“…Most segmentation algorithms have significant advantages and disadvantages, so it is challenging, even for marketing professionals, to consider the right technique or algorithm (Das & Nayak, 2022). Based on the surveys of Sari et al (2016), Cooil et al (2008), McKernan (2018), and Das and Nayak (2022), the current segmentation methods can be categorized as follows. - Ranking—methods that provide a full ranking of customers, versus methods that provide only the division of groups and segments without ranking.
- Updating—methods that are updated yearly or quarterly, versus dynamic methods that are updated frequently by demand, or continuously.
- Objectivity—subjective methods based on expert evaluation and salespeople (for example, AHP, Delphi), versus objective methods that are based on firmed past data, so everyone who uses the method will get the same result.
- Number of customers—methods suitable for classifying a relatively small number of customers with a small number of orders per year, versus methods that are appropriate to handle a high volume of data and an unlimited number of customers and orders.
- Technique—simple statistical methods such as descriptive analyses, versus more rigorous segmentation methods; these methods include k‐means clustering, decision trees, cluster analysis, cluster‐wise regression, multiple regression, linear programming, discrimination analysis, latent class structure, inductive learning techniques, soft computing techniques, Chi‐Square automatic interaction detector, and data mining. Techniques related to data mining include artificial neural networks, fuzzy logic, machine learning, and evolutionary methods.
- Forecasting—methods that can forecast the segment where a new customer will belong, versus methods that classify only customers with enough purchase history.
- Criteria—methods that classify customers according to a single criterion such as profit or purchasing volume, versus methods that classify customers according to multi‐criteria analysis (MCDA).
…”