Internet-based Infobroker for Electronic Commerce

The focus of this paper is on the role of business-to-consumer relationships and on the role of new communication technologies in establishing and maintaining relationships in electronic commerce. Therefore, a prototype of an Infobroker is presented using special services that allow individualized contacts with each customer. These contacts are carried out in the form of personalized information deliveries. Intelligent solutions also act as recommender systems tracing how the clients interact with the system and using this information for future contacts. In our case the goods sold are hardware components. 1. Basis of the Infobroker An Infobroker for electronic-commerce applications is generally understood as a system that collects product information from various sources and that presents it to customers according to their individual profiles. The enterprise that offers such an Infobroker will try to provide to its customers not just a product or service which fulfills the expressed specific customer need, but a set of products and services that cover the whole range of problems related to the specific problem submitted by the client. We chose hardware components as the product base and implemented such an Infobroker using modern concepts like individualized information deliveries or recommender systems. Reasons for selecting hardware components are manifold: They possess clear product attributes that can easily be stored in databases. Moreover, a lot of manufacturers already offer their products as well as information about them in the WWW. Thus, products of the same product category can be compared easily. We developed the Infobroker in the field of business-to-consumer relations with the product database of a leading European hardware distributor. The product catalogue represents a hierarchy consisting of three different levels. Altogether, about 5,000 products are categorized. As the system will be available online, this allows to study the impact of customer relationships quite respectably in an application that combines a fixed-price and an auction selling system. 1.1. Separate Offer and Product Data Using a standardized database, it is very convenient for the dealers to input their offers. They only have to select the relevant hardware component without adding a further description as it is already part of the database. The clients get the same quality of information for all available products because it does not depend on the suppliers’ input but on the underlying database. Furthermore, this approach reduces the likelihood of wrong data, i.e., incorrect product descriptions. The strict separation of offer and product data prevents the system from having different specifications for one product at different dealers. This is a drawback some other Internetbased systems have. Moreover, the process of shopping through our system is separated into three steps. First, the client selects the required product in the database. The given actual price helps her/him to do so. Secondly, the customer can choose the preferred dealer. Finally, the sale itself can take place online. 1.2. Integration of Fixed-Price and Auction Selling System Currently, the described approach is as far as we know the only system that integrates two different selling methods (fixed-price and auction) in one user interface. The customer can decide to see only one of the methods at any time which improves the transparency of the presentation. Only when the customer proceeds the sale there are differences. Using the fixed-price sale (we also use the expression ‘catalogue sale’ in the following) the client is always provided with a shopping basket in which the consumer can collect the products he/she is interested in. These articles are the basis for the order following later. In an auction the client only gets information on the current status of the auction, for instance, the highest bid. Thereafter, the offer can be submitted. These options offer clients two chances to get a ‘bargain’. On the one hand, they can easily identify the cheapest catalogue supplier of a certain product. On the other hand, they may participate in an auction to get the product at a reasonable price. Depending on their preferences the customers may choose one of the options only or they can use both simultaneously. The suppliers can use auctions, e.g., to sell phase-out models or shelf warmers at prices as high as possible. A sufficient number of potential customers is always at short distance as the auction is integrated in the fixed-price selling system. The fixed-price selling system’s focus lies on the product and offer selection. The customers are immediately directed to the order section of the system after the offer selection has taken place. At an auction, the two steps product and offer selection coincide because there usually is only one auction for a certain product available. The special attention lies on the bidding event here. 1.3. Fixed-Price Selling System 1.3.1. Means of Access In order to take care of the importance of customer satisfaction and easy navigation as key issues, we built up three different categories to access the hardware components: In the first category we unified a search containing product hierarchy and fulltext search to get to the interesting components. Secondly, there is the attribute-oriented search. And thirdly, the system offers a problem-oriented access for PC systems. Product Hierarchy and Fulltext Search The depth of the product hierarchy is important for the ‘browsing’ approach. The more levels the hierarchy comprises and the more complicated the structure of the offer gets, the better the user guiding has to be. A permanently available list of the products’ hierarchy helps to alleviate this problem. Using this list, the user will easily leave for new products or product groups. Clients that are looking for a specific product may use the fulltext search to find a certain product or supplier. Attribute-oriented Search After selecting a product group, the customer can further restrict the search by defining values for certain characteristics. Another option is to demand some characteristics of the product. An example for this could be to guarantee a perfect cooperation of parts that are bought to complete existing systems. Or one might just want to specify certain technical features. In either case, this approach is rather suitable for ‘experts’. A modification of the attribute-oriented search reduces the customers’ workload and offers potential for cross-selling. The ‘Find Similar Products’ option takes the characteristics of a certain product using them to start a new attribute-oriented search. The client can specify whether the product’s features have to be matched or if a slight deviation is allowed. Problem-oriented Search This approach is especially suited for ‘novices’. The system does not ask them for technical features but for the intended purpose of the product. Normally, the most frequently run applications (word processor, spread sheets, ...) or some technical features (e.g., range of use: home or office) have to be given. For instance, the customer selects an appropriate system out of a list of reference systems. As the list also offers a brief description of the products’ technical features, the customer can easily adapt them to one’s individual needs. It is also possible to state the name of an often used software. This permits the system to derive the relevant technical requirements and to find an appropriate computer system. 1.3.2. Intelligent Shopping Basket Shopping baskets allow the customer to make a note of an interesting product and to prepare one complete order instead of many small orders even though the items of the order were selected in different purchasing acts in different sessions. Differences are still in their persistence. Some systems do not make records of any basket, whereas other applications even ‘remember’ baskets of non-registered customers and present them to the user when she/he enters the system the next time. Intelligent additional functions are important, e.g., an adjustable budget limit or references to the customers’ individual shopping history to allow repeated sales or ‘standing orders’. Sometimes, a client might be able to work with two different shopping baskets: One could be called the ‘preliminary’ basket, the other the ‘real’ one. The basic idea is to collect all information the user deems interesting in the ‘preliminary’ basket and, after considering all options, to let the customer make the final decision at the end. The possibility to make proposals for future purchases based on customer’s former baskets will be explained later (see 2.2). 1.4. Auction Selling System At an auction system the bids have to be submitted as WWW forms. The bids can later be raised easily. This is possible because each participant in the auction can access the system simply by using a WWW browser to be informed on the current status of ‘one’s auction’. This and the time-independence of an auction that does not require a customer to watch the auction permanently make the auction a very convenient and attractive application. As opposed to an ordinary auction that stops when no further/higher bids are given, an online auction has to terminate at a previously fixed and announced point of time. 2. Realization of the Infobroker The Infobroker consists of an extensive evaluation system of customer activities, an adjustable newsletter and a sophisticated recommender system. Some more detailed elements are shown in figure 1. The aim of the system is to increase and to stabilize individual customer relationships by finding out what the customers need through interaction and feedback. As a distinguishing mark to conventional online shopping sites the system notices the special preferences of its customers in order to increase their loyalty. Various user activities are logged, e.g., which part of the hierarchy, which product groups or product subgroups are visited frequently, and whether the center of interest lies on the catalogue or on the auction part. Individual Profile customer data (profiles of interests) standard profiles (problem-oriented) Individual Profile individual profiles (self-defined) combine individual newsletter (Newsletter, Channel, Top 10) log data product data