Why the Best Ideas Fail

The opening discussion comes from Peter Senge’s account of organizational learning. Its workplace examples and time references belong to that account.

One thing all managers know is that many of the best ideas never get put into practice. Brilliant strategies fail to get translated into action. Systemic insights never find their way into operating policies. A pilot experiment may persuade everyone that a new approach improves results, yet the organization never adopts it widely.

Mental models help explain this “slip ’twixt cup and lip”. A new insight may conflict with a deeply held picture of how the world works, leaving familiar habits intact even when people intend to change. Bringing those assumptions into view allows an organization to test and revise them.

A mental model is a framework that a person carries in their mind to interpret concepts, explain occurrences, and predict how the world around them works. The concept can be traced back to Kenneth Craik’s research in 1943, which proposed that the human mind constructs a small-scale model of reality to anticipate events. None of us can carry an entire organization, family, or community in our heads. What we carry are images, assumptions, and stories. These models can be simple generalizations such as “people are untrustworthy,” or they can be complex theories about why family members interact as they do, or why an industry succeeds. Philosophers have discussed mental models for centuries, going back at least to Plato’s parable of the cave. “The Emperor’s New Clothes” is a classic story, not about fatuous people, but about people bound by mental models. Their image of the monarch’s dignity kept them from seeing his naked figure as it was.

In surveying the accomplishments of cognitive science, Howard Gardner writes that the major accomplishment has been the clear demonstration of a level of mental representation active in diverse aspects of human behavior. Our mental models determine not only how we make sense of the world, but how we take action. Harvard’s Chris Argyris, who has worked with mental models and organizational learning for forty years, puts it this way: although people do not always behave congruently with their espoused theories (what they say), they do behave congruently with their theories-in-use (their mental models).

Why are mental models so powerful in affecting what we do? In part, because they affect what we see. Two people with different mental models can observe the same event and describe it differently, because they’ve looked at different details and made different interpretations. When you and I walk into a crowded party, we both take in the same basic sensory data, but we pick out different faces. As psychologists say, we observe selectively. This is no less true for supposedly “objective” observers such as scientists than for people in general. As Albert Einstein once wrote, “Our theories determine what we measure.” For years, physicists ran experiments that contradicted classical physics, yet no one “saw” the data that these experiments eventually provided, leading to the revolutionary theories—quantum mechanics and relativity—of twentieth-century physics.

The way mental models shape our perceptions is no less important in management. I will never forget visiting with a group of Detroit auto executives after their first factory visits to Japan over twenty years ago. This was right around the time that US automakers were finally waking up to the fact that Japan was steadily gaining market and profit share in their industry—and it might be attributed to how they managed, not just because they had “cheap” labor or protected home markets. A little way into the conversation, it was clear that the Detroit executives were unimpressed. I asked why and one said, “They didn’t show us real plants.” When I inquired as to what he meant by this, he responded, “There were no inventories in any of the plants. I’ve been in manufacturing operations for almost thirty years and I can tell you those were not real plants. They had clearly been staged for our tour.” Today we all know that they were indeed real plants, examples of the “Just-in-time” inventory systems that the Japanese had been working on for many years that dramatically reduced the need for in-process inventories throughout the manufacturing system. Within a few years these same US firms would be racing desperately to catch up to these manufacturing innovations … but that evening the Detroit execs had not seen anything to alarm them.

Or consider the customer belief that dominated the Big Three of Detroit for decades (and many feel still does)—that American consumers mostly cared about styling. According to management consultant Ian Mitroff, these beliefs about styling were part of a pervasive set of long-standing and unquestioned assumptions for success at General Motors:

• GM is in the business of making money, not cars.
• Cars are primarily status symbols. Styling is therefore more important than quality.
• The American car market is isolated from the rest of the world.
• Workers do not have an important impact on productivity or product quality.
• Everyone connected with the system has no need for more than a fragmented, compartmentalized understanding of the business.

The point here is not that GM or the Detroit industry in general were wrong. As Mitroff points out, these principles had served the industry well for many years. But the industry treated these principles as “a magic formula for success for all time, when all it had found was a particular set of conditions … that were good for a limited time.”

All models simplify reality. A particular problem arises when their assumptions remain implicit and escape examination. The Detroit automakers didn’t say, “We have a mental model that all people care about is styling.” They said, “All people care about is styling.” Because we remain unaware of our mental models, the models remain unexamined. Because they are unexamined, the models remain unchanged. As the world changes, the gap widens between our mental models and reality, leading to increasingly counterproductive actions.

As the Detroit automakers demonstrated, entire industries can develop chronic misfits between mental models and reality. In some ways, close-knit industries are especially vulnerable because all the member companies look to each other for standards of best practice.

Failure to appreciate mental models has undermined many efforts to foster systems thinking. In a classic study many years ago, a leading American industrial goods manufacturer—the largest in its industry—found itself steadily losing market share. Hoping to analyze their situation, top executives sought help from an MIT team of system dynamics specialists. Based on computer models, the team concluded that the firm’s problems stemmed from the way its executives managed inventories and production. Because it cost so much to store its bulky, expensive products, production managers held inventories as low as possible and aggressively cut back production whenever orders turned down. The result was unreliable and slow delivery, even when production capacity was adequate. In fact, the team’s computer simulations predicted that deliveries would lag further during business downturns than during booms—a prediction which ran counter to conventional wisdom, but which turned out to be true.

Impressed, the firm’s top executives put into effect a new policy based on the analysts’ recommendations. From now on, when orders fell, they would maintain production rates and try to improve delivery performance. During a modest business downturn the following year, the experiment worked beyond anyone’s expectations; thanks to prompter deliveries and more repeat buying from satisfied customers, the firm’s market share increased during the downturn. The managers were so pleased that they set up their own internal systems group. But the new policies were never taken to heart, and the improvement proved temporary. During the ensuing business recovery, the managers stopped worrying about delivery service. Four years later, when a more severe recession hit, the company went back to their original policy of dramatic production cutbacks.

Why discard such a successful experiment? The reason was the mental models deeply embedded in the firm’s management traditions. Every production manager knew that there was no more surefire way to destroy one’s career than to be held responsible for stockpiling unsold goods in the warehouse. Generations of top management had preached the gospel of commitment to inventory control. Despite the new experiment, the old mental model was still alive and well.

The inertia of deeply entrenched mental models can overwhelm even the best systemic insights. This has been a bitter lesson for many a purveyor of new management tools, not only for systems thinking advocates. But if mental models can impede learning—freezing companies and industries in outmoded practices—they can also help accelerate learning, if we learn to surface and examine them.

One way to develop a model deliberately is to identify a system’s components, map their relationships, and test the resulting explanation against observations. Models also develop informally through experience. Declarative knowledge concerns facts and relationships we can describe; procedural knowledge concerns how to perform tasks, such as riding a bicycle. Practice can make performance more automatic. Both kinds of knowledge can inform a model, but collecting more facts does not by itself make the model more accurate.

Experience can help a person recognize relationships that a beginner misses and distinguish cases that initially seem alike. Concepts from other disciplines may offer useful comparisons, provided their assumptions fit the problem. A broad model can connect several situations; a narrow model can be more precise within its intended setting. Usefulness depends on what the model helps explain or predict and how well it survives testing.

Mental models shape ordinary expectations as well as technical judgments. One person may imagine an ideal home in a busy metropolitan area, while a partner pictures a quiet community with grass between houses and a single stop sign. A diner may expect local produce and moderate prices, while a friend expects fine wines and prime cuts of meat. Such differences often become visible only when people try to make a decision together.

In the workplace, mental models shape dynamics just as powerfully. An employee might think, “My supervisor never listens—she’ll ask for my opinion and then do exactly what she planned anyway.” The supervisor, in turn, might think, “My employees don’t care about their work; only the weekend motivates them.” Both perspectives are mental models that, left unexamined, reinforce a negative pattern of behavior that harms the organization.

An unexamined model can make our own interpretation seem like the only possible one. Identifying its assumptions helps us investigate both failures and successes: what did we expect, what happened, and what explains the difference? This form of self-examination fits the maxim of “know thyself.”

Some modern spiritual interpretations use alchemy—the transformation of lead into gold—as a metaphor for changing one’s experience through thought. Claims about the subconscious connecting people to a greater life-force belong to that spiritual framework; they do not establish how mental models function. The practical question here is how to identify an assumption and examine its limits.

To examine a mental model, state an assumption, identify what it predicts, and compare that prediction with what happens. In an organization, this also means asking whether routines and incentives support the revised understanding. A successful pilot will have little lasting effect if the old assumptions still govern everyday decisions.