Expert Systems

In the competitive world of e-commerce, high costs and scarce resources often drive companies to try self-service approaches to CRM, especially in customer support and contact center operations. Most self-service systems fail, however, by attempting to educate customers to answer their own questions rather than providing situation-specific answers or recommendations. In taking broad strokes at gathering and delivering the information customers require, these systems frequently leave customers overwhelmed and unfulfilled. Self-service CRM initiatives that fail to provide answers to customer questions lead to frustrated customers calling contact centers and help desks, which also may not lead to the answers they are looking for. On the opposite end of the spectrum, customers are no longer willing to spend time sifting through FAQ listings or to put up with Web searches yielding hundreds of thousands of hits. So well meaning initiatives that provide an overabundance of information can just as quickly bring a customer's loyalty to an abrupt end. What's missing is the element that can make self-service CRM successful, providing customers with interactive sessions that offer quick, complete, and of course, correct answers to their questions. Knowledge-automation expert systems pick up where traditional self-service CRM systems leave off, emulating discussions with top-level support staff and providing customer-specific recommendations and answers. The case for more satisfying customer interactions Self-service CRM education approaches have two main problems: 1. Most people do not want to invest the time to be taught or read pages of information; they just want a quick answer to their question. If you are sick, you're not looking to land a medical degree--you just want to speak with an expert (doctor) who can tell you what's wrong and what you should do. 2. Technical, regulatory, and other matters are far too complicated for an education system to teach customers enough to reliably make a good decision--and a little knowledge can be dangerous. Cased-based systems are an alternative resolution attempt. They try to match a new case with a previous one in a database, attempting to make educated guesses based on pattern-matching algorithms. However, such approaches typically omit analyzing the customer's situation logically, and the guesses are often wrong. To be truly effective, self-service CRM initiatives need a front-end knowledge-automation expert system solution. The ideal solution is an automated, Web-based system that provides specific advice and recommendations comparable to a human expert, and that emulates the one-on-one conversation a customer would have with an expert. Made available directly to the customer, or used by the first-level help-desk staff, such systems are allowing companies to effectively and efficiently deliver expert advice and satisfying experiences to their customers. Expert decision-making knowledge for commonly occurring questions can be captured in a way that allows it to be automatically delivered over the Web. A successful self-service CRM system should incorporate both expert knowledge and intelligent process automation (IAP) to analyze the customer's situation and provide advice or recommendations that reflect each customer's needs. Knowledge-automation expert systems have a well-demonstrated track record of providing correct, consistent, individualized answers for product recommendation, customer problem resolution, company compliance, and many other CRM and business decision-making processes. Easily incorporated into Web sites and intranets, these systems can tie into existing CRM and sales support systems, or can be very effective on their own and in kiosks. 100-to-1 ROI With current development tools, expert systems can be created quickly, successfully, and at relatively low cost. They don't require a costly or risky multitier development or implementation effort so common with other CRM projects, and benefits are well proven. For example, DuPont, which has probably built more expert systems than any other Fortune 100 company, reports a typical 100-to-1 ROI in fielding expert systems--for every $10,000 spent, it saved $1 million--adding up to over $2 billion. Expert systems are built by using development tools that simplify describing the decision-making logic and processes of human experts in rule form. These rules cover all aspects of a decision-making process, and can be applied to a wide variety of situations. In each customer session (typically via a Web browser), the expert system "inference engine" uses the rules to ask the customer questions that are needed to make a decision. The expert system interacts with the user, asking specific questions to reach a recommendation. Based on each customer's input, the system will skip irrelevant questions and drill down where appropriate. Data can also be obtained automatically from a database, or from other parts of the CRM system. The logic in the system may be very complex, but end-users only see simple questions that they can answer, and precise recommendations that they can use. They are not expected to learn or understand how the decision was reached (although that can be implemented). They are simply given a valid answer that they can act on immediately. Another benefit is that the advice is consistent--if two customers provide the same input, they will get exactly the same recommendation, and it will be based on the expert knowledge in the system. Expert systems can be added to CRM and sales support systems as individual modules to answer questions. They can also be integrated as a front-end to traditional contact center systems as a very capable first level of self-service support, with the ability to automatically open traditional tickets with all relevant data for problems beyond the scope of the system. Expert systems should be a key part of any Web-based self-service CRM. They are the only technology that can automatically provide precise, detailed recommendations for the wide range of increasingly complex issues that CRM systems have to deal with, and do it in a scalable way with fixed costs. About the Author Dustin Huntington is the president of Exsys Inc. Please visit www.exsys.com