One of the primary benefits of modeling with systems tools—whether causal loops or computer simulations—is the intense discussion it can generate around important variables and how they interrelate. Although managing this discussion can go smoothly, the process can also easily get bogged down at this stage. This is because modeling, especially computer modeling, requires explicit, logical statements of assumed cause-and-effect. If the variables, or how they relate to each other, are unclear, the process will stall. One factor which often slows down discussion—if it doesn’t derail it outright—is a lack of differentiation between actual, perceived, and desired variables.
Consider, for example, the difference between actual customer satisfaction, perceived customer satisfaction, and desired customer satisfaction. It is not uncommon for managers engaged in a modeling effort to use a single variable—”customer satisfaction”—to represent all three ideas. In part, this tendency simply reflects the fact that people often use the same word to mean different things, particularly when coming from different areas of the business. And one can argue that the benefit of the modeling process is that it provides an opportunity for these conflicts to surface and clarify areas of ambiguity. But from a process point of view, such multiple interpretations can bring the model-building effort to a grinding halt.
Modeling, especially computer modeling, requires explicit, logical statements of assumed cause-and-effect. If the variables, or how they relate to each other, are unclear, the process will stall.
Suppose a manager identifies a variable, such as expected delivery time, that she believes influences (perceived) customer satisfaction. Someone else on the team might respond to that statement by asking incredulously, “How on earth does expected delivery time affect the customer’s (actual) satisfaction once he or she receives the product?” Unless both people recognize that they are discussing two different variables, they could waste a lot of time and effort talking past each other. A team that has gotten bogged down in multiple interpretations of a single variable might want to use the following framework to clarify the distinction between the actual, perceived, and desired state of a particular variable.
“Actual” variables represent the actual system state. Whenever this distinction is raised, at least one philosopher on the team will ask whether the actual system state is ever truly knowable. While this is a valid issue, it is important to re-member that a model is simply a representation of our assumptions about a system, not a search for “the truth.” Because most of us do carry assumptions in our heads about how certain forces influence an actual state variable, we need to share these assumptions by distinguishing between the actual variable and the perceived (or desired) variable, even if the actual value can never be known.
At some level, the philosophers are right—we will never know the actual system state. All we have is our perception of what that state is. Although we can try to measure that variable, the perceived (measured) variable will differ from the underlying (actual) system state for at least two reasons: measurement error and measurement interaction with the system.
Our perception of the system state will always be limited by measurement error. Just as there is no perfect meter for measuring dimensions such as distance and time, there is no tool for accurately gauging an attitudinal variable such as customer satisfaction. Therefore, we can always expect perceptions to be different from actual system conditions.
What’s more, by trying to measure actual system conditions, we often aggravate the difference between perceived and actual values. This is because inserting a measurement process into the system adds additional structure to the system. And, as we all know, changing system structure changes the system behavior. By attempting to measure the system, we create new, often unintended feedback structures that alter the overall system. And unless we are learning about the system state faster than we are changing the system, the gap between our perceptions and the actual system conditions will remain and perhaps will even grow. Because of this inherent difference between the actual system condition and our perception of it, both of these variables should be included in a systems map.
Many balancing feedback processes in a system work to reduce the difference between a variable’s actual state and its desired state. In some cases, the desired variable is an explicit goal of a system agent (manager or other decision-maker). In other cases, the desired goal is set implicitly by the structure of a subsystem (as occurs when the growth of a population is limited by the available resources). Although the difference between a desired and actual state usually is clearer than the difference between actual and perceived, groups can still get stalled if they don’t make that distinction explicit on the causal map.
Making the Distinction
To understand how these distinctions might be used to clarify an important issue in a group, let’s look at the example of a corporate staffing department that has been caught in a roller coaster of hiring frenzies, followed by hiring freezes. When department personnel looked at the issue from a systemic viewpoint, they recognized that the problem stemmed from a lack of clarity throughout the company about the relationships between the company’s desired capacity, its actual capacity, and its perceived capacity.
Departments throughout the company made requests for additional personnel when they perceived a shortfall in capacity. But because each new hire was required to follow a three-month corporate training program, departments whose requests had already been acted upon by the staffing department still experienced understaffing and hence issued additional hiring requests. Once trainees came through the pipeline, perceived capacity quickly reached and exceeded actual capacity, as department planners tried to allocate the excess capacity they had unwittingly created (see “Capacity Distinctions”). By recognizing these differences between actual, perceived, and desired capacity—and how the three-month delay exacerbated the gap between perceived and actual capacity—the department was able to change the way it handled staff requests in order to reduce the staffing fluctuations.
In this company, a gap between perceived capacity and desired capacity was dosed by hiring new personnel. However, the three-month training delay for new hires led to a perception of chronic under-capacity throughout the company, which prompted continued hiring requests. Over time, this created a cycle of hiring frenzies, followed by hiring freezes.
Making the distinction between desired and actual states of a variable can also help identify gaps in a company’s performance. For example, a team of managers at a large service provider was discussing how the company’s various core competencies affected its ability to do business. Among the attributes was “relationship management”—how well the company’s representatives handled its key customer accounts. The team agreed on the simple linear explanation that relationship management skills affected customer satisfaction, which led to more business. But they began to get bogged down when they tried to address the implications of these relationships for the company’s strategic direction: What is the current level of relationship management skills in the company? How much investment in relationship management will be enough? To address these questions, they needed to break down the term “relationship management” into three components—their perceived expertise in this area, their desired expertise, and the actual (or current) expertise.
By breaking the variable down, the team was able to wrestle with a difficult issue that had not surfaced before: the gap between the company’s desired level of relationship management skills and its actual capacity. In fact, the company’s relationship with its largest customer was in serious trouble. Once they had raised this “undiscussable” issue, the managers were able to take action to save this relationship before the trouble reached crisis proportions.
As both of these examples illustrate, being aware of the distinction between actual, perceived, and desired variables—and looking for appropriate opportunities to clarify those terms—provides a level of precision that often sheds light on areas of inconsistencies, disagreements, or “undiscussable” issues in a company. Once a team achieves this level of clarity, it can then move into a more informed discussion of how to address the problems at hand.
Gregory Hennessy is an associate at GKA Incorporated. He has worked in a number of planning roles in both the energy and telecommunications Industries and was a strategy consultant with Monitor Company.
Jorge Rufat-Latre Is founder of JRL Learning Systems (Dallas. TX) where he works with clients to surface assumptions, build community practice conversational tools and build simulations of group mental models.
Editorial support for this article was provided by Colleen Lannon.