One of the critical issues that today's online firms face is to make sense of all the available data about their customers and to offer them customised and personalised services with affective features. There are numerous clustering methodologies that can help companies identify homogeneous groups of people among their potential customers so that they can design such services for each homogenous group. Because firms do not have prior external knowledge about the true clusters of their potential customers, deciding which clustering method to use becomes extremely challenging. This paper compared two most popular algorithms including k-means and fuzzy c-means clustering methodologies. The results showed that compared to fuzzy c-means clustering k-means clustering yielded an imprecise categorisation of as much as 72% of the potential shoppers of an online shopping service. Moreover, the results showed that compared to k-means clustering, fuzzy c-means clustering led to better cluster solutions based on multiple criteria. The paper shows how the results can help online businesses design their online offerings with effective features.