Nowadays, a lot of attention in e-commerce is paid to improving user experience. Due to a high competition on the market, e-commerce websites must provide services focusing on usability and quality of service. For this purpose they can detect possible problems by using data which represents customer behavior while navigating the website, described by the sequence of actions performed by a customer on the portal. The goal of this article is to propose an approach to apply process mining techniques for mining website logs, to discover user paths and patterns often seen in a website. Patterns retrieved from user's most frequent browsing behavior are then utilized to analyze usability issues. The paper presents a general model for improving an e-commerce website based on the application of process mining techniques. The findings of the article showcase that it is possible to analyze and improve a website based on results achieved by applying process mining techniques on the web logs the site produces. The usefulness of provided model is proven on logs from a Polish e-commerce portal.