Currently a corporate web site is not considered as a necessary business attribute, but as a marketing tool which should yield results. In this study we consider a web site as an instrument for attraction of new partners (customers, suppliers). Web site outputs are a number of visitors interested in contact information (reached the contact info page) and a number of visitors who sent a request via a special form on the web site. We build a sequential discrete choice model for web site outputs. Explanatory variables set includes a number of visited pages, seconds spent on the web site, and dummy variables for specific pages visited (a page with prices information, a portfolio page). Also we investigate an influence of search engines (Google, Yahoo, MSN), which refer a visitor to a corporate web site and keywords used for pay-per-click advertising campaigns. We estimate model parameters on the base of a small UK-based web development company's web site statistical data and discover strong dependencies, which allow improving web site organisation and its search engine positioning.
