Best Practices: Using Structural Equation Modeling (SEM) to Drive Better Business Outcomes

SEM provides more powerful insights to strengthen brand health.

Many businesses establish and track top-line metrics or dashboards to measure business health. While this a valuable tool for monitoring progress, perhaps more important is identifying the factors that will most influence and improve those key metrics. To date, this has been accomplished primarily by using simple driver analysis methods. Unfortunately, these methods tend to pale in comparison to the power of modern modeling tools and, in some cases, can incorrectly state the impact of the drivers.

As an alternative, progressive businesses are implementing more insightful driver models to:

·       Improve brand equity
·       Change purchase/spending behaviors
·       Grow product usage and customer engagement
·       Boost advocacy
·       Create broader concepts of loyalty

These models make leaders at all levels more effective by better prioritizing strategic investments, elements of the customer experience, and positioning or messaging ideas.

Structured Equation Modeling (SEM), in combination with Path Models, is one of the most powerful techniques for developing driver insights because of its unique ability to identify and focus on the big levers that most significantly impact top-line business outcomes.

Creating the Driver Model Structure and Identifying the Big Levers

SEM creates a structure of attributes that drive major concepts (often called the big levers). There are often multiple layers to these relationships, which is immensely insightful to understand as you assess a broader range of issues related to the customer experience — such as emotional elements, brand image, and purchase process attributes, along with more standard perceptions and behaviors. This model structure is typically depicted in a Path Model diagram that helps visualize and explain the impact. As the structure is changed, SEM computes the driver impact scores for all of the items at different levels. In doing so, it enhances your view of the world by providing guidance on how to best define your big levers and by creating an optimal, more accurate model that clarifies the relationships between attributes you can influence and critical business outcomes.

Achieving a Deeper, More Accurate Understanding of the Impact Relationships

SEM identifies relationships among all of the big levers and attributes, and measures their impact on each other, whereas other techniques only consider the impact that attributes have on a business outcome (dependent variable) or a single idea. This difference is critical. For example, product reliability and value can each have an impact on purchase — and value might have a higher direct impact on purchase than reliability. However, SEM can identify if reliability also impacts value and how much. SEM would correctly show the total effect that reliability has on purchase by measuring reliability’s direct impact on purchase plus the indirect impact it has, through the path of impacting value, which in turn further impacts purchase.  In the end, SEM would correctly answer whether investing in reliability is more effective than prioritizing value, while other methods might miss this indirect and total impact.

Delivering More Insight and Actionability

SEM delivers a full complement of actionable outputs to drive business decisions. By comparing your performance on attributes versus their impact, you can better prioritize where to invest. These dimensions are typically shown in a quadrant map or XY scatter plot. The drivers for your brand and your performance can be compared to those of your competitors to prioritize strategic moves or refine your messaging or positioning to effectively differentiate your brand or product offers.

Radius has deep experience applying SEM in many business situations, across a range of industries. In fact, this experience spans the entire class of methods, including Latent Class Regression, Partial Least Squares (PLS) and Ridge Regression. The result is an ability to help you apply the best tools for your needs.

Please contact us if you’d like to learn more about these tools and how they might best work for you.