Wednesday, February 13, 2013

Data can help us, even outside the web


Today, nearly every dimension of personal and professional life will produce, interact, and sometimes be drowned by data. With the ever-increasing amount of digital devices at hand, huge stores of data are now automatically recorded, providing detailed code-bits of information that can illustrate at fundamental levels the complex array of activity people engage with every day. This information is hugely valuable, yet our cognitive minds are poor at deciphering data to recognize relevant patterns that can provide insight to improving our activities. The use of customer analytics and data mining processes can provide insight into how customer behavior is affected by particular sets of decisions. Data mining is a process that extracts previously unknown, interesting, valid, and actionable data patterns from a large set of data for supporting decisions and providing Business Intelligence (BI). This data can be analyzed to make key decisions in customer behavior, site selection, customer relationship management, and even something apparently unrelated such as physical retail design. But how can customer analytics and digital data translate into a physical world and aid us in something such as a retail store design? In order to do so we much first understand what problem we are trying to solve and identify what data can augment our understanding of a particular business scenario. What we find is that the way in which a scenario plays out is often different than the subjective observations we make.
            Most retail stores use a basic principle of association patterns to locate items of relationship in close proximity to each other in order for the customer to identify related items that they will likely also purchase. For example, furniture shopping is largely a contextual experience. That is, one envisions the environment that a piece will fit into and shop based upon not only the piece itself, but how it complements other pieces in a setting. For this reason, the purchase of furniture will often lead to the purchasing of other pieces that go in a set. An obvious example is the purchase of a bed; customers will often consider case pieces such as dressers, nightstands, and armoires. Therefore, strategic settings of furniture provide customers with examples of complementary items that can provide them with a vision of how a room will look and feel (this envisioning is often difficult for customers to do themselves). Using insights from data mining, we can apply the principles of association patterns analysis to determine which items were often purchased together, and provide insight into the most advantageous methods for store layout.
            But how can customer analytics help in this arrangement? Many companies that specialize in luxury goods maintain a robust historical database of computerized customer account information (such as a SQL software application). For companies even operating outside of the web, this data can provide us with the first preliminary steps in our analysis. For the furniture retail store, data can still provide insight into furniture categories (bedroom, office, kitchen, etc.). By mining the data for association patters, we can determine the most frequently purchased items that occur with individual customers. This will have to be evaluated over time as people rarely purchase all their home furniture in one transaction. Therefore, queries into the data should the customer accounts within certain time frames (3-12 months). This will yield scenarios that are more likely a result of customers experience on the retail floor rather than situations where customers returned a year or more later to purchase an unrelated item. By analyzing the correlation among pieces purchased, the company can utilize the patterns that show high-correlation to improve the design of the store layout. By repeated analysis, the company can analyze the effectiveness of a particular layout and refine their strategies to locate more and different patterns.
            This is only one dimension that customer analytics can be enhanced by data mining processes. But the analysis should reveal how effective data can be in providing an objective window into patters of customer behavior if the right questions are asked and the right data acquired.

Eric Siegel, Predictive Analytics with Data Mining:
How It Works, 2005

No comments:

Post a Comment