The objectives of this research are to identify customer purchase behavior, obtain the number of customer clusters, and form customer profile in order to find situation-based customer purchase be- havior pattern. From the given data, the RFM method and K-Means clustering are used to identify the customer purchase behavior and profiles. The result of this research showed that the customer clusters are formed differently in every product category based on the RFM value and K-Means clustering. There are also differences in peak hour for each customer cluster. The best time to deliver notifica- tions and personal messages is near the peak hour. Indeed, this matter is useful to create contextual marketing and targeted advertising that is designed based on customer cluster and purchase behavior.