Four Reasons to Analyze Customer Behavior
How enterprises can engage, retain, and strengthen bonds with their customers.
- By Fern Halper, Ph.D.
- August 25, 2015
TDWI research consistently reveals that a major driver for analytics, and especially advanced analytics such as predictive analytics, is to deepen an understanding of customer behavior. In fact, marketing and sales are often the first departments to start using advanced analytics. These groups want to do more than understand and gain insight into customer behavior. They want to engage, retain, and strengthen bonds with these customers.
Gain insight: Clearly, one of the main reasons organizations analyze data is to gain insight. Exploring your data for insights about customer behavior may involve segmenting your customer base, which often uses cluster analysis, a technique that organizes a set of observations into two or more groups that are mutually exclusive based on combinations of variables.
Typically, organizations do discovery and segmentation analysis using structured data. However, unstructured text data, such as social media data or internal text data, can also provide great insight into customer sentiment and behavior. More often, organizations are performing social media analytics, such as voice-of-the-customer analysis, using text analytics technology to gain insight about what customers are saying and how their brand resonates with existing and potential customers. TDWI sees increasing interest in these technologies.
Attract and engage: If you’ve segmented your customer base, you can target customers and engage them because you have a better sense of what they might be interested in. For instance, an organization wants to make customers the right offer when it launches a product campaign across various channels (online, e-mail, mobile, in-store, etc.). By analyzing historical purchases and profiles, companies can predict the likelihood, or propensity, of future activity at a customer level. For instance, a company might use a propensity model and past purchase behavior to gauge the probability of a customer making a certain purchase. This data can be used when developing the new campaign.
Improve retention: Customer retention is a key marketing activity, especially when it comes to profitable customers, and predictive analytics can be extremely helpful. For instance, decision trees can be useful where there are discrete target or outcome variables of interest (leave or stay, for example). Typically, a set of historical training data is provided to the predictive analytics algorithm. The data might consist of different kinds of information about customers (demographics, purchase history, even past sentiment) and it is used by the decision-tree algorithm to determine decision rules that describe the relationship between the input and outcome variables. These rules can be used against new data where the outcome is not known (for instance, leave or stay). These models are often operationalized -- for instance, in a call-center where agents can use them to try to retain customers at risk.
Strengthen bonds with customers: Organizations want to continually strengthen relationships with their existing customers while attracting new ones. Customer lifetime value models help organizations understand the future worth of customers and segments. It is an important part of a customer strategy. Techniques like affinity analysis using market basket analysis to understand combinations of products bought together can be very useful in driving e-mail marketing and recommendation engines.
Of course, I’m only scratching the surface here, and many of these techniques can be used in multiple situations. If you want to learn more about how successful organizations are using customer analytics, think about attending the TDWI Executive Summit in San Diego September 21–23! We’ll be talking specifically about customer analytics and the data and infrastructure to support these analytics.