Content integration for customer-focused applications

Most corporate applications focused on delivering client information on an enterprise basis-particularly customer portals and CRM installations-fail to achieve anywhere near their full potential. They don't fail for lack of technical expertise or commitment of people and dollars. They fail because

Laura Ramos, analyst at Giga Information Group, states in a March 30, 2002 report titled "Taxonomy, Thesaurus, Tagging: Balancing Automation and Editorial Review" that as companies deploy portals and enterprise applications "the need to deploy content organization and classification technology becomes more important. Without ways to organize content, categorize it automatically and provide relevant navigation, applications that focus on collaboration and communities can fail to provide effective user experiences or to deliver returns to the business."

We have observed that several leading West Coast technology firms-which lead the market with cutting-edge software and hardware solutions-cannot deliver the envisioned breadth and depth of company information through their customer portals for this reason. A leading strategy-consulting firm identified a minimum of five separate company schemes (dare we call them taxonomies?) for different applications across one such company.

Most taxonomy solutions are limited to unstructured textual documents, like news. They run deep on business and topic terms but don't perform well for tagging companiesthe key entity to a customer-focused application. They lack global breadth below the top few thousand companies and they almost all lack the hierarchical information of corporate families necessary to identify and tag references to subsidiary operations of diverse multi-national corporations.

Tools and services to apply taxonomies have likewise focused on auto-classification of textual information, using advanced techniques that analyze the distribution or presence of key words and phrases. Just read a dozen papers on taxonomy, classification, or knowledge management and you will find that over 95% of all references are to "documents," meaning unstructured text. While textual classification approaches have become quite sophisticated, they do not solve the problem of matching or integrating structured database-driven information.

The problem is that much of the critical information on one's clients-corporate profiles, key contacts, opportunity assessments, purchase history, etc.-will likely reside in relational or other structured databases. Integrating structured company data is not an issue of taxonomy generation and auto-classification of documents, but is rather one of maintaining a controlled universe of key corporate data and deploying sophisticated matching algorithms. Such algorithms utilize data points like address, phone, and postal code and don't just rely on company name. And because no company limits its sales to publicly traded companies, common taxonomies, like stock ticker, don't cover the necessary universe of companies for customer-facing applications.

A complete content integration solution can handle all types of data-whether internal or external data, structured database or unstructured text-so that all information on a client can be accessed from a single point of access in a customer portal or CRM system. The complete solution therefore includes:

A broad, integrated taxonomy that covers companies, topics, industries/lines of business and geographies that can be used both to describe metadata for textual documents (such as call reports and news stories) and to organize structured company data (such as corporate profiles and sales histories).

A strong company taxonomy that supports aliases, former names, common identifiers (like ticker and CUSIP), and subsidiary names. It should also support mapping to a variety of commercial company identifiers should the user want to integrate multiple commercial sources.

A taxonomy solution that embraces industry standards like SIC and ISO codes as well as other proprietary schemes. Any taxonomy that ignores standard codes will fail to support client data solutions (that were most likely built to such industry standards).

Auto-classification tools to apply a taxonomy to unstructured text documents with a high degree of accuracy.

Automated matching tools to map structured company files to a common company taxonomy.

Services to perform editorial review in support of automated matching and classification techniques, because no tool alone can perform matching to the high degree of certainty and completeness required.

The bottom line is that whatever content integration solution a company chooses, it is critical that they understand their full set of needs before selecting a provider. And to do this, they must first ensure that their portal or CRM team includes someone experienced in content integration and knowledgeable about taxonomies and classification solutions.

OneSource provides Web-based business information to professionals who need quick access to reliable corporate, industry, and market intelligence. Our products are designed to address information needs of leading professional and financial services firms, technology companies, and other large organizations. Representative customers include Bain & Company, Citibank, Bank One, Cisco Systems, Compaq, Deloitte & Touche, Grant Thornton, Harvard Business School, Akamai Technologies, Merrill Lynch, Oracle, SAP, Sun Microsystems and TNT.