Managerial Decision Making: Insights from Control Theory

By Tariq Samad

Control theory has developed a body of knowledge that has resulted in remarkable accomplishments with complex dynamical systems, such as landing spacecraft on distant planets, keeping continental-scale power grids stable, regulating human hearts with pacemakers, and more. In these applications, control systems make decisions under uncertainty, relying on mathematical models, data analysis, and optimization algorithms.

But engineered systems are not the only complex dynamical systems we need to control. Organizations are complex dynamical systems too, and we use the term management to refer to their control. Mathematical modeling is, with rare exceptions, infeasible in the organizational context, but the concept of models is directly relevant.

Consider a manager of an engineering group. She has objectives which could include delivering a product design or prototype on time and within an allocated budget. She has engineers (and others, including lower-level managers) in her organization to whom she gives direction and allocates resources. She monitors performance based on information she can access and redirects resources accordingly. She may receive changes in direction or objectives from her superiors too. The organization isn’t working in isolation; the rest of the world impinges too—perhaps a pandemic restriction is announced. All of these will require decisions to be made.

The quality of decision making is intimately connected with how well the manager understands her organization and its environment. This understanding encompasses the groups and people under her, the organizational and broader context, and, also, herself.

The manager is, in effect, conducting model-based control, relying on her mental models. Figure 1 illustrates this concept, showing a manager’s mental models involved in her decision making.

Figure 1. A manager as a model-based controller of an organizational system (S), subject to disturbances from the outside world (W). The manager relies on her mental models to decide on actions to be taken. The “hat” symbols represent the fact that the mental models—of S, W, and the manager herself—are approximations of reality. (Source: [1], © IEEE)

Mapping control systems to managerial decision making enables several insights from the former to be applied to management. I note a few of these here (see also [1], [2]):

  • Feedback combined with “feedforward” actions: Feedback is essential for counteracting uncertainty, but it requires time to work—signals must travel around the control loop. Feedforward actions can improve response time, but require more accurate models to be most effective. Managers should combine the two, and the proportion of each depends, among other things, on the accuracy of mental models.
  • Effect of delays on managerial performance: Delays arise from various sources in organizations—e.g., decision making itself, the implementation of decisions, the time involved for decisions to be manifest in observable outputs, etc.—but regardless, they make the task of the manager harder. Reducing delays can dramatically improve performance, especially in feedback loops.
  • States versus outputs: Some organizational parameters are relatively easy to measure, but they may not provide direct information about what a manager really needs to know: the state of the organization. The distinction between outputs and states is crucial in control theory, and specialized techniques are used to derive states from measured signals. Similarly, managers need to be aware that they need to relate what information they receive (e.g., lines of code completed in the week) to what they need to know (e.g., the level of completion of the software deliverable).
  • Robustness-performance tradeoff: Other things being equal, there is an inherent tradeoff between the performance that can be achieved in a controlled system and its robustness to uncertainty. High performance is achieved at the cost of lower resilience to noisy data, model mismatch, and disturbances, and if a manager desires a maximally robust organization, performance under nominal conditions will likely suffer.

Control theory is the only rigorous approach for effective decision making in complex dynamical systems. Even if the mathematical machinery of control theory isn’t directly applicable to management, insights from control, leveraging the analogy between system models and mental models, can be helpful to managers for improving their decision making.

Digging Deeper

[1] T. Samad, “Control Systems — Concepts and Insights for Managerial Decision Making.” Proc. 2020 IEEE 10th Int. Conf. on Intelligent Systems, Varna, Bulgaria, 2020
[2] T. Samad et al., “Managerial Decision Making as an Application for Control Science and Engineering,” Proc. American Control Conf., Atlanta, U.S.A., 2022


About the Author

Tariq Samad holds the Honeywell/W.R. Sweatt Chair and is the director of graduate studies at the Technological Leadership Institute at the University of Minnesota. Prior to joining the Univ. of Minnesota, he was with Honeywell, retiring as Corporate Fellow. During his 30-year career with Honeywell, he contributed to and led automation and control technology developments for applications in electric power systems, clean energy, building management, the process industries, automotive engines, unmanned aircraft and advanced manufacturing. He is a past president of IEEE Control Systems Society and the American Automatic Control Council. Please contact him via https://cse.umn.edu/tli/tariq-samad-phd or his Email.

TEMS – 5 Focus Areas

Moving Product/Services from Idea to Market

Identifying and Implementing Successful Projects, and Systems

Integrating Technology for Capability and Productivity

Developing from Engineer to Leader

Balancing the Norms of Society, Government, and Regulators

Attend upcoming Conference

IEEE International Conference on Smart Mobility (IEEESM'23)

Join IEEE TEMS