It's in the Research
As you go through this site, keep in mind that I‘m not asking you to digest these ideas just because I say so. Rather, I'm asking you to thoughtfully consider them because they are derived from rigorous research. Built to Last required six years of research, conducted at Stanford University Graduate School of Business with my colleague and mentor Jerry Porras. Good to Great required five years of effort with 21 research associates at my management laboratory in Boulder, Colorado.
Turning mountains of data into useful concepts is an iterative process of looping back and forth, developing ideas and testing them against the data, revising the ideas, building a framework, seeing it break under the weight of evidence, and rebuilding it yet again. That process is repeated over and over, until everything hangs together in a coherent framework of concepts.
Allow me to make a few points regarding our research process:
Comparison Method: The cornerstone of our research method is the selection of a credible study set, the direct comparison of that set to a carefully selected control set, and the study of the contrasts between each set over a long period of history. The critical question is not “What do successes share in common?” or “What do failures share in common?” The critical question is “What do we learn by studying the contrast between success and failure?” Think of it this way: Suppose you wanted to study what makes gold medal winners in the Olympic Games. If you only studied the gold medal winners by themselves, you‘d find that they all had coaches. But you if looked at the athletes that made the Olympic team, but never won a medal, you‘d find that they also had coaches! The key question is, “What systemically distinguishes gold medal winners from those who never won a medal?”
As Jerry Porras and I first mentioned in Built to Last, our comparison method has proven to be the key for calling into question powerfully entrenched myths and discerning fundamental principles that apply over long stretches of time and across a wide range of circumstances.
Correlations, Not Causes: The variables we identify in our research are correlated with the performance patterns we study, and we cannot claim a definitive causal relationship. If we could conduct double-blind, prospective, randomized, placebo-controlled trials, we would be able to create a predictive model of corporate performance. But such experiments simply do not exist in the real world of management, and therefore it‘s impossible to claim cause and effect with 100-percent certainty. That said, our contrast method does give us greater confidence in our findings than if we studied only success, or only failure.
Historical Analysis: We employ a historical method, studying each company from its founding up to the end point of our investigation, focusing on specific eras of performance. We gather a range of historical materials, such as financial and annual reports, major articles published on the company, books, academic case studies, analyst reports, and industry reference materials. This is important because drawing solely upon backward-looking commentary or retrospective interviews increases the chances of fallacious conclusions.
Using a well-known success story to illustrate, some retrospective accounts attribute Southwest Airlines success to pioneering a unique and innovative airline model (in part because the authors writing today believe the winners must be the innovators); but in fact, a careful reading of historical documents shows that Southwest largely copied its model from Pacific Southwest Airlines in the late 1960s. If we were to rely on only retrospective accounts, we would be led astray about why Southwest became a great company.
We therefore derive our frameworks primarily from evidence from the actual time of the events, before the outcome is known, and we read through the evidence in chronological order, moving forward through time. Documents published at each point in time are written without foreknowledge of the company‘s eventual success or failure, and thereby avoid the bias of knowing the outcome. So, for instance, the materials we have on Zenith that were published in the early 1960s, when Zenith sat on top of its world, give us perspective on Zenith at that time, uncolored by the fact that Zenith would eventually fall. Where they are included at all, backward-looking interviews play a minimal part in our research method.
Not that historical information is perfect—corporations can selectively exclude unhappy information from their annual reports, for example, and journalists may write with a preconceived point of view. Nor am I entirely immune from having some retrospective bias of my own, as I always know the success or failure of the company I‘m studying, and I cannot erase that from my brain.
But even with these limitations, our comparative historical method helps us see more clearly the factors correlated with the rise and fall of great companies. Yet always keep in mind: I offer the ideas on this site and in all my work for your thoughtful consideration, not blind acceptance. You‘re the judge and jury. Let the evidence speak.
- Jim Collins