OR/MS Today - August 2000

August 2000

Cyberspace

E-Biz Opportunities Abound

By ManMohan S. Sodhi

And now, as in Monty Python, it's time for something completely different. My last two articles have been about supply chain management in the context of the traditional, albeit Internet-enabled, buy-make-move-sell framework. According to David Ross, e-commerce manager at IBM, the new supply-chain model is "sense and respond" instead of "make and sell." Certainly, operations research can help with traditional responses like redeploying inventory using real-time information on supply and demand, or reconfiguring orders to meet or accelerate delivery date. OR can also help product lifecycle management in conjunction with the Internet. But there are also new opportunities for OR in this new model.

Firms can use operations research:

In design by optimizing product attributes, improving the use of design and development resources, and planning the phase-in for new products as well as the phase-out of old products. Analysts can use customer data collected in Web surveys to determine desirable product characteristics quickly and improve product designs. Planners can access design and capacity information over a firm's intranet or extranet to plan for improved resource usage.

In sales by predicting demand both at the individual and aggregate levels more accurately than it would be possible otherwise, and deploying inventory effectively to meet changing demands. Analysts can analyze data on Web site visitors' navigational paths to predict sales at the individual customer level. They can also predict aggregate sales and redeploy inventory if need be.

In customer relationship management by designing call centers and other service center facilities to improve service and the use of resources. CRM permits managers to integrate existing and new channels to support customers, sales and marketing. These managers can use OR strategically to design call centers and to integrate their customer relationship infrastructure and resources. They can use OR tactically to deploy algorithms to aid in automated and manual, representative-aided responses to customer requests.

One-on-One Marketing

The large amount of consumer data collected by retailers combined with a desire to target individual customers has spurred the growth of data mining. Web sites are collecting further data as users navigate around them. This has motivated development of OR-based tools for predicting an individual consumer's purchasing behavior in real or delayed time.

Collaborative filtering is a way to establish what ads to display to Web users who have browsed ads or made purchases. Collaborative filtering software compiles customers' purchasing information to pool them into clusters and uses some cluster members' purchasing patterns to predict the buying habits of others in the same cluster. It does this in real time and, for instance, can put an ad on the customer's screen while he or she is making a purchase. (For more information, see www.acm.org/siggroup/collab.html.)

Personalization is the real-time or delayed modification of a Web site by attempting to display content or ads of products and services that likely interest the user. As mentioned, collaborative filtering is one way to achieve this, but there are others, like Markov chain models.

"Clickstream" analysis involves collecting and analyzing users' navigational paths on Web sites, mining the data they create. Analysis at an aggregate level is useful for improving Web site design, and analysis at the individual level is useful for personalization.

Companies that provide tools and services in this area include DoubleClick for advertising management, Vignette for automated delivery of content, Net Perceptions for real-time personalization, net.Genesis for mining customer data, HNC Software for software to predict customer behavior, IBM for decision support and advanced data mining, Quadstone for predicting customer purchases, and WebTrends for analyzing marketing campaigns and integrating visitor data with data from other sources, such as ERP.

Customer Self-Service

Customer service applications that allow customers to directly access technical or sales information on Web sites or through automated e-mail response also provide opportunities for operations research. Automated self-service or other Web-based resolutions to customer queries cost a tenth of the resolutions requiring a customer service representative on the telephone. Existing applications reportedly use expert systems, knowledge engineering and case-based reasoning. These applications typically respond to e-mail or assist a customer service representative by matching a query against a knowledgebase that includes both content and rules for routing and problem-escalation. For instance, the software may react to a customer's e-mail by deciding that the customer is reporting a problem. It could respond by first searching the knowledgebase for information it considers of potential use to this customer and then e-mailing it to him or her. It could also route the original e-mail and its response to a specific engineer to follow up. Software companies with products in this area are Acuity, Edify, eGain, Inference, Net Effect, Silknet and ServiceSoft.

Conclusion

E-business in all its various forms is creating new opportunities for operations research. There are new opportunities for Markov models and statistics and extended opportunities for linear programming. These opportunities may not be completely different, but as in Monty Python, the attraction comes from the mix of the familiar and the unfamiliar.

Dr. ManMohan S. Sodhi (MohanSodhi@AOL.com) is director of enterprise e-business strategy with Scient in Chicago. He is president of the Logistics Section of INFORMS.

Table of Contents

OR/MS Today Home Page

Lionheart Publishing, Inc.
506 Roswell Street, Suite 220, Marietta, GA 30060, USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com
URL: http://www.lionhrtpub.com