Example Application: May I Help You next up previous
Next: The Vortex Language and Up: Personalizing E-commerce Applications with Previous: Introduction

2. Example Application: May I Help You


In a department store, customers are free to browse. In a good department store, a salesperson will sometimes approach customers with the gentle question ``May I help you?". In an excellent department store the timing and manner in which this question is asked is guided largely by the browsing behavior of the customer. The May-I-Help-You (MIHU) system provides a functionality for web-based storefronts that is analogous to this kind of service in excellent department stores. The MIHU system has an important advantage over a department store salesperson, which is that many businesses know the identity of customers during their visits on the web.

MIHU is a Customer Relationship Management system that interfaces to a business' web storefront. MIHU can keep track the interaction of a customer with the storefront. To be more specific, using the high-level Vortex language business analysts and managers can program the MIHU system to use customer interaction information (e.g., shopping cart content, sequence of pages visited), coupled with information available in enterprise databases (e.g. customer profile, contact history, current orders, and results from off-line decision support tools) to build a model of the customer and the current interaction. Based on the individual characteristics of each customer interaction, MIHU may choose to present to the customer an icon or window offering help relevant to the current context. This help might be automated, or it might be an offer to chat with a live Customer Service Representative (CSR). In the first case, if the customer takes up the offer (e.g., by clicking on the icon or window), then appropriate context-dependent information will be delivered to the customer. In the second case, a CSR will be assigned, appropriate information will be forwarded to that CSR, and some kind of interaction with the customer will be initiated. Of course, providing live CSR help with an interactive session brings with it the opportunity to help close the sale, and also the opportunity to attempt cross-sells or up-sells.

Figure 1: Overview of MIHU functionality



Figure 1 summarizes the operation of the MIHU system. There are four phases or aspects to the operation. In the first phase (shown with lines having small dots) a customer has ``normal'' interaction with the web store-front. In particular, the web server supporting the store-front presents pages to the customer's web client, and the customer fills in blanks and submits page requests to the server. However, before the web server presents a new page to the customer, the on-line decision engine is asked whether or not the customer should be presented with the ``May I Help You'' option (or some other optional assistance or customer service such as a targeted discount). The decision server can access information about the customer's current web session (e.g., pages visited, shopping cart contents), and may access data from an enterprise database (including results from off-line decision support systems). The decision server may also gather information from decision engines using alternate reasoning paradigms, such as an expert system or, e.g., a specialized system for determining customer preferences.

The second phase (shown with lines having large dots) occurs if and when the decision engine determines that the customer should be given the MIHU option. In that case, the web server presents to the customer's client an applet that asks whether the customer would like assistance from a CSR.

The third phase (shown with lines having short and long dashes) arises if the customer does want assistance. In that case, some or all of three forms of interaction can be established between customer and CSR: voice conversation, text chat, and ``escorted'' or ``collaborative'' web browsing (where the CSR can select a URL and both the CSR and customer clients go to that URL, or visa-versa).

The fourth phase (shown with lines long dashes) occurs in parallel with the other ones, and at a more deliberate pace. This stage involves tuning for business performance, i.e., the continued examination of the decisions made for the web-storefront, with the ultimate goal of making improvements on the underlying decision policies. As will be described below, novel aspects of the Vortex language make it possible to quickly modify a Vortex program in order to achieve a desired effect.

At a superficial level, it might seem that since the MIHU system monitors a customer's progress through a web site, and peeks at the interaction between her and the web site, there are serious privacy issues involved here. However, this is not the case since the MIHU system is not getting any extra information that is not already available to the web site. However, it might be a good idea to let the customer know that such monitoring might be going on (e.g., by allowing her to opt-in when she registers with the site).

next up previous
Next: The Vortex Language and Up: Personalizing E-commerce Applications with Previous: Introduction
Rick Hull