Segmentation is one of the most frequently used words in marketing but actually refers to many things. The background photo above is one popular way of thinking about segmentation; this article looks at it from a somewhat different perspective.

At its most fundamental level, it means categorizing objects. The “objects” are often people – customers, shoppers, general consumers – but not necessarily. For example, we can segment companies by industry type, country of origin, for-profit or nonprofit, Business-to-Business (B2B) or Business-to-Consumer (B2C), as well as by size. We can segment products into categories, brands into sub-categories, and brands based on their users or image.

Segmentations also differ according to their purpose. While often associated with targeting, segmentation is often used to help us to better understand a market or identify a potential market. Regarding targeting, personalized marketing or hypertargeting may come to mind. Recommender systems are used for targeting products, services, news content, people we may wish to connect with on LinkedIn, and many other ways.

Targeting can be longer-term and strategic too, for example, aimed at people who shop a category in certain ways or use a product or service for particular reasons. These types of segmentations, which usually have a small number of groups (e.g., 5-6), are typical in marketing research. They are utilized in strategic marketing and normally re-benchmarked every 3-5 years.

Segmentations differ by the kinds of data used, as well. Demographics have been used for years in direct marketing. Strategic segmentations often make use of consumer surveys, whereas hypertargeting and recommender systems draw heavily upon behavioral data​: think Amazon. Data from different sources can also be combined in most types of segmentation.

Method is another way to segment segmentations. In supervised segmentation (aka a priori segmentation) segments are pre-defined, for example based on recency, frequency and monetary value (RFM) or ever purchasers and never purchasers. Various statistical and machine learning techniques are employed to develop predictive models for targeting purposes and/or to better understand the ways important consumer groups differ.

Unsupervised segmentation (aka post hoc segmentation) reverses this process. Here, the segments are discovered, usually with a cluster analysis method. Many kinds of data can be used and merged as needed. Predictive models known as typing tools in marketing research are frequently used to classify respondents in future waves of tracking studies into their “nearest” segment, new customers in a data base, and in various other ways.

This has been merely a snapshot of a topic that could easily fill a large textbook, but I hope you’ve found it interesting and helpful.

 
Bio: Kevin Gray is President of Cannon Gray, a marketing science and analytics consultancy. He has more than 30 years’ experience in marketing research with Nielsen, Kantar, McCann and TIAA-CREF.

Original. Reposted with permission.

Related:

• Where is Marketing Data Science Headed?
• Customer Segmentation Using K Means Clustering
• Top Data and Analytics Trends