Finding New Customers Using Unstructured and Structured Data
Abstract
Identifying new customers is a critical task for any salesoriented company. Of particular interest are companies that sell to other businesses, for which there is a wealth of structured information available through financial and firmographic databases. We demonstrate that the content of company web sites can often be an even richer source of information in identifying particular business alignments. We show how supervised learning can be used to build effective predictive models on unstructured web content as well as on structured firmographic data. We also explore methods to leverage the strengths of both sources by combining these data sources. Extensive empirical evaluation on a real-world marketing case study show promising results of our modeling efforts.
