Information Systems Research, 2005, Volume 16,
Issue 4, Page 334-351.
Internet firms frequently employ a two-stage approach to promotional activities. In Stage 1, they attract customers to their websites through advertising. In Stage 2, firms generate sales transactions or sales leads through their website. Comprehensive assessment of the promotional performance of pure online firms requires the study of Stage 1 and of Stage 2 jointly. In this paper we develop a joint two-stage conceptual and econometric model for assessing website promotion on three important dimensions: (1) how advertising response can be measured by linking media schedules to website log files; (2) how advertising and website characteristics jointly affect the desired system outcome of the promotion; and (3) whether the joint investigation of advertising response and desired system outcomes is essential to assess the results of website promotion. Three general findings follow from application of our model to a pure online firm's campaign to generate sales leads through print advertising. First, advertising and website characteristics affect sales leads in different ways. A characteristic may influence sales leads directly, or indirectly, or both. Second, assessing advertising effectiveness in an online environment may not require costly survey research data. Instead, secondary data available from website log files may be used for such assessment. Third, the interaction between the first and second stages of our two-stage model can lead to misspecifications that produce misleading inferences. This occurs because the unobserved characteristics in generating website visits and sales leads may be correlated.
Keywords: e-commerce; sales leads; sample selection; two-stage model; website traffic