Consumers in e-commerce acquire information through search engines, yet to date there has been little empirical study on how users interact with the results produced by search engines. This is analogous to, but different from, the ever-expanding research on clickstreams, where users interact with static web pages. We propose a new network approach to analyzing search engine server log data. We call this searchstream data. We create graph representations based on the web pages that users traverse as they explore the search results that their use of search engines generates. We then analyze the graph-level properties of these search network graphs by conducting cluster analysis. We report preliminary evidence the presence of heterogeneity among users in terms of how they interact with search engines. This suggests that search engine users may not all benefit from the same functionality in the search engines they rely upon. We also offer additional evidence on the empirical regularities associated with a variety of relevant issues that arise in the business-to-business (B2B) e-market context that we have studied.