Modeling the Clickstream: Implications for Web-Based Advertising Efforts

• Published in: Marketing science (Providence, R.I.) R.I.), 2003-10, Vol.22 (4), p.520-541
• Main Authors: Chatterjee, Patrali, Hoffman, Donna L, Novak, Thomas P
• Format: Article
• Language: English
• Subjects: Advertising and Media Research; Advertising expenditures; Advertising research; Advertising to sales ratios; Clickstream Data; Coefficients; Computer-Mediated Environments; Consumer advertising; Consumer behavior; Consumers; Internet; Internet/Web advertising; Intersession; Marketing; Mathematical models; Modeling; Online advertising; Online Consumer Behavior; Parametric models; Random-Coefficient Models; Statistical analysis; Studies; Sumer; Websites; World Wide Web
• ISSN: 0732-2399; 1526-548X
• DOI: 10.1287/mksc.22.4.520.24906

In this paper, we develop an analytical approach to modeling consumer response to banner ad exposures at a sponsored content Web site that reveals significant heterogeneity in (unobservable) click proneness across consumers. The effect of repeated exposures to banner ads is negative and nonlinear, and the differential effect of each successive ad exposure is initially negative, though nonlinear, and levels off at higher levels of passive ad exposures. Further, significant correlations between session and consumer click proneness and banner exposure sensitivity suggest gains from repeated banner exposures when consumers are less click prone. For a particular number of sessions, more clicks are generated from consumers who revisit over a longer period of time, than for those with the same number of sessions in a relatively shorter timeframe. We also find that consumers are equally likely to click on banner ads placed early or late in navigation path and that exposures have a positive cumulative effect in inducing click-through in future sessions. Our results have implications for online advertising response measurement and dynamic ad placement, and may help guide advertising media placement decisions.