Customer segmentation is considered the cornerstone for personalisation, target advertising, and promotion assisting both researchers and practitioners to enhance customers’ buying behaviour understanding. Pertinent literature mainly exploits one distinct segmentation type such as behavioural to segment customers solely under one lens. We develop a two-stage business analytics approach that introduces a combination of geographic and behavioural customer segmentation. Our approach is based on data mining and machine learning techniques. We evaluate the suggested approach using e-commerce home delivery data. First, we segment customers based on the products ordered to identify behavioural customer segments with similar product preferences. Then, we perform geographic segmentation. By applying the approach developed we also identify challenges that affect the segmentation process and results. The suggested approach can serve as a guide to business analysts to understand which are the steps that they should perform when analysing similar datasets. Whereas its results may assist third-party logistics (3PL) companies, retailers, and brands in supporting decision making.