Customer segmentation, or market segmentation, at a basic level, is the partitioning of a broad range of potential customers in a given market into specific subgroups of customers, where each of the subgroups contains customers that share certain similarities. The following diagram depicts the formal definition of customer segmentation where customers are identified into three groups:

Customer segmentation needs the organizations to gather data about customers and analyze it to identify patterns that can be used to determine subgroups. The segmentation of customers could be achieved through multiple data points related to customers. The following are some of the data points:

• Demographics: This data point includes race, ethnicity, age, gender, religion, level of education, income, life stage, marital status, occupation
• Psychographics: This data point includes lifestyle, values, socioeconomic standing, personality
• Behavioral: This data point includes product usage, loyalties, awareness, occasions, knowledge, liking, and purchase patterns

With billions of people in the world, efficiently making use of customer segmentation will help organizations narrow down the pool and reach only the people that mean something to their business, ultimately driving conversions and revenue. The following are some of the specific objectives that organizations attempt to achieve through identifying segments in their customers:

• Identifying higher-percentage opportunities that the sales team can pursue
• Identifying customer groups that have a higher interest in the product, and customize the product according to the needs of high-interest customers
• Developing very focused marketing messages to specific customer groups so as to drive higher-quality inbound interest in the product
• Choosing the best communication channel for various segments, which might be email, social media, radio, or another approach, depending on the segment
• Concentrating on the most profitable customers
• Upselling and cross-selling other products and services
• Test pricing options
• Identifying new product or service opportunities

When an organization needs to perform segmentation, it can typically look for common characteristics, such as shared needs, common interests, similar lifestyles, or even similar demographic profiles and come up with segments in customer data. Unfortunately, creating segments is not that simple. With the availability of big data, organizations now have hundreds of characteristics of customers they can look at in order to come up with segments. It is not feasible for a person or few people in an organization to go through hundreds of types of data, find relationships between each of them, and then establish segments based on several different values possible for each data point. That's where unsupervised ML, called clustering, comes to rescue.

Clustering is the mechanism of using ML algorithms to identify relationships in different types of data, thereby yielding new segments based on those relationships. Simply put, clustering finds the relationship between data points so they can be segmented.

The terms cluster analysis and customer segmentation are closely related and used interchangeably by ML practitioners. However, there is an important difference between these terms.

Clustering is a tool that helps organizations put together data based on similarities and statistical connections. Clustering is very helpful in guiding the development of suitable customer segments. It also provides useful statistical measures of the potential target customers. While the objective for an organization is to identify effective customer segments from data, simply applying a clustering technique on data and grouping the data in itself may or may not offer effective customer segments. This essentially means that the output obtained from clustering, that is, the clusters, need to be further analyzed to get insight into the meaning of each of the clusters, and then determine which clusters can be utilized for downstream activities, such as business promotions. The following is a flow diagram that helps us to understand the role of clustering in the customer-segmentation process:

Now that we understand that clustering forms a stepping stone to performing customer segmentation, in the rest of the chapter, we will discuss various clustering techniques and implement projects around these techniques to create customer segments. For our projects, we make use of wholesale customer dataset. Before delving into the projects, let's learn about the dataset and perform exploratory data analysis (EDA) to get a better understanding of the data.