Emerging issues in e-commerce: security, advertising, and personalization

Vijay Mookerjee,Dengpan Liu

2006-01-01

Abstract:The ever expanding realm of E-Commerce has seen the advent of three key phenomena---(i) the increasing urge to secure sensitive information stored at web sites, which have suffered great loss from attacking in the past few years, (ii) the need to deploy the optimum advertising strategy considering IT capacity constraints, and (iii) the adoption of personalization technology to deliver personalized service to each on-line customer. My dissertation consists of three essays, each of which separately addresses one of these three important aspects. Essay I is to analyze how managerial decisions, such as decisions on information security investment and decisions on security information sharing, get affected by the nature of information stored at different sites. Essay II is to examine the impact of limited information processing capacity at an electronic retailer's web site on the optimal effort in advertising, or any related activity that generates traffic to the site. Essay III is to address the problem of allocating personalization server resources for content delivery sites. In essay I, I found that the nature of sensitive information stored at two firms does affect certain characteristics of the system, e.g., system vulnerability and the amount of investment, when firms make security investment and knowledge sharing decisions. I also developed coordination schemes which can make firms achieve the social optimum level security investment. In essay II, I examined how the IT capacity constraints affect the advertising strategy in both monopoly and duopoly cases. In monopoly case, I show that when the capacity constraint is imposed, firms should advertise less in certain regions and sometimes, take exactly opposite decisions; and as a result, the steady state market share also reduces. In duopoly case, I show that when the capacity constraint is imposed, at the Nash Equilibrium, the heterogeneity of steady state market share reduces, i.e., IT capacity constraint will dilute the advantage of the firm which has higher advertising efficiency; the advertising rate also changes and the changing pattern is identified. In essay III, I presented the batching approach to allocate personalization service to arriving requests at content-delivery websites, and identified that requests selected for personalization within a batch should be scheduled in reverse order of their profile quality. When the arrival rate is relatively low and high, our approach leads to substantially improved profits as compared to two viable heuristics. When the request arrival rate is moderate, the maximum selection heuristic slightly outperforms our proposed batching approach. We also develop an experimental procedure to determine the optimal batch length based on data about visitors of a site. The optimal batch length is sensitive to the request arrival rate when the rate is close to or less than the server capacity. The batch length is relatively stable for high arrival rates (peak times), which is when our policy is most useful.