the formal way of doing a job and the informal shortcuts that separate strong performers from average ones, are necessary for the job to be done well at all. Individual capacity, the physical, mental or emotional capability to do the work, sets a final limit; a salesperson who cannot handle rejection, for instance, may simply lack the capacity the role demands regardless of training or support.

Taken together, job management comes down to six questions: do performers understand their job goals, do they have the resources and job design needed to succeed, are they rewarded for hitting those goals, do they know whether they are meeting them, do they have the necessary skills and knowledge, and would they have the capacity to succeed even if every other condition were met.

The nine boxes in practice

Crossing the three levels with the three factors produces the nine boxes: organization goals, organization design and organization management; process goals, process design and process management; and job goals, job design and job management. Each box is a distinct variable that can be diagnosed and improved on its own, but all nine stay connected, so a change in one typically ripples into the others.

Consider a company that sets an ambitious new strategic goal to cut order-to-delivery time in half. That decision at the organization level immediately raises questions in every other box. Does the order fulfillment process, as currently designed, have any realistic path to that speed, or does it need a redesign built around a fresh process map? Do process owners have the authority and measurement systems to manage cycle time directly, rather than tracking it after the fact? Do the individual performers, the warehouse staff, the customer service representatives and the logistics coordinators, have job goals, tools and training that reflect the new target, or are they still being measured against the old one?

If the organization sets the goal but never touches process design, the strategy stalls exactly in the white space the model is built to expose. If it redesigns the process but never updates individual job goals or training, performers keep working the old way regardless of what the new process map says. Rummler and Brache's central argument is that durable performance improvement requires deliberate alignment across all nine boxes, not a single intervention at whichever level happens to be most visible.

Where the model gets used

The nine-box framework applies across a range of situations: diagnosing and correcting underperformance, such as excess cycle time in manufacturing or shrinking margins in retail; continuously improving a system that already works adequately, such as how an airline adapts to requests for unusual aircraft configurations; guiding an organization into new territory, such as entering the software business; or designing an entirely new operation from scratch, such as a next-generation factory or a new marketing function. It can be applied by executives shaping vision and driving change, by managers rolling out companywide initiatives, and by analysts designing the specific systems and procedures that make the higher-level goals real.

Organizations using the model as a diagnostic tool typically start at the process level, since that is where the most cross-functional value gets created and where the largest gaps tend to hide. From there, they trace upward to confirm organizational goals and structure actually support the process, and downward to confirm individual jobs are equipped to deliver it.

This systemic view stands in deliberate contrast to piecemeal improvement efforts, which Rummler and Brache treat as potentially as damaging as taking no action at all, since a local fix that ignores the surrounding system often just creates a new bottleneck elsewhere. A manager who improves one function's local metrics can easily degrade the end-to-end process customers actually experience, unless someone is explicitly managing the white space between functions.

The model has been applied across manufacturing, service and technology organizations, and its consulting applications reportedly span more than 250 documented projects with companies including Hewlett-Packard, 3M, Shell Oil and Citibank. 2 Decades after its introduction, it remains a foundational reference in business process management and organizational design, largely because it forces a specific question that most frameworks skip: where, precisely, does the handoff break down, and who is accountable for fixing it.

• 1Improving Performance, 3rd Edition, Wiley
• 2Improving Performance, Amazon