This is the second posting in a four-part series on business insights.
My first posting in this series introduced the reader to a recent article on extracting business insights from data, presented their definition of an insight, and laid out some of my thoughts on the importance of business insights. As a reminder, this series of posts will not be summarizing the entire article (you should read it for yourself) but will extract and expand upon specific points I find most interesting or compelling.
Let me restate the definition of an insight here as a starting point for part two of this series.
A thought, fact, combination of facts, data and/or analysis of data that induces meaning and furthers understanding of a situation or issue that has the potential of benefiting the business or re-directing the thinking about that situation or issue which then in turn has the potential of benefiting the business.
Embedded in this definition is the assumption of two different types of insights.
- Insights that support a specific decision or provide additional understanding of a specific situation or issue and will directly benefit the business.
- Insights that enhance market understanding or strategy definition without directly impacting a specific decision.
The article spends only a few lines defining and discussing these two types, but I think there are important differences in how these two types of insights are produced.
Business insights of type #1 should be derived using a deductive method. You begin with a specific issue or question; find a model or framework (aka, theory) that addresses the question; determine the possible answers (aka, hypotheses); use data to find the “right” answers; and develop a recommended solution to the original question.
Here’s a common example. When a company wants to reduce customer attrition and build customer loyalty we apply a standard empirical model of customer loyalty (experiences -> attitudes -> loyalty -> behavior); collect data and validate each relationship (hypothesis) proposed by the model; and then create priorities and recommendations.
The second type of insight is the result of an inductive, exploratory research process. This process usually begins with a fuzzy question, or an issue that doesn’t have well-developed theories or frameworks. The goal is to learn something new or provide better definition of an issue by observing specific instances or data; looking for patterns; developing hypotheses based on those patterns; and looping back through the process until you can develop meaningful, supportable conclusions.
The conclusions or insights derived from this process should impact a company’s strategies. But this type of research is difficult and messy, and the observations-patterns-hypotheses cycle may not produce meaningful insights that impact the business. Consequently, businesses often avoid this research, but they do so at their own peril.
A vast majority of business insights fall into the first category, and rightfully so – businesses have to maintain focus, not “boil the ocean.” However, most business professionals and customer researchers do not conduct this type of research correctly. We often ignore the steps of applying a framework or model to an issue and using it to generate testable hypotheses unless it is a standard, rote issue. The reason so many consultants have jobs is because good ones excel at exactly this part of the process.
In practice, many people take a deductive process and replace the second step with an inductive process. This can still result in useful insights, but in a very inefficient manner that often decreases the chances of finding insights that will be properly utilized by the business. That’s because the two approaches require different questions, assumptions, and tools; and we need to know which insights we are seeking in order to structure the research project appropriately. The article I am referencing in this blog series does not provide details on the best ways to approach these two different processes, but you can find more details in this book chapter (see pp. 7-16).
A majority of business insights will, and should, come from deductive processes. However, the true “eureka” moments, the insights that define new and successful strategies, the analyses that uncover undervalued customer segments or new product niches, will most often come from inductive insights.
If you avoid inductive research, or do it poorly, you are at risk of the business management version of a “Minsky moment” – if certain theories, frameworks and strategies are successful, competitors will adopt them and your competitive advantage will collapse. Deductive insights will keep your company alive, but inductive insights are necessary if you want to thrive – just ask Google.
Troy Powell, Ph.D.
Vice President, Statistical Solutions