Purpose – This study explores the role of construal levels in predicting online consumer behavior on a retail website. It builds on the conceptualization that simply browsing a website and making actual purchases can be an outcome of how abstractly or concretely one thinks about that experience. This study examines the differential effects of intermediary websites’ attributes and seller’s product offerings in predicting frequency of visits and actual purchases. Research design, data, and methodology – Data were collected from 188 undergraduate students in a large university of Korea. Hierarchical regression model was utilized to test the proposed effect of website characteristics and seller attributes on visit and purchase. Results – We propose and find that online shopping website visits and purchase frequency have different antecedents. The results reveal that website visit frequency and purchase have different predictors and this can be explained through construal level theory. Specifically, we find purchase frequency is predicted more by website image and financial benefits can be more predictive in actual purchases. Conclusions – Consumer behavior on the internet can be delineated into website visits and actual purchases. First, uplifting the image of the website itself is much more important than just making offerings cheaper. Online shopping website should try to match its features to mental representations that customers go through from just visit (abstract) to purchase (concrete).