|
Activity Number:
|
156
|
|
Type:
|
Invited
|
|
Date/Time:
|
Monday, August 4, 2003 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Business & Economics Statistics Section
|
| Abstract - #301183 |
|
Title:
|
Predicting Online Purchase Conversion Using Web Path Analysis
|
|
Author(s):
|
Alan Montgomery*+ and Shibo Li and John C. Liechty and Kannan Srinivasan
|
|
Companies:
|
Carnegie Mellon University and Carnegie Mellon University and Pennsylvania State University and Carnegie Mellon University
|
|
Address:
|
Graduate School of Industrial Admin., Pittsburgh, PA, 15213-3815,
|
|
Keywords:
|
clickstream data ; path analysis ; multinomial probit ; hierarchical Bayes models ; hidden Markov chain models ; vector autoregressive models
|
|
Abstract:
|
Clickstream data provide information about the sequence of pages viewed by a visitor as they move through a web site. A valuable facet of this data is the navigation or web path the user has chosen to traverse the web site. This path reflects a user's goals, which we use to predict a user's potential to purchase. One application of path analysis is to predict which users are likely to make a purchase as they browse through web site. An online retailer could use path analysis to improve the design of their web site and better target customers. In our research we propose a dynamic multinomial probit model to predict the path that a user will take through a web site. Our model is formulated within a hierarchical Bayesian framework to account for consumer's observed and unobserved heterogeneity. Additionally, the model incorporates a mixture process whose multiple states are governed by a hidden Markov switching model to capture within user heterogeneity. Our results show that online retailers should use different marketing mix tools, web design, and navigation paths to target the right customers at the right time to bolster their purchase conversion rates.
|
