Developers of e-commerce applications are often unrealistic about how their site will be used, and about possible outcomes during site usage. The most commonly considered outcomes of a user's visit to a site are firstly that the visit culminates in a sale, and secondly that the user leaves the site without buying anything -- perhaps to return later. In the second case sites often "remember" any accumulated items so that a shopper can return at a later stage to resume shopping. In this paper we consider certain disruptions, such as breakdowns, problems caused by human errors, and interruptions, which could affect the outcome of the e-commerce shopping experience. These events have definite and possibly long-lasting effects on users, and applications should therefore be developed to cater for these eventualities so as to enhance the usability of the site and encourage further usage. We develop a model for analysing e-commerce application usage, and using this model, propo...