Span of Control
What is the right span of control? That question may irritate you because some consultant, armed with a management theory but little understanding of the local situation, may tell you that your span of control is wrong. There are certainly many cases where the classic 3–7 span of control is not appropriate, so why do we cling to it? Let’s consider how to use theory, along with knowledge of the specific situation, to determine appropriate spans of control.
Why we are pushed to use span-of-control measures
The theory that span of control should normally be between 3–7 is sound. Too large a span of control is hard to manage; too narrow a span is inefficient and may create unnecessary management layers.
Unfortunately, as organizations evolve, spans of control often drift too narrow or too broad. A manager whose span has grown too large may be reluctant to give up the power that comes from directly controlling a large group. At the other end of the spectrum, a manager may want to promote a high performer by creating an assistant-manager role in a one-over-one reporting relationship. These are just a couple of many possible reasons why span of control goes awry.
We want a method for identifying situations where the span of control is ineffective, yet applying a strict guideline is too simplistic. We need an alternative.
An alternative to span-of-control guidelines
The alternative to theoretical span-of-control guidelines, which cannot account for local conditions, is to look for signals of a real problem. Start with the counterexample: if a team is working well despite an unusual span of control, why disrupt it? On the other hand, managerial burnout, complaints that employees rarely see their manager, or projects going off the rails may signal that the span is too wide. We can watch for those signals and troubleshoot where there is a real problem, rather than intervene simply because an abstract theory says we should.
For overly narrow spans of control, we might see unclear accountability—for example, employees are told they report to an assistant manager but know that, in practice, they are better off going directly to the manager. Other symptoms include duplicated management work, slow decisions because of unnecessary layers, or managers doing little genuine managerial work.
The shortcoming of this approach is that it is hard to apply at scale. In an organization-wide review, it is difficult to examine every team for signs of dysfunction. And when you do find dysfunction, there is no guarantee that changing the span of control is the solution. You still have to troubleshoot the underlying problem.
We need the simplicity of span-of-control guidelines with the specificity of an analysis of local conditions.
Combining the approaches
The problem with span-of-control theory arises when it is treated as a firm rule of good design rather than as a tool for spotting potential issues. An organization-design review can quickly identify unusual spans of control and then investigate whether there is an actual problem. Theory is a spotlight for telling us where to focus, not a rule for telling us what to do.
An aside about how we build theories
It’s no surprise that a simple model such as “span should be between 3 and 7” has limitations. It would be natural to want to add dimensions to the model such as “Let’s refine the model to consider the complexity of the work” and perhaps “Let’s also consider the impact of the industry”. You’ll notice we didn’t do that. We jumped straight from a simple model that helped us spot potential problems to a specific troubleshooting step grounded in the unique situation. The reason for this is that there are an endless number of possible factors and trying to build a model that covers them adequately is impractical. For example, in some cases the reason for a small span of control is that the manager just isn’t that good at people management, so you keep the number of direct reports small. Given all the nuances that can occur, you will have to go from theory to specifics at some point, and often a simple model is all you need to get pointed in the right direction.
Will the future be different?
We can improve our organizational design today by occasionally scanning the organization for unusual spans of control and then thoughtfully investigating individual situations to see if the unusual span is really a problem. However, the future may be different. If AI can cheaply scan engagement data, calendars, workflow, turnover, organizational-network data, employee comments, project delays, and so forth, perhaps we can move from rule-based organization design toward exception detection and diagnosis at scale.
You can ask your people analytics team to begin experimenting with loading readily available data for your department into an LLM and asking, “Are there any signals of problems caused by too wide or narrow a span of control?” The goal isn’t necessarily to have a working organization design diagnostic in the short-term; you just want to get a sense for what the LLM can do with available data. Once you see that, you can plan the next steps in using LLMs to change how we handle organization design.
The future of organization design looks exciting; I encourage you to dive in.