Chapter XIII Shopping Agent Web Sites: A Comparative Shopping Environment

XIII Shopping AgentWeb Sites A Comparative Shopping Environment

Ming Wang
California State University - Los Angeles, USA


This chapter introduces the shopping agent technology as a new Internet marketing trend. The recent development of shopping agent Web sites has offered online shoppers an excellent comparative shopping environment. Shopping agent Web sites, also called shopping agents or shopping bots, are software programs that search the Internet stores on the Web and find products that meet a buyer’s specifications. Not only do these agent Web sites bring comparative product and price information from individual merchants’ Web sites, but also provide the online merchant ratings to customers. In summary, these agent Web sites take a query, search the Web sites of individual merchants that may have the product sought, bring back the results, and present them in a consolidated and compact format that allows comparison shopping at a glance.


The enormous amount of information available to online shoppers on the Internet makes it difficult to find relevant products and merchant information. The recent development of shopping agent Web sites has offered a practical solution for this information overload problem and provided online shoppers an informative, convenient and comparative shopping environment. Recent research has shown that consumers prefer to shop and buy on the Internet for three major reasons — convenience, time-savings, and comparative shopping (Oz, 2002). The features of shopping agent Web sites fit these customers’ needs.

Shopping agent Web sites are created with agent software that performs functions such as information gathering and information filtering on behalf of a person or entity. In the comparative shopping arena, shopping agents can provide the customer with comparative prices for a desired product, a review of the product and a review of the number of shopping alternatives before they make their purchase decisions. Once a customer selects a product with a merchant, the individual merchant Web site will process purchase and delivery details. Like a personal digital assistant, an agent provides personalized assistance in a specialized task for customers. From the customer’s point of view, an agent reduces complexity, increases efficiency, and supports user mobility (Liu, 1998). For example, the agent will first locate the merchants’ Web sites selling specific products. Then, the agent will collect information about the prices of a targeted product, and its features from these merchants. Most shopping-bots claim to eliminate the searching necessary to identify the right product at the best price. It has been proposed that the availability of agent Web sites is one of the reasons why e-markets should be more efficient (Mougayar, 1998).

Some auction agent Web sites provide a negotiation service. Agents will represent both buyers and sellers. Once a buyer identifies a seller, the agent can negotiate the transaction. The agents will negotiate a price and then execute the transaction for their respective owners. The buyer’s agent will use a credit card account number to pay for the product. The seller’s agent will accept the payment and transmit the proper instructions to deliver the item under the terms agreed upon by the agent.

The shopping agent Web sites work as a third party between online customers and merchants providing services for both sides. Member merchants can benefit from the shopping agent’s market report. The report provides brand, price trends, market share and customer buying habit information based on real-time data from the database that monitors over 40,000 of the most actively traded products. In addition, merchants can also receive feedback in the form of online evaluations from customers. Positive evaluations will draw more customers to the evaluated merchants. Negative evaluations will help them learn lessons and improve their service quality. In the auction environment, eBay makes reputation reports about each other available to its bidders and sellers.


The software agent with roots in problem solving and knowledge representation is not a new concept, but had not grown until the past decade when the Internet created a perfect environment for e-commerce. Dr. Pattie Maes, founder of the MIT Media Lab Software Agent Group, and other researchers have developed a number of intelligent shopping agents. Based on the customer buyer behavior (CBB) model, Maes and Guttman (1999) identified and implemented six stages of the buying process: need identification, product brokering, negotiation, purchase and delivery, and product service, and evaluation. Need identification characterizes the buyer’s need. Product brokering includes retrieval of product recommendations to help determine what to buy. Merchant brokering utilizes merchant ratings to help determine who to buy from. Negotiation considers how to settle on the terms of transactions. Purchase and delivery signal termination of the transaction process. Product service and evaluation involve post-purchase services and evaluation of satisfaction with the overall buying experience and the decision.

Shopping agent Web sites represent the newest trend in Internet shopping technology and have been accepted by a large number of e-merchants and online shoppers. The most popular shopping agent Web sites are bestWebBuys, bizRate, dealTime, mySimon, RUSure, and nextTag. Although it is difficult to determine how these agent Web sites have affected online sales of individual merchants, more and more merchants have subscribed to agent Web sites to attract more customers. For example, the PriceGrabber company, which added software agents for e-shopping and raised the standard for online shoppers’ comparison Web sites on November 8, 1999, had 1,050 subscribed merchants by the end of October 2003. The growth rate for the number of subscribed merchants was about 10% each month.

Academic interest in shopping agents has grown rapidly. Resnick et al. (2000) predicted that reputation systems that rely on the participation of large numbers of individuals accumulate trust simply by operating effectively over time. With clear reputation ratings, merchants with good reputations could charge a premium for their services, since some users may be willing to pay for the security and comfort of high-quality services. Cooke et al. (2002) explored how consumers respond to recommendations of unfamiliar products made by shopping agents. Ganesh and Amit (2003) examined the impact of Internet shopping agents (ISAs) on market competition and indicated that an ISA can potentially prevail over the competition due to the mass of consumers attracted to the site. The equilibrium of inside pricing is such that the average price charged can increase or decrease when more retailers join. In April, the Sloan School of Management at the Massachusetts Institute of Technology hosted a symposium sponsored by the National Science Foundation on the Online Reputation Mechanism. The potential financial implications of online reputation and fraud risk management were discussed.

Issues and Problems

Most agent shopping Web sites currently offer three features: competitive prices for a searched product, product recommendations, and merchant reputation ratings. These features help customers search for a product with the lowest price and choose merchants they can trust and feel comfortable with.

Competitive Prices

Most shopping agents claim to eliminate the search necessary to identify the right product at the best price. They take a query, visit their member e-tail stores that may have the product sought, bring back the results, and present them in a consolidated and compact format that allows for comparative shopping at a glance. The impact of this comparative price system lowers the search cost not only for consumers but also for suppliers who wish to find out what prices their rivals are charging. This makes it easier for merchants to operate price-trigger strategies that may enable them to sustain high prices. This also makes it more difficult for sellers to undercut each other secretly (Vulkan, 1999).

Recent research shows that price and promotion are no longer the main reasons for a customer’s decision to make a purchase. More and more sophisticated online customers prefer to pay higher prices to merchants who provide high quality e-service (Schneider, 2002). The most experienced and successful merchants are beginning to realize that the key determinants of success or failure are not merely Web presence or low price, but e-service quality (Zeithaml, 2002). The customer’s e-satisfaction rating is an important measure of e-service quality.

Product Review

The product reviews on an agent Web site are contributed by customers shopping on different e-tailers’ Web sites linked to the agent Web site. Thus, the product review system is able to provide a large volume of product reviews from customers. Avery et al. (1999) studied the potential of a market for evaluations and indicated that personal experience with products is enormously powerful in forming the customers’ decisions. These personal recommendation networks enable consumers to learn from the experience of others.

Ganesh and Amit (2003) focused their research on how product review systems affect customer preferences of unfamiliar products since recommendations of unfamiliar products represent an important new source of business. Behavioral research suggests that consumers view recommendations of unfamiliar products negatively (e.g., Park and Lessig, 1981). The result demonstrates that positive contextual recommendations based on a customer’s prior purchase interest do not always produce positive effects on the sale of new products. It is important that designers of electronic agents create contexts judiciously. The designers offer tentative guidelines for product review systems. Contextual recommendations may be beneficial when they are known to be attractive to the customer and are likely to be perceived as similar to the target item. The context can be provided in a manner that makes it salient when customers first encounter the unfamiliar recommendations. Conversely, new recommendations should be presented when little is known about the shopper and when familiar recommendations are likely to be perceived as different from the unfamiliar recommendations.

Merchant Reputation Systems

In addition to obtaining competitive prices and product recommendations, many shopping agent Web sites have developed systems to keep track of merchant reputation ratings as well as service quality simply because consumers want to know that the merchant with whom they are transacting an order is reliable, and will deliver the product as specified in the delivery schedule, will maintain confidentiality, and will have appropriate product packaging and handling arrangements. Resnick et al. (2000) have defined a reputation system as one which collects, distributes, and aggregates feedback about participants’ past behavior. Though few producers or consumers of the ratings know one another, these systems help people decide whom to trust and also encourage trustworthy behavior. So far, the merchant reputation system is best known as a technology for building trust and encouraging trustworthiness in e-commerce transactions by taking past behavior as a publicly available predictor of likely future behavior (Dellarocas, 2003).

Merchant reputation mechanisms initially attracted attention as a mechanism for building trust. Trust is an essential concept that an e-business should attend to since trust has been considered a building block that strengthens relationships between customers and merchants (Siebel and Hous, 1999). In the B2C e-commerce environment, trust is more difficult to establish and even more critical for success than in traditional business. The retailers down the block will likely be there tomorrow, but the merchant that exists in cyberspace is often not real in the customer’s eyes (Head and Hassanein, 2002). The customers’ lack of inherent trust in “strangers” in the e-stores is logical and to be expected. If an e-tail store wants to do business, it has to prove its trustworthiness by satisfying customers for many years as it grows (Schneider, 2002).

The merchants’ reputation is built with customers’ shopping experiences and comments. The comments from an e-tail store’s previous customers could be a valuable asset to other prospective customers. The goal for online reputation systems is to encourage trustworthiness in e-commerce transactions by using past behavior as an accessible predictor of likely future behavior. Taylor (1974) found that consumers tend to regard information obtained by “word of mouth” as more objective and possibly more accurate. A satisfied customer will tell three people about his or her experience, but a dissatisfied customer will complain to thirty people. Therefore, consumer comments can be a powerful influence on the purchasing decisions of others (McGaughey and Mason, 1998). The merchant reputation has significant impact on customers’ trust and on their intentions towards adopting e-services (Ruyter et al., 2001). In February 2003, Jeff Bezos, the CEO of, decided to cancel all plans for any television or general print advertising because he believed that his company was better served though word-of-mouth generated through the Internet than by paid advertising. Despite its undeniable importance and widespread adoption, the current merchant reputation system still encounters some problems.

Fraud Risk Management

Key facets of the usefulness and successful adoption of emerging reputation systems are their accuracy, consistency, and reliability. Can, should, and indeed, to what extent, do consumers rely on the information on the agent Web site and use it for their decision-making processes? One party could blackmail another by threatening to post negative feedback unrelated to actual performance. A group of academic scholars also conducted research on the risk management of merchant reputation systems on the agent Web sites. Kollock (1999) states that online rating systems have emerged as an important risk management mechanism in the e-commerce community. Dellarocas (2000) identified several scenarios (“ballot stuffing,” “bad-mouthing,” positive seller discrimination, negative seller discrimination and unfair ratings “flooding”) in which buyers and sellers can attempt to “rig” an online rating to their advantage. Some important management mechanisms have been developed. Friedman and Resnick (1998) discuss risks related to the ease with which online community participants can change their identity. They conclude that the assignment of a lowest possible reputation value to newcomers of agent Web sites is an effective mechanism that discourages participants from misbehaving and, subsequently, changing their identity.

Most agent Web sites have developed countermeasures to counteract the above potential threats in various ways. Some have created registration systems to restrict online evaluation writers only to its member customers. Member customers have to log on to their accounts before they evaluate their merchants. Some agents have claimed that they are able to detect and eliminate fraudulent ratings by using a combination of sophisticated mathematical algorithms and a large number of reviews from a variety of sources that have been checked for consistency.

Rating Consistency

Most online reputation systems have adopted an ordinal rating system to rate merchants as illustrated in Figure 13-1.

click to expand
Figure 13-1: Merchant overall rating system

As Zacharia et al. (2000) indicate, individual standings are developed through social interactions among a loosely connected group that shares the same interest. Through this interaction, the users of online communities form subjective opinions of each other. These opinions may differ greatly among different users, and their variance is large enough most of the time to make the average opinion a rather unreliable prediction.

On the other hand, as an increasing number of online reputation systems become available in the shopping agent Web sites, one customer may be exposed to multiple online merchant reputation systems for the same merchant. Since different reputation systems may provide different ratings for the same merchant, customers may get confused. Thus, questions are raised on the consistency of the ratings provided by different reputation systems. Are online ratings on merchants consistent between different reputation systems? Are online ratings consistent on merchants on the same reputation system? Do they prove to be misleading to customers to a certain extent? If not, to what extent do consumers rely on the ratings and use them for their decision-making processes?

It is crucial for online customers to get to know the reliability ratings of merchant service quality from merchant reputation systems before they make a purchase decision. Consistent ratings are able to provide more reliable evaluations of a merchant. Wang and Christopher (2003) conducted an empirical study on the rating consistency of reputation systems using three sets of data collected online from two popular merchant reputation systems. The results of the study showed that individual customer ratings are not consistent for the same set of merchants across reputation systems. However, the averaged customer ratings are consistent on the same merchant across different merchant reputation systems. The averaged customer ratings on different occasions in the same reputation system are consistent, too.

The solution to the rating discrepancy problem across different reputation systems is to develop a new agent Web site listing the averaged rating for each merchant on different reputation systems as a comparison at a glance reference resource for customers.

Biased Rating

Customers with exceptionally positive or negative views of e-tail stores are more likely to respond than the general customer population. The implication is that those customers who wrote an online comment are strongly opinionated. This results in a biased response that would be more likely to identify current quality problems than a controlled survey of equal sample size. The customers who remained neutral might not be motivated to rate the online store.

People may not bother to provide feedback at all since there is little incentive to spend another few minutes filling out the form. Most people do it because of their gratitude or their desire to take revenge. People could be paid for providing feedback. Further research needs to be done on the portion of customers who did not choose to either express or share their post-shopping satisfaction.

Table 13-1 presents a frequency distribution of 106 overall star ratings classified by customers’ post-shopping satisfaction level. This exhibit lists the frequency of occurrence of each classification of rate. The total number of given ratings was 106. The total percent was computed by dividing the number in each level by the number of ratings.

Table 13-1: Overall rating classified by satisfaction level (Wang & Huarng, 2002)

Level of Satisfaction (Overall Rating)

Number of Overall Rates at Satisfaction Level

Percent of Total

Very Satisfied



Somewhat Satisfied






Somewhat Dissatisfied



Very dissatisfied



Total Reviews




The shopping agent is still relatively new, and there is also significant scope for its further development through the development of its intelligent agent functions. Future agents will automatically build models of shoppers, recommend products to shoppers, negotiate on behalf of shoppers, and personalize the shopping experience.

MySimon is an intelligent agent that can imitate human navigational behavior and can be taught to shop at thousands of merchants in hundreds of product categories. Tete-a-tete is a project within the MIT Media Lab’s Agent-mediated Electronic Commerce initiative. It engages consumer-owned shopping agents and merchant-owned sales agents in integrative negotiations over the full value of each product offering (Guttman and Maes, 1998). These intelligent agent functions take the concept of the shopping agent beyond that of a purpose-built search engine. The future agents will reside on individual computers, create personal accounts for the owners, learn about their owners’ shopping preferences, create a personalized, or individualized “virtual store” and perform more efficient inquiries and purchases in subsequent requests.

Much of the relationship between retailers and customers in the future may be conducted via handheld mobile devices such as palm computers and cellular phones. PDAs already are combined with mobile phones from which users can log onto the Internet. Mobile agents are well suited for e-commerce since a commercial transaction may require real-time access to remote resources. Lang and Oshima (1999) specified seven main benefits for using mobile agents. They reduce the network load, overcome network latency, encapsulate protocols, execute processes asynchronously and automatically, react autonomously to changes, provide optimal conditions for seamless system integration, and are robust and fault-tolerant. Therefore, the design and development of the future shopping agent Web sites should suit the need of m-commerce.

Future Research

From a customer behavior point of view, the impact of shopping agents on the customer shopping decision-making process needs to be studied. Some customers will not use shopping agent Web sites either because they don’t know about them or because they are reluctant to delegate decisions to an artificial agent. One implication of this is that even though customers can determine the prices of those suppliers who choose to allow their site to be interrogated by search agents, they have to decide whether, in addition, they should also engage in a costly search for merchants who are not listed on the agent Web sites. In addition, the proportion of online customers who really rely on product recommendation systems and online reputation systems remains mysterious. The acceptance rate of the product recommendation systems and online reputation systems needs to be determined.


Avery, C., Resnick P., & Zeckhauser, R. (1999). The market for evaluation. The American Economic Review.

Cook, A., Sujan, H., Sujan, M., & Weitz, B. (2002). Marketing the unfamiliar: The role of context and item-specific information in electronic agent recommendations. Journal of Marketing Research, 39(4), 488-458.

Dellarocas, C. (2000). Immunizing online reputation reporting systems against unfair ratings and discriminatory behavior. Proceedings of the Second ACM Conference on Electronic Commerce, Minneapolis, Minnesota, October 17-20.

Dellarocas, C., & Resnick, P. (2003). Online reputation mechanisms: A roadmap for future research. A summary report of the symposium on online reputation mechanisms, April 26-27, MIT, Cambridge, Massachusetts.

Friedman, E., & Resnick, P. (1998). The social cost of cheap pseudonyms. An earlier version was presented at the Telecommunications Policy Research Conference, Washington DC, October.

Ganesh, L., & Amit, P. (2003). Internet shopping agents: Virtual co-location and competition. Marketing Science, 22(1), 85-107.

Head, M., & Hassanein, K., (2002). Trust in e-commerce: Evaluating the impact of third party seals. Quarterly Journal of Electronic Commerce, 3(3), 307-325.

Kollock, P. (1999). The production of trust in online markets. In Advances in Group Processes, (16). Greenwich: JAI Press.

Lang, D., & Oshima, M. (1999) Seven good reasons for mobile agents. Communications of the ACM, 42(3), 88-89.

Liu, S. (1998). Strategic scanning and interpretation revisiting: Foundations for a software agent support system – Part 2: Scanning the business environment with software agents. Industrial Management & Data Systems, 98(8), 362-372.

Maes, P., & Guttman, R. (1999). Agents that buy and sell. Communications of the ACM, 42(3), 81-90.

McGaughey, R., & Mason, K. (1998). The Internet as a marketing tool. Journal of Marketing Theory and Practice, 6(3), 1-11.

Mougayar, W. (1998). Opening digital markets. Battle plans and business strategies for Internet commerce. New York: McGraw-Hill.

Oz, E. (2002). Foundations of e-commerce. Englewood Cliffs, NJ: Prentice Hall.

Park, C., & Lessig, P. (1981). Familiarity and its impact on consumer decision biases and heuristics. Journal of Consumer Research, 8(2), 223-30.

Resnick, P., Zeckhauser, R., Friedman, E., & Kuwabara, K. (2000). Reputation systems. Communications of the ACM, 43(12), 45-48.

Ruyter, K., Wetzels, M., & Kleijnen, M. (2001). Customer adoption of e-service: An experimental study. International Journal of Service Industry Management, 12(2), 184-207.

Schneider, G. P., & Perry, J. T. (2002). Electronic commerce (3rd ed.). Course Technology.

Siebel, T., & Hous, P. (1999). Cyber rules. New York: Currency/Doubleday.

Taylor, J. (1974). The role of risk in consumer behavior. Journal of Marketing, 38, 389-398.

Vulkan, N. (1999). Economic implications of agent technology and e-commerce. Economic Journal, 109(453), 67-91.

Wang, M., & Christopher, D. (2003). The consistency of customer online ratings of etailers on shopping agent Web sites. Research of Business Review, 1(1), 14-20.

Wang, M., & Huarng, S. (2002). An empirical study of Internet store customer post-shopping satisfaction. Issues in Information Systems, 3(1), 632-638.

Zacharia, G., Moukas, A., & Maes, P. (2000). Collaborative reputation mechanisms for electronic marketplaces. Decision Support Systems, 29(4), 371-388.

Zeithaml, V. (2002). Service excellence in electronic channels. Managing Service Quality, 12(3), 135-139.

Web Systems Design and Online Consumer Behavior
Web Systems Design and Online Consumer Behavior
ISBN: 1591403278
EAN: 2147483647
Year: 2004
Pages: 180 © 2008-2017.
If you may any questions please contact us: