Patrick Barwise. Handbook of Marketing. Editor: Barton A Weitz & Robin Wensley. 2002. Sage Publication.
Introduction: The Promise of Digital Marketing
How Important is the Internet in Marketing?
The Internet has become one of the most discussed topics in business and academia. The speed of development of electronic marketing has been fast by any standards, and especially compared with the slow process of academic research and publication in marketing and other social sciences.
At the time of writing, however, business and financial markets are recovering from earlier overexuberance about the Internet. It is now recognized that the fundamentals of business have not changed; that most pure-play dotcom business models were wildly over-optimistic, especially in business-to-consumer (B2C) markets; and that the future role of the Internet in marketing will be largely as part of an integrated combination of ‘bricks and clicks.’
Despite this necessary reassessment, the Internet remains the most wide-ranging and significant area of current development in marketing:
- It allows faster, cheaper, more personalized interactions (and raises more privacy concerns) than any previous medium.
- It can dramatically reduce customer search costs and even support purchase decisions made on behalf of the customer by intelligent software agents.
- It allows seamless communication over any distance, local or global. Eventually, it will also support effective automatic language translation.
- It is becoming ubiquitous, allowing ‘24×7’ communications with customers at home, at work, at the point of purchase, on the road, or anywhere else—including location-sensitive communications.
- Increasingly, it is evolving beyond a single, limited channel (a PC connected to a regular telephone line) to exploit a range of new, high-capacity fixed and mobile networks and ‘convergent’ devices such as interactive digital television (iDTVs), online games computers, next generation cellphones and PDAs, in-car telematics, online vending machines, and utility meters.
These capabilities have the potential—in principle -to transform many aspects of marketing: segmentation and targeting, bundling, pricing, customer service and customer relationship management, marketing communication, promotion, channels and value chains, brand communities, global marketing, and the importance of brands.
Marketers are all still trying to discover the long-term implications for strategy and execution, but it is possible to draw some initial conclusions, including on many of the topics just listed. Our overall assessment is that, despite the earlier hype, the Internet remains the most important development in B2C markets since the growth of television and supermarkets 50 years ago, and the most important in business-to-business (B2B) markets since the railroad and telegraph 100-150 years ago.
Hoffman (2000: 1) described the Internet as ‘the most important innovation since the development of the printing press,’ with the potential to ‘radically transform not just the way individuals go about conducting their business with each other, but also the very essence of what it means to be a human being in society.’ Peppers and Rogers (1993, 1997) argued that digital marketing represents a complete transformation of the marketing paradigm, from a predominantly one-way broadcast model to a model of totally interactive, totally personalized one-to-one relationships. However, the extent to which digital media such as the Internet will revolutionize business, home life, and the relationship between marketer and consumer is still controversial. Earlier innovations such as the electric telegraph, the railroad, electricity, the telephone, the automobile, the airplane, radio, and television have all had widespread impact on both business and everyday life, although perhaps only electricity quite matches the combined speed and scale of the Internet’s impact (Barwise & Hammond, 1998).
Many of the features associated with the Internet have appeared before in the context of technologies such as the electric telegraph (Standage, 1998) and radio (Hanson, 1998, 2000). Ethnographers such as Venkatesh (1985) have long studied people’s everyday use of technology in the home. Mick and Fournier (1998) and Fournier, Dobscha and Mick (1998) found that much so-called technophobia among consumers is entirely rational and based on their previous experience of technology making life more complicated—not simpler, as claimed. Certainly many of the early claims about the likely speed of the Internet’s impact, for example on the use of other media (Gilder, 1994; Negroponte, 1995), have turned out to be wide of the mark. Exaggerated visions of a wired society go back at least to E.M. Forster writing in 1909 (Baer, 1998).
What is clear is that the Internet combines many of the features of existing media with new capabilities of interactivity and addressability, as well as making it much easier for both companies and individuals to achieve a global reach with their ideas and products. By the start of the twenty-first century it had already been adopted on a massive scale, especially in North America, Australia and Northern Europe, and its effects will continue to be felt in almost every market and on almost every aspect of marketing. These impacts range from the most micro (such as ad agencies needing to experiment with the design of interactive advertisements) through to the most macro (such as whether corporate profitability will be lower in ‘frictionless’ markets).
A few researchers were quick to recognize the potential of new interactive media to affect all aspects of marketing. Blattberg and Deighton (1991: 8) noted that, by the early 1990s, technology was already allowing interactive marketing to take place to individually identifiable consumers: ‘When a firm can go back to a customer to respond to what the customer has just said, it is holding a dialogue, not delivering a monologue.’ They noted that good database design was key, that profiles of customer histories should be collected, and that privacy concerns would grow. Deighton built on this work by gathering together a collection of thoughts from marketing academics and industry experts on how interactivity might reshape the marketing paradigm (Deighton, 1996). He also foresaw the emergence of a new marketing paradigm that would bring about a convergence between consumer marketing and business-to-business marketing (Deighton, 1997). He suggested that
the discipline of marketing, whose stock of knowledge amounts to a fund of insights on how to compensate for the imperfections of two kinds of tools—broadcast tools and sales agent tools—now has a new tool without some of those imperfections but with a whole new set of imperfections yet to be discovered. (Deighton, 1997: 348)
Hoffman and Novak (1996, 1997) identified the unique ‘many-to-many’ property of computer-mediated environments such as the web. They suggested that marketing activities will be difficult to implement in their traditional form, and predicted the evolution of a marketing paradigm compatible with the increased role of the consumer and of interactive technologies.
Other researchers have suggested similar far-reaching changes in business: Haeckel (1998: 69) proposed that the collaborative potential of information technology and the Internet might recast business as ‘a game with customers, rather than … a game against competitors.’ Taking a similar view, Achrol and Kotler (1999) suggested that, as the hierarchical organizations of the twentieth century disaggregate into a variety of network forms, customers will enjoy an increasing capacity to become organized, with marketers becoming agents of the buyer rather than the seller, and adopting the role of customer consultant rather than purveyor of goods and services. They cite examples such as Baxter Travenol in hospital supplies, McKesson in pharmaceuticals, Travelocity in flight booking, and Amazon (Achrol & Kotler, 1999: 157-8). Similarly, Nouwens and Bouwman (1995) explored the conditions likely to give rise to ‘network organizations’ such as the recorded music retail industry. They concluded that the role of a system integrator is crucial, but that a dominant actor (e.g., a copyright holder) could hinder the effective use of industry-wide communication networks.
In exploring the research agenda for marketers, Winer et al. (1997) examined the potential for marketing research associated with computer-mediated environments (CMEs). They suggested that the CME provides a new context in which to study existing theories as well as being an entirely new phenomenon meriting research in its own right. They identified five key areas of choice research likely to be impacted by the development of CME technology: decision processes, advertising and communication, brand choice, brand communities, and pricing. This review chapter covers research on all five, as well as on other topics.
Aims and Scope of this Chapter
Our aim is to review the research to-date on how the Internet is impacting marketing. The speed of development and the shortage of new theory to support hypothesis-testing necessarily mean that our review deals more with empirical research than with theory. There has, however, been a growing stream of theoretical development and discussion, from early essays on marketing in an information-intensive environment (Glazer, 1991), on the nature of interactivity in both marketing (Blattberg & Deighton, 1991; Rust & Oliver, 1994) and communication (Dutton, 1996; Morris & Ogan, 1996; Neuman, 1991; Pavlik, 1996), through ethnographic work which explores household-technology interactions (Venkatesh et al., 1996) to the work of Hoffman and Novak (1996)—the main pioneers of Internet research in marketing—who proposed one of the first models of consumer behavior in computer mediated environments—and to Bakos and Brynjolfsson (1999, 2000a, 2000b), and their work on online pricing strategies.
The Internet impacts marketing strategy, channel management, pricing, marketing communications, customer service, decision support systems, database marketing, global marketing, and business-to-business marketing. We here focus on research which looks at how the Internet is or can be used by both firms and consumers to support the marketing process. We concentrate on research in consumer behavior, advertising, pricing, channels, and marketing strategy, because empirical research is most advanced in these areas. We aim to provide an overview of the current state of research and briefly to identify opportunities for further work. We do not cover the extensive literature on supply chain management, information management, organizational behavior, or the broader impact of the Internet on productivity, profitability, employment, and international trade.
We searched for relevant literature in a number of ways. First, we used databases such as JSTOR, ABI Inform/ProQuest Direct and Web of Science, searching for key terms covering the topics under investigation. Second, we browsed volumes of relevant marketing journals (e.g., the International Journal of Research in Marketing, Journal of Consumer Research, Journal of Electronic Commerce, Journal of Interactive Marketing, Journal of Marketing, Journal of Marketing Research, Management Science, and Marketing Science) published after 1994—roughly the time at which articles about the Internet started to appear. Third, we searched for articles on the Internet, usually starting at popular search engines, web pages published by academic institutions (e.g., the MIT E-commerce Forum, Wharton’s Forum on Electronic Commerce, and references on UCLA’s Anderson School website), as well as individual researchers’ personal web pages. Fourth, we collected literature based on suggestions by colleagues who had read an earlier draft of our manuscript (which we had made publicly available on the web). In our search, we did not limit ourselves to published articles, book chapters and books; given the dynamic nature of the topic, we also sought to include working papers and manuscripts under review. In total, we collected nearly 400 publications.
The rest of this chapter is organized as follows. Section 2 concerns Internet adoption, usage and the consumer’s experience. Section 3 builds on this to discuss the evidence on consumers’ online purchasing behavior. Section 4 covers Internet advertising. Section 5 then explores wider issues related to Internet economics and pricing, including the evidence on whether the Internet is, as is often claimed, leading to ‘frictionless’ markets characterized by fierce price competition. This leads into Section 6, on the impact on channels and intermediaries. Section 7 reviews the strategies and business models that firms are developing in response to the new threats and opportunities created by the Internet. Finally, Section 8 briefly summarizes the chapter and discusses future prospects and research opportunities.
Internet Adoption, Usage, and the Consumer Experience
In this section we start by reporting research on the factors that influence customers’ adoption and usage of the Internet in general. We next discuss its adoption specifically for e-commerce, and factors such as site design, service quality and fulfillment, that can affect its continued use as an e-commerce channel. Finally we introduce the relevant literature on software agents.
Predictors of Internet Adoption and Usage
The use of the Internet as a marketing channel depends both on the growth in general Internet penetration and usage, and on how the Internet then influences the adoption and diffusion of other products and services. In both these contexts, there is the usual caveat that one must be careful not to assume that the predictors of early adoption will also hold for later adopters.
Rangaswamy and Gupta (1999) draw on the adoption and diffusion literature to propose how the Internet will develop. They use the GVU database2 and other surveys to explore early adopters’ attitudes toward the Internet and perceptions of online versus offline vendors. They found that among very early adopters, especially those who were heavy users, the primary reason for using the web was for shopping.
Hammond et al. (2000) compared Internet users and non-users to determine if there were differences between these two groups in their attitudes toward technology, ownership of different technologies, and information versus entertainment needs. They found that, compared to non-users, Internet users were more interested in technology in general and the benefits it provided, especially if they thought it would save them effort or time. They were also more likely to think that technology was both important and fun. Internet users were primarily motivated by communication/information needs but this did not appear to be because they felt ‘time-pressured’ compared to non-users. A tentative implication of these findings is that the Internet may need to become more entertainment-oriented in order to attract a broader active user base.
Emmanouilides and Hammond (2000) used logistic regression to explore four successive waves of survey data on Internet users. They found that the main predictors of active or continued use of the Internet were: time since first use (very early adopters were the most likely to be active users, but this relationship was curvilinear, with middle adopters more likely than other groups not to have used the Internet in the previous month); location of use, particularly at home; and the use of specific applications, such as information services. However, the main predictors of frequent or heavy Internet use were: use of email for business purposes; time since first use of the Internet; and location of use (either at work or at home with two or more other people).
As the Internet matures as a consumer medium we begin to see studies which evaluate how it is used and what makes for a compelling experience. Novak et al. (2000) built on Hoffman and Novak’s (1996) discussion of ‘flow’ to develop a model of the components of a compelling online experience. The model was validated using a web-based consumer survey. A compelling experience was found to be positively correlated with fun, recreational and experiential uses of the web, with expected use in the future, and with the amount of time consumers spend online, and negatively associated with work-related usage. Faster download and interaction were not, per se, associated with a compelling experience.
Other evidence supports the importance of the fun or hedonic aspect of Internet use. Hammond et al. (1998) explored the differences between novices and more experienced users and their appreciation of the web’s informational and entertainment value. They found that, while prior experience was an important moderator of users’ attitudes towards the web, its influence was nonlinear. The heaviest users were enthusiasts for the medium, while moderate and light users perceived it as a source of information, but not for entertainment or fun.
Exploring the different reasons people have for using the Internet, Hoffman et al. (2000) used the locus of control (LOC) construct to study differences in usage between those with an internal versus an external focus. They found that those who use the web as a substitute for other activities tend to have an external orientation, whereas those with an internal focus use it in a more goal-directed manner as a supplement to other information gathering activities.
Applying the uses and gratifications perspective to web users, Eighmey and McCord (1998) found similarities to the uses and gratifications reported for other media. However, there were additional factors related to personal involvement and continuing relationships that were associated with users’ reactions to websites. As Venkatesh (1998) notes, consumers not only consume new technology and its products, but are also consumers of the market processes, which are themselves affected by new technology, and all these processes can affect the social order. Related to the experience of consumers in online environments, Gould and Lerman (1998) analyzed early consumer-to-consumer (C2C) exchanges on an online discussion forum, NetGirl, in a bid to describe and understand, in phenomenological terms, the consumer experience.
These adoption and usage studies treat the Internet as an additional medium in consumers’ lives. We now turn to research which specifically evaluates its adoption and use as a channel for commerce.
Adoption of the Internet as an E-Commerce Channel
Hoffman et al. (1995) were the first researchers to propose a framework for examining the commercial development of the web. They explored its role both as a distribution channel and as a medium for marketing communication, evaluated the resulting benefits to consumers and firms, and discussed the barriers to its commercial growth from both supply-and demand-side perspectives. Their classification scheme categorized websites as either destination sites (online storefronts and other content sites), or as web traffic control sites (malls, incentive sites, and search agents). They proposed that the interactive nature of the web freed customers from their traditional passive role as receivers of marketing communications, giving them access to greater amounts of dynamic information to support decision-making.
In an early study comparing web shopping with buying through other channels, Palmer (1997) tested the buying of 120 different products across four retail formats: in store, catalog, cable television, and the web. Total product cost did not vary significantly between the four formats, but there were significant differences in product description, availability, delivery, and time taken to shop.
Consumer substitution between online, traditional retail, and direct mail has been explored by Ward and Davies (1999) using a transaction cost approach. Ward and Davies developed a model to investigate distribution channel choice and tested it using survey data, finding that consumers considered online shopping and direct marketing to be closer substitutes than either of them with traditional retail. Degeratu et al. (1999) hypothesized how online and traditional grocery stores differ in their influence on consumer choice. They found that brand names were more valuable online in categories where information on fewer product attributes was available; that ‘non-sensory’ attributes (e.g., the fat content of margarine) had more impact on online choice than ‘sensory’ ones (e.g., visual cues such as paper towel design); and that price sensitivity was higher online because online promotions were stronger signals of price discounts. The combined effect of price and promotion on consumer choice was found to be weaker online than offline.
Other research has covered the battle between bricks-and-mortar and Internet companies. Interesting in this regard is a study by Goolsbee (2000) on competition in the computer industry. Based on Forrester Research Survey data covering 90,000 households, he constructed a price index measuring the offline costs of a computer in different cities, and then calculated how likely a computer buyer would be to purchase online, taking into account prices of computers offered by offline retailers. He found that if local (offline) and online prices were initially equal, and local prices were to rise by 10%, bricks-and-mortar retailers would see their share of unit sales fall from 68% to 63%, assuming the total volume of purchases remained unchanged. Although the precise effect may vary among sectors and products, and the findings indicate that online and offline purchases are to some extent substitutes, these findings suggest that there is less rivalry between offline and online retailers than between retailers operating nearby stores.
Building on the channels literature, Morrison and Roberts (1998) explored the determinants of consumer consideration of new delivery channels for existing services in terms of preference for the service, preference for the delivery method, and the fit between the service and the delivery method. They suggested that adoption of new channels was being slowed by consumer doubts over the relative advantage of these new channels, and also the lack of fit between the service (their example was banking) and the new channel. They concluded that consumer e-commerce firms need to spend more effort explaining to consumers the fit and appropriateness of new delivery methods for their goods and services.
Using US survey data from the GVU Center, Bain (1999) examined factors related to the adoption of the Internet as a purchase medium. A strong (negative) relationship was found between consumer perceptions of the risk of web-shopping and purchase behavior, but not between perceptions concerning information privacy and purchase behavior. This suggests that consumers will not buy online if they are worried that there is a financial risk but they are less concerned about their personal details being kept, used, or traded by retailers.
The most likely candidates to be early adopters of retailer websites have been found to be consumers who were familiar with the Internet and already used to other modes of home shopping (Balabanis & Vassileiou, 1999). Consumers in higher income brackets were more enthusiastic about shopping on the Internet, but for this segment strong brands were key to shopping online. Li et al. (1999) found that education, experience, views on convenience, channel knowledge, perceived distribution utility, and perceived accessibility were robust predictors of the extent to which an Internet user was a frequent online buyer.
Bellman et al. (1999) surveyed over 9,000 online users and used logistic regression to identify factors that predicted whether an online user bought products online, and if so, how much they spent. The best predictors were ‘time starvation’ (how many hours a week the user worked) and the extent of their ‘wired’ lifestyle. Lohse et al. (2000) re-surveyed the same respondents to test whether and how their attitudes and behavior had changed over time. They found that the average annual spend per purchaser had increased over time; time starvation and a wired lifestyle were still major determinants of the amount of online spending; but time starvation no longer appeared to influence whether a person chose to buy from an online store. A further finding was that, while the percentage of respondents who made a purchase online increased over time, a significant minority (14%) who had previously bought online had ceased to do so. The authors suggest that this was mostly due to respondents having had bad experiences with online retailers.
Swaminathan et al. (1999), using an email survey of GVU Center respondents, found that concerns about privacy and security had minimal effect on consumers’ online purchasing behavior, but that those who purchased frequently online were more likely to support new laws to protect privacy (perhaps because they knew more about what companies do with the data). The main determinant of online purchase frequency was perceived vendor characteristics, especially price competitiveness and the ease of canceling orders. Another factor was that consumers motivated by convenience (mostly men) were more likely to buy online than those who valued social interaction (mostly women).
Another key factor that has been found to determine whether consumers buy online is whether this fits into their lifestyle, and the extent to which they perceive it as easy and convenient (Becker-Olsen, 2000). She found that those not buying online felt that traditional shopping was easier, quicker, cheaper, and more convenient for their particular lifestyle. Even those who did buy online did not perceive it as quicker or less expensive, nor did they feel that they received better service. Neither group (online buyers and non-buyers) seemed strongly concerned with security risks, although the overall credibility of the company/site was seen as important, especially by those who had not purchased online. Other factors were the need to see/touch the product (in some categories) and consumers’ need to have the product immediately.
Service Quality, Fulfillment, and Site Design
Building on Hoffman et al.’s (1995) framework for classifying Internet commerce sites, Spiller and Lohse (1998) surveyed 44 website features across 137 women’s apparel retail sites. Using cluster and factor analysis they identified five distinct web catalog interface types: superstores, promotional stores, plain sales stores, one-page stores, and product listings. Differences between online stores centered on size, service offerings, and interface quality.
Again, building on the Hoffman and Novak framework, it has been proposed that there are three major components to a consumer’s online shopping experience: interface quality, encounter quality, and fulfillment quality (Chang, 2000). We can think of these as: process, experience, and results, although in the online world these elements are often intertwined more than in traditional retail channels. To build brand equity, it is suggested that firms need to ensure excellence on all three dimensions. We now consider empirical research findings in these areas.
One of the key aspects of the online experience investigated by researchers is the time taken for consumers to access the information they require. Dellaert and Kahn (1999) investigated whether the waiting time experienced by consumers (e.g., while pages downloaded) affected subsequent evaluation of the web content. They found that the potential negative effects of waiting can be neutralized by improving the waiting experience. Consumers may feel negative effect as a result of waiting, but this did not necessarily impact on their evaluation of the web material itself, as long as the waiting time was signaled and expected.
This finding was amplified by Weinberg (2000), who showed that the perceived waiting time could be significantly influenced by providing a waiting time anchor lower than the actual waiting time. In an experiment where all subjects experienced an actual wait of 7.5 seconds, those given the message ‘Please wait about 5 seconds’ on average estimated it was 5.6 seconds; those told ‘Please wait about 10 seconds’ estimated 8.7 seconds. In a second experiment, subjects exposed to a 5-second wait anchor rated the quality of the homepage higher, and were more likely to continue searching the website, than those exposed to a 10-second wait anchor.
Related to consumer perceptions of the time they spend waiting for information to download is the time spent waiting for a response to an information request. Voss (2000) reported a survey of different sites’ responsiveness to email inquiries. It was found that, in both the US and Britain, websites’ response tended to be poor for web startups and even worse for established businesses. There was, however, wide variation, with some sites never responding and others having an excellent auto-acknowledge function followed by relatively fast full response. This paper proposed a set of key metrics in areas such as trust, response time, response quality, and navigability. Similarly, in the context of complaints, Strauss and Hill (2001) found that customer satisfaction was enhanced by quick responses to complaint emails. They, too, found a huge range in company response.
As many web startups discovered, service quality also includes fulfillment, which, for physical products, requires expensive bricks-and-mortar logistics. In many markets this can be subcontracted at low cost, but in some, notably groceries, home delivery is likely to be the limiting factor. Two articles in McKinsey Quarterly (Barsh et al., 2000; Bhise et al., 2000) and another in the Booz Allen Strategy and Business(Laseter et al., 2000) explored the issue of fulfillment. Barsh et al. describe how most e-tailers lose money on every transaction. The articles conclude that successful online retailing requires a scalable national sales and distribution channel, and that the main discriminator between those companies that succeed and those that fail will be large order volumes and deep reserves of capital.
The online experience has also been related to interface quality, specifically to website design. Ghose and Dou (1998) found that the more interactive the website, the more likely it is to be rated a ‘top site.’ Against this finding, however, are the crucial issues of simplicity, navigability, and especially the download and response times, as described above.
Bellman and Rossiter (2001) introduce and test the concept of a web ad schema, defined as the consumer’s set of beliefs about information locations, and routes to those locations, for a web advertising site. They argue that consumers already have well-developed schemata for finding useful information in traditional media (e.g., they expect to find detailed product specifications towards the back of a brochure, and to see or hear the brand name in the final frames of a TV commercial). But on the web, some consumers are much better than others at finding information from an advertising website. Bellman and Rossiter contend that these adept consumers have web ad schemata which are both well developed and congruent with the structure of the site. Their paper reports a series of three studies which support the existence of web ad schemata and their influence on communications effectiveness (brand knowledge). The results also suggest that these schemata are motor-associative -acquired by exploring the site one click at a time -rather than map-like. The implications are first, that the site structure should be as simple as possible, and second, that (except for a very large site) in-site navigation should also be simple and motor-associative, providing ‘local’ information about the pages immediately accessible from the current page and not a ‘menu’ of the overall site structure, which is the usual approach (Hofacker, 2001). Although this research focuses only on advertising sites, the conclusions about navigability seem likely to generalize to other types of site.
Research by Mandel and Johnson (1999) showed that even minor peripheral cues such as background color and pictures could influence consumers’ response to a site. For example, in one experiment, subjects who were shown a ‘money’ background gave more weight to price when asked to evaluate products than those who were shown a different background. Such priming effects were found to affect both search order and choice.
Another element of site design is the tools that are provided to help consumers choose between different products/brands. Building on the extensive literature on consumer information processing and consideration sets, Häubl and Trifts (2000) explored the impact that interactive decision aids have on consumers’ decision-making. In a controlled experiment they introduced users to two decision aids. The first was a recommendation agent to allow consumers to screen a large set of alternatives and reduce these to a short list or consideration set. The second, a comparison matrix, helped users make in-depth comparisons among the selected alternatives. The study found that both these interactive decision aids had an impact on consumer decision-making, enabling consumers to make better decisions with less effort. The findings also suggest that the use of such tools can lead to an increase in the quality (but decrease in the size) of consumers’ consideration sets.
Also on this topic, early work by Widing and Talarzyk (1993) on how consumers use decision aids suggested that not only did they like computer assistance and feel that it helped them to make better decisions, but that aids which enabled them to weight the importance of rated attributes (such as ease of use, or performance), and obtain a personalized rank order of brands, were more useful than other types of decision aid.
Ariely (2000) used a series of experiments to explore the characteristics and likely impact of information control on consumers’ decision quality, memory, knowledge, and confidence. He found that controlling the flow of information can help consumers better match their preferences, have better memory and knowledge about the topics they are exploring, and be more confident in their judgments. However, he warns that controlling the information flow can create additional demands on the consumer’s ability to process information, and so may not be suitable for situations where the task is difficult or novel.
Almost all the previous research focuses specifically on B2C interactions. Berthon et al. (1998) developed a conceptual framework for evaluating web communication activities (e.g., number of active users as a percentage of hits, number of purchases as a percentage of active users, etc.), and suggested how this might be applied to the B2B industrial purchasing decision-making process.
Devices such as the recommendation agents in Häubl and Trifts’ (2000) study, discussed above, focus on improving the navigability and convenience of choosing a product or service supplied by the site owner. We now turn to the more advanced intelligent agents which act on behalf of individual consumers, knowing their preferences and searching the web for products that best meet their needs. Such an agent could significantly influence brand choice and/or which distribution channel is chosen to supply a particular brand. If requested, it can negotiate price and/or delivery, possibly by inviting suppliers to bid (‘reverse auction’) and/or by forming a cartel with other consumers (or their agents) to negotiate the best price (Dolan & Moon, 2000). Finally, the agent may be empowered in some contexts to make the actual purchase on behalf of the consumer. Initially, agents (or ‘bots,’ e.g., ‘shop-bots’ in the case of those aimed at shopping applications) were controlled through the consumer’s PC. Increasingly, they will also be controlled through mobile phones and interactive digital TVs, and eventually by voice rather than a keyboard or keypad (Barwise & Hammond, 1998).
The best known center for research on agent technology is the Media Laboratory at MIT. Patti Maes, who heads its software agent group, expects agent technology to have dramatic effects on the US economy (Maes, 1999). She predicts the disappearance of existing intermediaries and the emergence of some new types of intermediary, a reduction in the capital required to set up a business (and therefore more scope for small niche businesses), and efficiency gains for both buyers and sellers by dramatically reducing marketing and selling costs, but with a clear overall shift in the balance of power from sellers to buyers.
Guttman et al. (1998) summarize the Media Lab’s research on e-commerce agents. These include: C2C ‘smart’ classified ads, merchant agents that provide interrogative negotiation, agents that facilitate expertise brokering and distributed reputation facilities, agents for point-of-sale comparison shopping, and agents for mobile devices. The Media Lab research has focused on the development, prototyping and evaluation of the agent technology itself, including the launch of an agent-based business, http://Firefly.com, which was bought by Microsoft in 1998 but shut down in 1999 in preparation for Microsoft’s Passport service. Other researchers have sought to explore the potential and likely impact of agent technology from a marketing perspective (discussed below). Some of this research overlaps with work on buyer search costs and information economics, and on disintermediation, discussed in Sections 5 and 6.
Recommendation systems are assessed by Ansari et al. (2000) who investigate the two main methods of gathering recommendations: collaborative filtering (based on other users’ weighted preferences) and content or attribute-based filtering (based on information provided by the user). Researchers at the Sloan School at MIT have investigated the use of recommendation agents as ‘trust-based advisors’ in both B2B and B2C contexts. Urban et al. (1999) described a prototype trusted agent system, and reported that consumers who are not very knowledgeable about the product, who visited more retailers, and who were younger and more frequent Internet users, had the highest preference for a virtual personal advisor.
Other researchers have proposed frameworks for thinking about the design of electronic agents. Gershoff and West (1998) and West et al. (1999) suggest a set of goals for agent design that include both outcome-based goals (e.g., improving decision quality) and process goals (e.g., increasing consumer satisfaction and trust). Kephart et al. (2000) describe an IBM research program into the potential impact of dynamic pricing agents on the economy. Iacobucci et al. (2000) focus on intelligent agents that compare a user’s profile to data on other users to determine which users in the database are similar, in order to develop relevant recommendations of value to the focal user. Iacobucci and her colleagues characterize this as ‘rediscovering the wheel’ of cluster analysis and therefore draw from the cluster analysis literature to begin to address the questions being posed in this new application area.
All these decision-making aids and recommendation agents are provided with the aim of supporting purchase and repeat-purchase. The next section reviews the evidence on consumer purchase behavior online.
Online Purchasing Behavior
Retailer/Brand Choice and Loyalty on the Web
The impact that interactive shopping might have on consumer behavior, and therefore on retailer and manufacturer revenue, was addressed by Alba et al. (1997). They considered the relative attractiveness to consumers of alternative retail formats. They noted that technological advances offered consumers unmatched opportunities to locate and compare product offerings, but that price competition may be mitigated by the ability of consumers to search for more differentiated products better fitted to their needs. The authors examined the impact of the Internet as a function of both consumer goals and product/service categories, and explored consumer incentives and disincentives to purchase online versus offline. They also discussed implications for industry structure in terms of competition between retailers, competition between manufacturers, and retailer-manufacturer relationships. They concluded with a list of research questions raised by the advent of interactive home shopping.
Continuing the theme of the impact that interactive shopping might have on consumers, Peterson et al. (1997) suggested that many predictions regarding the growth of the Internet for consumer retail are overstated because they failed to consider the heterogeneity and complexity of consumer markets. Peterson et al. analyze channel intermediary functions that could be performed on the Internet, classify its potential impact by category type, and discuss how price competition might evolve. They propose a consumer decision-making process that takes account of the Internet as both an information channel and a purchase channel. They give a detailed list of questions to motivate and guide the development of research into the use of the Internet and its implications for marketing theory and strategy.
As Rubini et al. (1996: 70) argued, even with the arrival of new technology, ‘the process of shopping remains unchanged. The underlying basis for all retail, whether physical or virtual, remains a problem-solving process with a value-for-value exchange at its heart.’ They discuss consumer shopping behavior from a design perspective, stressing the role played by problem recognition, search and evaluation of alternatives, navigation, purchase, content, identity, infrastructure, and social space. They propose further exploration of these issues using a test virtual world through which shoppers navigate. Such a world is explored by Burke (1996), who described early test virtual stores as used by a tire company, a snack-food maker, a frozen-food company, and a fast-food restaurant chain. Burke (1997) reviewed the ways in which conventional retailers can enhance the shopping experience for their customers using both in-store and virtual applications. He addresses the potential weaknesses of the Internet as a shopping medium and suggests how these might be overcome.
Moving on from the decision whether to shop online, to which retailer site to patronize, Ilfield and Winer (2001) explore the decision-making process behind consumers’ website choices and the relative effects of the communication channels which aid this process. They compare the traditional ‘persuasive’ hierarchy of effects (think-feel-do) model with a ‘low involvement’ (think-do-feel) model and a ‘no involvement’ (do-think-feel) model. The models were tested across a wide range of Internet companies. The data include spending by each company on a variety of online and offline media, measures of website attractiveness, number of visits and page views, public relations mentions, website quality, links, and whether the firm had a non-Internet presence. The low involvement model gave the best fit. The authors concluded that high brand awareness is essential for an Internet firm’s survival.
In line with this conclusion, Adamic and Huberman (2000) studied the distribution of website visitors by examining usage logs covering 120,000 sites. They found that, both for all sites and for sites in specific categories, the distribution of visitors per site follows a universal power law similar to that found by Pareto in income distributions. This implies that a small number of sites command the traffic of a large segment of the web population, which suggests that only a few winners will emerge in each market.
Once the prospective customer has visited a website, managers wish to maximize the breadth (time spent on a site) and depth (number of pages viewed) of site visits, plus the repeat-visit rate and, of course, where appropriate, the amount of money spent per customer. Evidence on the importance to consumers of the perceived value of time can be found in a study by Lohse and Spiller (1999) which assessed different features of website interface design that affect store traffic and amount spent. Product list navigation features that save consumers time online (i.e., reduce the time to purchase) accounted for 61% of the variance in monthly sales.
Several researchers have applied traditional repeat-buying models to the investigation of online brand loyalty in order to evaluate customer lifetime value and to test whether consumer loyalty operates in a similar manner online compared with offline. Fader and Hardie (2001) presented a non-stationary experiential gamma (NSEG) model, similar to a negative binomial distribution (NBD) model but with dynamic individual-level buying rates. The NSEG model outperformed the NBD in the context of repeat buying on the Internet, in terms of both forecast accuracy and parameter stability across calibration periods of different lengths. Fader and Hardie describe a modeling exercise applied to repeat sales at the online music retailer CDNOW. Their main finding is that most of CDNOW’s sales growth was due to a constant stream of new customers rather than from earlier trialists increasing their loyalty (measured as share of category requirements) over time. The implication is that it will be hard for online stores to sustain their earlier rapid sales growth since most consumers who use them will continue to do so in combination with established channels, rather than gradually switching over to purchasing exclusively online.
Using similar techniques, Moe and Fader (2000) developed an individual-level model for online store visits based on Internet clickstream data. This model captured cross-sectional variation in store-visit behavior as well as changes over time as consumers gained experience with the store. The results confirmed that people who visit a store more often are, as is widely assumed, more likely to buy. However, changes in an individual’s visit frequency over time can also provide additional information about which consumer segments are more likely to buy. The implication is that marketers should use a sophisticated segmentation approach that incorporates how much an individual’s behavior is changing over time, rather than simply targeting all frequent shoppers.
Other modelers include Wu and Rangaswamy (1999), who used data from online grocer Peapod. Their model uses Fuzzy Set Theory to capture the two-stage process of (1) consideration set formation, and (2) evaluation of alternatives in the consideration set. They showed that previous models had failed to capture the richness of the choice processes that are increasingly feasible for consumers in online markets.
A key question for retailers and brand managers is whether consumers exhibit greater or lesser brand loyalty online compared with offline. Danaher et al. (2000) compared consumer brand loyalty (share of category requirements and average purchase frequency) for packaged goods purchased online versus at traditional grocery stores. They used a segmented-Dirichlet model with latent classes for brand choice to compensate for systematic deviations in the offline retail setting. This provided the benchmark against which loyalty was tested in the online setting. Brand loyalty for high market-share brands was significantly greater in the virtual environment (a ‘winner-take-all’ effect), with the reverse being the case for low-share brands. In an online purchase setting, ‘niche’ brands (those with a small but highly loyal following) did better than expected, while ‘change-of-pace’ brands (those bought infrequently but by a large segment of the relevant population) did worse than expected.
Brands, Trust, and Customer Relationships
In the context of the Internet, ‘the brand is the experience and the experience is the brand’ (Dayal et al., 2000: 42). As in physical environments, the key goal of online marketing is to use the web, usually in combination with a range of other channels and activities, to build a positive and profitable long-term relationship with the customer.
Steinfield et al. (1995) argued that B2B electronic networks can be used either to support transactional marketplaces or to strengthen commercial relationships. Their review of the literature suggests that the latter relational (‘electronic hierarchies’) approach is more prevalent. They examined the theoretical rationales behind these competing approaches and presented evidence on the conditions under which electronic marketplaces or electronic hierarchies are likely to prevail. Their conclusions are supported by Bauer et al. (1999), who focused on the contribution the Internet can make to relationship marketing, and especially to commitment, satisfaction and trust. This paper provided empirical evidence that consumers’ trust is reduced if their expectations are not met.
There is, however, also evidence that the Internet is used by consumers as ‘merely’ an alternative shopping channel rather than a means of strengthening their relationships with brands or other consumers. Becker-Olsen (2000) found that, among those who had bought online, the most important factors determining their purchase behavior were the ability of the site to load quickly, availability of familiar brand names, and a clear return policy. These findings throw some doubt on the conclusions of Steinfield et al. (1995), and Bauer et al. (1999), and on the claims of Armstrong and Hagel (1996) and Hagel (1999), who suggested that online brand communities can play a major role in creating loyal customers (discussed more fully below). Becker-Olsen’s results imply that consumers are more interested in purchasing conveniently and quickly than in browsing and developing a relationship. This supports Peterson’s (1997) view that marketing relationships are by nature mostly exchange-oriented rather than relational.
The need for online retailers to pay more attention than their offline counterparts to establishing trust in the minds of consumers is seen as key by many researchers. Reichheld and Schefter (2000) suggested that the outlays to acquire a customer are often considerably higher for online retailers than for traditional channels, but that encouraging repeat (i.e., loyal) customers is key to success as these customers not only increase their spend over time but contribute to further customer acquisition through positive recommendation (which is easier in an online environment, via email, etc.). In Reichheld and Schefter’s words, ‘Loyalty is not won with technology. It is won through the delivery of a consistently superior customer experience’ (2000: 113). Part of their argument is that the Internet, like all database relationship-marketing channels, enables marketers to target resources on their most profitable customers and prospects. This argument is similar to that of Peppers and Rogers (1993), who proposed the ultimate segmentation approach: reducing the market to what is sometimes called ‘segments of one’. (Peppers and Rogers themselves avoid this term, however, since they see one-to-one marketing as fundamentally different from even the most targeted ‘push’ marketing, being based on an ongoing dialogue between the supplier and customer.)
Hoffman et al. (1999) argued that part of consumers’ low trust in online vendors arises from their perceived lack of control over web businesses’ access to their personal information and the secondary use of this information. The solution they propose is a radical shift toward more cooperative interaction between a business and its customers. A reasonable hypothesis based on cooperation is that consumers are willing to disclose personal information and to have that information subsequently used to create customer profiles for business use, if they also perceive there to be fair processes in place to protect individual privacy. Support for this hypothesis was found by Culnan and Armstrong (1999). They concluded that privacy concerns need not hold back the development of consumer e-commerce, provided that firms observe procedural fairness.
Milne and Boza (1999), however, presented evidence that improving trust and reducing concerns are two distinct approaches to managing consumer information. Further, contrary to existing self-regulation efforts, they argued that, when managing consumer information, improving trust is more effective than efforts to reduce concern. Researching gender differences regarding privacy concerns, Sheehan (1999) found that women were generally more concerned about online privacy, but that those men who were concerned were more likely to adopt behaviors to protect their privacy.
The issue of when and how consumers use brands as a source of information when shopping online was addressed by Ward and Lee (2000). Applying theory from information economics, they hypothesized that recent adopters of the Internet would be less proficient at searching for product information and would rely more on brands; as users gained experience with the Internet they would become more proficient searchers, more likely to search for alternative sources of information, and so less reliant on brands. These hypotheses were tested and supported using claimed usage and opinion data collected in one of the GVU Center’s regular surveys of web users (see note 2). Ward and Lee suggest that these findings are consistent with the substitutability of brand advertising for searches, especially for consumers with high search costs. Their results support the view that branding does not merely reinforce loyalty, but conveys useful product information that tends to make markets more efficient.
Returning to the theme that the Internet is better able to support buyer-seller relationships than offline media, Dellaert (1999) explored how it can facilitate increased consumer contributions to product/service design processes, branding (e.g., by discussing consumption experiences in online groups), service (helping other consumers in product searches and product usage), and the production process (by ordering electronically). In particular he examined the difference between the drivers of these consumer contributions and the drivers of online ordering. For instance, he found that consumer experience of the web was a driver of consumer contributions but not of online ordering. Similarly, online ordering increased linearly with consumer income, whereas consumer contributions had an inverted u-shaped relationship with income.
Also in support of the relationship approach, and in contrast to Becker-Olsen (2000) (see above), are two other studies. Mathwick (2000) reported a survey of online purchasers in the GVU database which found that not all respondents were ‘exchange’ orientated. For some users, a ‘communal’ orientation towards other users was claimed to be a defining characteristic of the online experience. Responses were categorized on a two-by-two matrix based on exchange orientation and communal orientation. Although a particular characteristic of the Internet is the ability to target consumers using behavioral data, Peltier et al. (1998) explored the use of relationship-oriented attitudinal data (trust, commitment, and relationship benefits) as the basis of market segmentation. This approach is also relevant to the use of online communities (see below) as a vehicle for increasing brand loyalty.
Taking a more strategic overview, Wind and Rangaswamy (2001), in a conceptual paper, propose that the next stage in the evolution of mass customization is customerization—‘a buyer-centric company strategy that combines mass customization with customized marketing’ (Wind and Rangaswamy, 2001: 13). They state that customerization requires the effective integration of marketing, operations, R&D, finance, and information, plus a substantial change in the firm’s orientation, processes and organizational architecture.
Turning to methodologies less often seen in marketing, network methods and analysis tools have been proposed by Iacobucci (1998) as useful for examining interactive marketing systems. She describes the content properties of interactive marketing: technology, intrinsic motivation, use of interactive marketing information, and the real-time aspect of interaction, together with the structural properties of interactive marketing: customization, responsiveness, interactions amongst relevant groups, and a structure of networked networks. Iacobucci suggests that interactive marketing is network-like, and can therefore be analyzed using the tools developed to help understand the meaning of intricate network structures.
Using Community to Reinforce Loyalty
Ever since the publication of Howard Rheingold’s (1993) book The Virtual Community: Homesteading on the Electronic Frontier, commentators have been struck by the Internet’s unprecedented capacity to support global communities of interest. In a business context, e-commerce models such as Hotmail’s free email service and (to some extent) eBay’s online auction system are centered on the firm’s ability to use the web to facilitate consumer-to-consumer (C2C) communication. More controversial is the scope for building an online brand community based on a major established brand. The most prominent advocate of this strategy is John Hagel of McKinsey, who has argued that a virtual community can expand the market, increase the brand’s visibility, and improve profitability (Hagel, 1999; Hagel and Armstrong, 1997; Hagel and Singer, 1999).
Research on online brand communities, like much Internet-related research, is still somewhat atheoretical. However, Wilde and Swatman (1999) explored a number of theories to support the concept of a telecommunications-enhanced community and attempted to develop an integrative model. McWilliam (2000) discussed the advantages, disadvantages and limitations of online brand communities, and the new and unfamiliar skills brand managers will need if they are to succeed in managing such communities. The problems identified include issues of scale (e.g., the number of active participants in discussion groups is tiny compared to the potential customer base) and especially control. She concluded that it is extremely hard for the firm, accustomed to controlling communications about its brand, to strike the right balance here.
Supporting the view that community, and especially C2C interactions, can help build loyalty, Dellaert (2000) explored how such contributions can be modeled and measured, and tested a model in the context of Dutch tourists’ preferences for Internet travel websites with and without other tourists’ contributions. The main finding was that consumers’ evaluation of other consumers’ contributions to a site was high relative to other website characteristics.
As part of a study aimed at developing theory and providing evidence on brand communities, Muniz and O’Guinn (2001) used both face-to-face ethnographic research and reported/observed computer-mediated communications to gauge the use and role of brand communities. They found evidence that involvement in brand communities (e.g., Saab, Apple Macintosh, Ford Bronco) was prevalent, and that the web provided much positive reinforcement; giving consumers a ‘greater voice,’ providing an important source of information (from the brand but more particularly from other members), and enabling social benefits (e.g., sharing rituals and traditions, promoting brand legitimacy). Membership of an online brand community was shown to positively affect the components of brand equity (e.g., perceived quality, brand loyalty, brand awareness and brand associations). However, the authors highlight that a strong brand community can also be a threat to marketers. For example, the community could collectively reject marketing efforts or product change, or it could be sabotaged by brand ‘terrorists’ or competitors.
The Role of Advertising Online
For most established businesses, the web’s main role is either to reduce costs or to add value for existing customers, but it also has a potential role in customer acquisition, and in the case of a web startup, this role is crucial. Both large corporates and web startups see driving traffic to the site as one of the most important, as well as one of the most difficult, determinants of the site’s success (Rosen and Barwise, 2000). A few startups such as Amazon have been extremely successful at generating free publicity. Others have been adept at so-called viral marketing (i.e., electronic word-of-mouth), the classic case being Hotmail: anyone with a free email account has a motive to encourage their friends to set one up too. More generally, however, Langford (2000) found that, although Free Traffic Builders (FTBs: search engines, directories, news groups, listservs, bulletin boards, and chat rooms) offer free online promotion, none of these had much impact in generating traffic. Indeed, the scale of their marketing expenditure relative to their revenue is one of the main causes of failure among web startups, especially B2C dotcoms (Higson and Briginshaw, 2000).
Several studies have looked at managers’ perceptions of the Internet as an advertising medium (Ducoffe, 1996; Leong et al., 1998; Schlosser et al., 1999). Bush et al. (1999) found that, while advertisers were generally keen to use the web to communicate product information, they were concerned about security/privacy and uncertain how to measure the effectiveness of online advertising. Leong et al. (1998) reported that website managers perceived the web to be a cost effective advertising medium, well-suited for conveying information, precipitating action, creating brand or product image and awareness. However, it was seen as ineffective for stimulating emotions or getting attention.
Silk et al. (2001) explored the role of the Internet as an advertising medium in competition with other media. Their analysis built on an earlier econometric study of the substitutability and complementarity of traditional US national media, as revealed by cross-elasticities over the period 1960-94 (Silk et al., 1997). They reviewed the early evidence on four aspects of online advertising: household penetration, consumer demand for information, pricing and measurement, and fit with different product/service categories. They concluded that
[the Internet’s] longer-term impact on intermedia rivalry will be broad and substantial. [It] is emerging as an adaptive, hybrid medium with respect to audience addressability, audience control, and contractual flexibility … [and] a potential substitute or complement for all the major categories of existing media. (Silk et al., 2001: 145)
Consumer Attitudes toward Advertising on the Web
If we turn from firms’ use of the web as an advertising medium to users’ perceptions of web advertising, early researchers (Hoffman et al., 1995; Rust and Oliver, 1994) predicted that consumers might abandon their traditionally passive role and actively seek out advertisements of relevance to them. It has also been suggested that a decrease in consumers’ search costs, coupled with technology to enable them to filter and block unwanted advertisements, and the ability of advertisers to offer targeted rewards for viewing ads, may lead to an ‘unbundling’ of advertising and content (Yuan et al., 1998).
Mehta and Sivadas (1995) found that, while early Internet users had a fairly negative attitude toward online advertising, they were more likely to respond to targeted than to non-targeted ads. Ducoffe (1996) reported that business users perceived web advertising as more informative than valuable or entertaining. Survey respondents were asked to rank seven media in terms of their value as a source of advertising. The web was placed near the bottom.
As Internet use became more widespread, however, Schlosser et al. (1999) found wide variation among users’ attitudes toward Internet advertising -equal numbers of respondents liked, disliked, and felt neutral toward it. Enjoyment of looking at web adverts contributed more than the informativeness or utility of the ad toward developing positive consumer attitudes to web advertising. This finding was mirrored by the responses of a demographically matched sample who answered questions on advertising in general, showing that the reported perceptions of Internet advertising were not just a reflection of the demographics of early Internet users. Other research has looked at the relationship between the complexity of websites/ads and attitudes towards them (Bruner and Kumar, 2000). Positive attitudes were associated with greater web experience and with sites perceived as interesting but not complex.
Advertising Effectiveness Online
While it has been shown that the greater the degree of interactivity, the more popular the website (Ghose and Dou, 1998), interactivity does not always enhance advertising effectiveness as it can interrupt the process of persuasion, especially when ads are targeted (Bezjian-Avery et al., 1998). Related to this, Griffith and Krampf (2000) found that consumers viewing a retailer’s product offering through a print ad were more involved with the offering, and recalled more about the product and the brand, than did consumers viewing the same offering online.
Most Internet users use search engines to find product or brand information. The ability of popular search engines to locate specific marketing/management phrases was modeled by Bradlow and Schmittlein (2000). They concluded that, in addition to the sheer size of the search engine (i.e., total number of pages indexed), its sophistication (depth of search, ability to follow frame links and image maps, and ability to monitor the frequency with which a page’s content changes) also affected the probability that a given engine could locate a given web page.
Unsurprisingly, the nature of the ad copy also affects the clickthrough rate (Hofacker and Murphy, 1998). Building on this research the same authors modeled clickthrough probabilities and surprisingly found that the addition of an extra banner ad on a page did not reduce the clickthrough rate of the first banner ad (Hofacker and Murphy, 2000). Early work by the Millward Brown company (Briggs and Hollis, 1997) established that banner ads were not only a direct marketing vehicle but also worked much like offline advertising in that, even when no clickthrough occurred, they helped to build brand awareness and image for the advertised brand.
Flores (2000) reported results of extensive research on Internet advertising effectiveness at Ipson-ASI Interactive. He stressed that traditional banners accounted for only about half of online advertising expenditure in 1999 (versus 95% in 1997). Other forms included full-page ads (‘interstitials’), ‘rich media’ ads, pop up ads, sponsorships, email ads (e.g., in electronic newsletters), and company or product websites. He argued that evaluation should depend on whether the main aim was direct response or brand-building. His findings include:
- Larger ads (interstitials etc.) generated greater recall and clickthrough rates, but took longer to download, causing more annoyance.
- Consumers found broadband ads almost as engaging as TV ads, and much more engaging than narrowband or print.
- Copy quality was crucial (and generally low).
- Experienced online consumers were less tolerant of online ads (and recall rates were generally declining over time).
Flores (2000: 17) also argued that the future challenge would be not only to understand how each medium (and format) works, but also how they work together, including in combination with new media such as advertising on wireless devices (Barwise and Strong, 2002) and iDTV (Brodin et al., 2002).
Drèze and Zufryden (1997) developed and evaluated a web-based methodology for evaluating the effectiveness of promotional websites. Using conjoint and efficient frontier analysis, the four site attributes tested were background, image size, sound file display, and celebrity endorsement. Their model aimed to provide a means of evaluating different trade-offs to achieve website configurations that result in the greatest time spent at the site plus the highest number of pages viewed.
Exploring the measurement of Internet advertising gross rating points (GRPs), reach and frequency, Drèze and Zufryden (1998) suggested that the two main measurement problems to be addressed were identification of an individual, and counting revisits of cached (i.e. stored) content. Leckenby and Hong (1998) continued the search for appropriate audience measures by developing and testing six reach and frequency models. They concluded that models developed for magazine or television data generally performed equally well with Internet data, with the simplest model, the Beta Binomial, providing the greatest accuracy. Wood (1998) also discussed how reach and frequency measures were evolving on the web. Other authors who conducted studies into the effects of web ads include Bellizzi (2000), who studied business-to-business advertising and found that mentioning or simulating the website in print ads significantly increased site traffic, Sen et al. (1998), who discuss the different variables available and appropriate for segmenting web users, and Yang (1997), who reported on an early study comparing the effect of interactive ads on students whose country of origin was Taiwan, China, and the US.
Advertising on the web has also been evaluated in terms of its comparative effectiveness compared with offline advertising. Hoffman and Novak (2000) discussed a range of customer acquisition methods used by CDNOW: traditional media (radio, television, and print), online advertising (e.g., banner ads), a revenue-sharing affiliate program, strategic partnerships with traffic generators such as AOL, plus PR, freelinks, and word-of-mouth. They concluded that revenue-sharing, a very different model from the impression-based advertising which still dominates broadcast media, was the most cost-effective means of acquiring customers online.
The Internet is more a ‘pull’ medium like classified advertising in print media (i.e., where interactions are initiated by the customer) than a ‘push’ medium like TV, but marketers are gradually learning how to use it in both modes (Braunstein and Levine, 2000; Hofacker, 2001). Overall, Internet advertising (including email) is still a growth area within marketing communications, despite the justifiable reaction against earlier over-optimistic expectations. Marketers are still learning how to use it in terms of brand strategy, creative execution, and evaluation.
Internet Economics and Pricing
Some commentators expect the Internet to lead to ‘frictionless markets’ in which empowered customers, increasingly supported by intelligent agents, trusted intermediaries and third parties, shop around with minimal effort, playing one supplier off against another and relentlessly driving down prices. In economic terms, the Internet can reduce buyer search costs, decreasing the ability of sellers to extract monopolistic profits and increasing the ability of markets to optimally allocate resources (Bakos, 1997). In this section we explore theoretical and empirical research to date on the impacts of the Internet on market prices and price dispersion. We start with theoretical notions on buyer search costs and the likelihood of ‘frictionless’ markets.
An early analysis of web-related information economics was given by Wigand and Benjamin (1995), who argued that the Internet holds great potential for efficiency gains along the whole industry value chain, primarily because of transaction cost savings (see also Rayport and Sviokla, 1995). The potential effects Wigand and Benjamin discussed include disintermediation, reduced profit margins, and consumer access to a broad selection of lower-priced goods, but also various opportunities to restrict consumers’ access to the vast amount of available information and potential commerce opportunities. They developed an integrated model of electronic commerce and discussed implications for public policy ‘to mitigate risks associated with market access and value chain reconfiguration.’
Bakos (1997) also addressed the impact of reducing search costs. He modeled the role of buyer search costs in markets with differentiated product offerings in the context of electronic marketplaces, and explored the implications for the incentives of buyers, sellers, and independent intermediaries to invest in such marketplaces. Among other things, his analysis provides formal support for the proposition that electronic marketplaces promote price competition and reduce the market power of sellers, as argued by Wigand and Benjamin (1995), Malone et al. (1987), and others who expected online markets to have less ‘friction’ than their offline counterparts.
More recently, Sinha (2000) argued that the ease of collecting and comparing information on the web, regarding prices, features and quality, means that costs are becoming increasingly transparent. This, according to Sinha, will impair sellers’ ability to obtain high margins, turning most products and services into commodities. Sinha suggests that the Internet ‘encourages highly rational shopping,’ eroding the ‘risk premium’ that sellers have been able to extract from wary buyers. It also demands that companies with varying prices in different countries re-examine their price structure. One response for firms is ‘smart’ pricing through versioning and other mechanisms such as auctions. Sinha (2000: 50) argues that such ‘smart’ pricing may be extremely risky in the long term, as it may create perceptions of unfairness among consumers, now able to share price information easily. The solutions he recommends are a combination of product quality, innovation, and bundling.
Sinha’s succinct but wide-ranging article touches on several topics that have been explored by researchers: price levels and price dispersion, bundling and versioning, and auctions. In the following sections, we look at studies in each of these areas.
Price Levels and Price Dispersion
Early empirical evidence on the Internet’s impact on prices is reviewed by Smith et al. (2000), who found that Internet markets are more efficient than conventional markets with respect to price levels, menu costs, and price elasticity. However, despite the presence of conditions to foster efficiency, they also found substantial and persistent dispersion in prices. They suggested that this may be partly explained by heterogeneity in retailer-specific factors such as trust and awareness (i.e., brand equity). In addition, Internet markets are still at an early stage and may change dramatically in the coming years with the development of cross-channel sales strategies, intermediaries and shopbots, improved supply chain management, and new information markets.
Brynjolfsson and Smith (2000), one of the studies reviewed by Smith et al. (2000), analyzed the prices of books and CDs on 41 Internet and conventional retail outlets. They found that prices on the Internet averaged from 9% to 16% lower than in conventional outlets (depending on whether taxes, shipping and shopping costs are included in the price). They also found substantial price dispersion among Internet retailers, although weighting the prices by a proxy for market share reduced this dispersion. They concluded that, ‘while there is lower friction in many dimensions of Internet competition, branding, awareness and trust remain important sources of heterogeneity among Internet retailers’ (Brynjolfsson and Smith, 2000: 563).
Signs of less friction are present in several studies. For example, in a study on life insurance, Brown and Goolsbee (1999) found that prices for term life policies fell 8% to 15% between 1995 and 1997. Although other factors contributed to lower prices, the rising use of specialized websites for comparison shopping explained up to half of the total decline. Similarly, Scott-Morton et al. (1999) found that car buyers who used the http://Autobytel.com online referral service paid, on average, 2% less than customers who bought offline.
In another study of price and non-price competition in the online book industry, Clay et al. (1999) collected prices of 107 titles sold by 13 online and 2 physical bookstores. Controlling for book characteristics, prices in online and physical bookstores were found to be the same. In line with Brynjolfsson and Smith (2000), this study found significant price dispersion among online bookstores, providing indirect evidence of perceived product or brand differentiation, enabling Amazon in particular to charge a ‘substantial premium, … even relative to http://barnesandnoble.com andhttp://Borders.com’ (Clay et al., 1999: 1).
Empirical research suggests that the Internet’s impact on pricing has been limited. A theoretical study by Lal and Sarvary (1999) provides insight into possible explanations. Their model distinguishes between ‘digital’ product attributes (which can be communicated online at low cost), and ‘non-digital’ attributes (for which physical inspection of the product is needed). It assumes that consumers are faced with a choice of two brands but are familiar with the non-digital attributes of only the brand bought on the last purchase occasion. Based on this assumption, Lal and Sarvary showed that when (1) the proportion of Internet users is high enough, (2) non-digital attributes are relevant but not overwhelming, (3) consumers have a more favorable prior knowledge about the brand they currently own, and (4) the purchase situation can be characterized by ‘destination shopping’ (i.e., the fixed cost of a shopping trip is higher than the cost of visiting an additional store), the use of the Internet can lead not only to higher prices but also discourage consumers from engaging in search. Their explanation is that, under these conditions, an online consumer who wishes to do so can avoid visiting any stores at all and therefore also avoid comparing the non-digital attributes of competing brands. A further insight is that physical stores may have a growing role for product demonstration and online customer acquisition. The underlying theory is based on Nelson’s (1970, 1974) distinction between search and experience goods.
In addition to Nelson’s (1970) initial distinction between search goods (whose quality can be judged by inspection), and experience goods (whose quality can only be judged only through usage), a third category was added by Darby and Karni (1973), namely ‘credence’ goods, whose quality cannot be determined reliably even after usage. A classic example is wine. Online wine sales have been researched by Lynch and Ariely (2000) who found that first, lowering the cost of search for quality information reduced price sensitivity, and second, price sensitivity for goods common to two online stores increased when cross-store comparison was made easy. However, easy cross-store comparison had no effect on price sensitivity for unique goods. Third, making information environments more transparent by lowering all three search costs (for price information, for quality information within a given store, and for comparisons across the two stores) produced welfare gains for consumers. The implications are that retailers should aim to make information environments maximally transparent but try to avoid price competition by carrying more unique or differentiated merchandise.
Shankar et al. (1999), using data from the hospitality industry, also found that the Internet could dampen price sensitivity in some contexts. Specifically, the Internet increased consumers’ price search, although it had no main effect on the importance they attached to price, and reduced price sensitivity by providing in-depth (price and non-price) information. The Internet also increased the range of products and prices offered and product/price bundling by an intermediary, thereby reducing price importance. Finally, it reduced the amount of price searching, thereby increasing the effects of brand loyalty—very much as Lal and Sarvary (1999) hypothesized, and in line with the empirical results of Brynjolfsson and Smith (2000) and Clay et al. (1999) for books and CDs.
Bundling and Versioning
In relation to the Internet, it is often said that ‘information wants to be free.’ Here, ‘free’ can mean both liberated and priced at zero. On the Internet, the marginal cost of providing information to a customer is usually zero, so any pricing model for an information product based on equating marginal cost to marginal revenue would eventually lead to the information being given away. This raises the question of how a firm can make money from content creation or packaging. Although this is not a new issue—it is one which has always been faced by broadcasting, publishing, and other media industries—the power and ubiquity of digital technology are increasing the scale of the problem.
Arthur (1996) argued that new knowledge-based industries are characterized by increasing returns to scale, i.e., that if a product gains a dominant market share its advantage is magnified by increasing returns. In this world, ‘success accrues to the successful’ and ‘market share begets market share.’ In contrast, traditional resource processing industries are characterized by diminishing returns. Arthur compared and contrasted these two interrelated worlds of business and offered advice to managers in knowledge-based markets. One important source of increasing returns in information and communication industries is network externalities, whereby the value of a product to one user depends on how many other users there are. Network externalities were first defined and discussed by Rohlfs (1974); Katz and Shapiro provided a comprehensive survey (Katz and Shapiro, 1994).
Turning our attention to bundling, optimal bundling strategies for a multi-product monopolist information supplier (i.e., with zero marginal cost) were modeled by Bakos and Brynjolfsson (1999). They suggested that bundling large numbers of unrelated information goods might be surprisingly profitable because the law of large numbers makes it easier to predict consumers’ valuations for a bundle of goods than for the individual goods sold separately. They modeled the bundling of complements and substitutes, bundling in the presence of budget constraints, and the scope for offering a menu of different bundles if the market is highly segmented. They argued that the predictions from their analysis appear to be consistent with empirical observations of the markets for online content, cable TV programming, and music.
In Bakos and Brynjolfsson (2000a, 2000b) the authors extended their bundling model to a range of different settings. They argued that bundling can create ‘economies of aggregation’ for information goods, even in the absence of network externalities or economies of scale or scope. They draw four implications: (1) when competing for upstream content, larger bundlers are able to outbid smaller ones; (2) when competing for downstream consumers, bundling can discourage entry even when the prospective entrant has a superior cost structure or quality; (3) conversely, bundling by the new entrant can allow profitable entry; and (4) because a bundler can potentially capture a large share of profits in a new market, bundlers may have higher incentives to innovate than single-product firms.
Shapiro and Varian (1998a, 1998b) argued that ‘the so-called new economy is still subject to the old laws of economics.’ As they noted, the fixed costs of information products tend to be dominated by sunk costs—costs that are not recoverable if production is halted. They suggested that information providers therefore need strategies both to differentiate their products and to price them in such a way that the price varies between buyers, reflecting the sometimes markedly different value that the different segments place on the same, or almost the same, information product. The solution they proposed isversioning, i.e., offering the information in different versions targeted at different types of customer. This is similar to the series of release windows for movies and the way publishers often release a book first as a high-priced hardback and later in paperback. They described a wide range of dimensions of versioning: convenience, comprehensiveness, manipulation, community, annoyance, speed, and support. Some of these issues were also explored by Adar and Huberman (2000), who focused on the possibility of exploiting the different and regular patterns of surfing demonstrated by different Internet users, by implementing ‘temporal discrimination’ through dynamically configuring sites and versioning information services.
Online auctions, with eBay as the best-known example, were a hot topic of the late 1990s. They offer a rich area of study for marketing researchers and economists. Lucking-Reiley (1999a) presented what he called ‘an economist’s guide’ to online auctions, including a brief history, and the results of a survey of 142 auction sites that were online in the Fall of 1998. His paper summarized the various business models they used, what goods they offered for sale, and what kinds of auction mechanism they employed. Lucking-Reiley argued that established auction theory from economics could be used to improve Internet auctions. He also presented detailed data on the competition between the incumbent eBay and the two well-funded entrants into the B2C online auction arena, Yahoo! and Amazon, in 1999. Among other things, he showed that the different auctioneers’ fee structures had measurable incentive effects on sellers’ choices and transaction outcomes.
Building on this review, Lucking-Reiley (1999b) tested different auction formats using field experiments in which collectible trading cards were auctioned off. In addition, Lucking-Reiley et al. (2000) presented an exploratory analysis of the determinants of prices in online auctions for collectible coins at eBay. Three findings stand out. First, a seller’s feedback ratings, reported by other eBay users, have a measurable effect on his auction prices. This is particularly true for negative feedback ratings. Second, minimum bids and reserve prices tend to have a positive effect on the final auction price. However, this finding does not take into account that these instruments also decrease the probability of the auction resulting in an actual sale. Third, when a seller chooses to have his auction last for a week or so, this significantly increases the average auction price.
A preliminary literature review and frameworks for analyzing auctions are also given by Klein and O’Keefe (1999) and Chui and Zwick (1999). Klein and O’Keefe described an example (http://Teletrade.com) of a telephone-based auction which now also uses the web; explored possible theoretical implications; and developed seven hypotheses for future empirical research. Chui and Zwick explored the scope and scale of online auctions and the range of business models, including B2C, B2B, and C2C auctions. DeKoning et al. (1999) explored consumer motivations in using C2C online auctions, focusing especially on the behavioral differences between global and local/regional online auctions.
Conclusion: Little Impact to Date and Brands are Not Dead
Overall, the dramatic early predictions that the Internet would lead to frictionless markets characterized by commoditization and the death of brands (e.g., Gates, 1995; Negroponte, 1995; Wigand and Benjamin, 1995) have so far proved wide of the mark. There is evidence that it has increased competition and reduced average prices slightly in some markets, but the impact on both price levels and price dispersion has been small. Brands are still here, and strong enough to attract critics (Klein, 2000).
The underlying reasons why brands exist have not gone away (Barwise, 1997). In fact, trusted brands may be even more important in a world of information overload, and money-rich, time-poor consumers, where product quality still cannot usually be reliably judged online (Barwise, 1997; Brynjolfsson and Smith, 2000; Dayal et al., 2000; Lal and Sarvary, 1999; Shankar et al., 1999). Also some firms can counter price competition by providing unique products that prevent direct comparisons (Lynch and Ariely, 2000). Finally, concerns about brand reputation should limit firms’ enthusiasm for aggressive ‘smart’ pricing—a euphemism for charging more to less price-sensitive customers (Sinha, 2000).
The continuing importance of brands has been one lesson from the dotcom bubble. Another is that romanticized notions that the Internet would abolish economies of scale were largely mistaken. In addition to economies of scale in branding and fulfillment, and network externalities (e.g., in standards), even pure information businesses can gain from economies of bundling (Bakos and Brynjolfsson, 1999, 2000a, 2000b) and will likely also have more opportunities for versioning (Shapiro and Varian, 1998a, 1998b).
One area where the Internet is having more impact is auctions. In the B2C and C2C markets, eBay is a new type of business, but the number of viable firms seems likely to be small. In contrast, online markets and auctions are having a significant impact on B2B markets (Bloch and Catfolis, 2001).
It may be that the muted impact of the Internet on prices is still the quiet before the storm. As intelligent agent software becomes more powerful and more widely used, we may see more pressure on prices and on brand loyalty. Even this, however, may have more impact on channel choice (e.g., finding the cheapest place to buy the brand) than on the choice of brand itself.
We now turn from ‘horizontal’ competition within a market to ‘vertical’ competition along a value chain, i.e., channels and intermediaries. This is another area for which early predictions of dramatic change have proved overstated.
The Impact on Channels and Intermediaries
The ability of easily accessible electronic information to increase the efficiency of markets was an early topic addressed by marketing academics. Bakos (1991) used economic theory to develop models which showed that, where product quality and price information are easily available (as in electronic markets), search costs are reduced and benefits for buyers increased, which, in turn, can reduce sellers’ profits. Following on from Bakos’s work, Benjamin and Wigand (1995) suggested that the so-called national information infrastructure (or NII, of which they believed the Internet was only a part) would cause a restructuring and redistribution of profits among stakeholders along the value chain, threatening all intermediaries between the manufacturer and consumer.
This issue of the role of intermediaries in buyer-seller relationships has been a recurring theme, with much of the earlier work suggesting disintermediation and later papers generally arguing for re- or cyber-intermediation. Rayport and Sviokla (1994) described how physical interactions in the marketplace were being replaced by virtual ‘marketspace’ transactions. They argued that the conventional value proposition was being disaggregated and that its three basic elements—content (the firm’s offering), context (how the content is offered), and infrastructure—could now be managed in new and different ways. Building on these ideas, Rayport and Sviokla (1995) suggested ways of managing and exploiting this new virtual value chain. Weiber and Kollman (1998) also evaluated the significance of virtual value chains and concluded that information, in its own right, will become a factor of competition in future markets.
Shaffer and Zettelmeyer (1999) and Wigand and Benjamin (1995) both sketched a pessimistic scenario for intermediaries. According to Shaffer and Zettelmeyer (1999), manufacturers traditionally have had to rely on retailers to provide product and category information that is either too technical or too idiosyncratic to be communicated effectively via mass media. The emergence of the Internet as a medium for marketing communications now makes it possible for manufacturers (and third parties) to also provide such information. Shaffer and Zettelmeyer show that this may lead to channel conflict. Specifically, manufacturers gain and retailers lose from information that makes a retailer’s product offerings less substitutable. In an earlier paper, Wigand and Benjamin (1995), using a transaction cost perspective, suggested that intermediaries between the manufacturer and the consumer may be threatened as electronic commerce becomes ubiquitous and as information infrastructures reach out to the consumer. Profit margins, they posit, may be substantially reduced. The consumer is likely to gain access to a broad selection of lower-priced goods, but there will be many opportunities to restrict consumers’ access to the potentially vast amount of commerce. An essential component of the evolution of the future world of electronic commerce, the authors suggest, is the ‘market choice box’—the consumer’s interface between the many electronic devices in the home and the information superhighway (Barwise, 2001; Wigand & Benjamin, 1995).
Reintermediation and ‘Cybermediaries’
In contrast to these views, Sarkar et al. (1998) argued against the idea that intermediaries are likely to disappear. Drawing on channel evolution literature and transaction cost economics, they proposed instead that virtual channel systems and new ‘cyber-mediaries’ would emerge. In a short Harvard Business Review perspectives article, Carr (2000) took this argument a step further, arguing that, far from the widely predicted disintermediation, the Internet is in fact leading to ‘hypermediation,’ in which transactions over the web, even very small ones, routinely involve many intermediaries—not only wholesalers and retailers, but also content providers, affiliate sites, search engines, portals, internet service providers, software makers, and many others. He suggested that it is these largely unnoticed intermediaries who stand to gain most of the profits from electronic commerce.
Jin and Robey (1999) focused on B2C cybermediaries such as Amazon, Virtual Vineyards, and 1-800-FLOWERS. They proposed six theoretical perspectives on cybermediation: transaction cost economics, consumer-choice theory, retailing as an institution, retailing as social exchange, retailers as bridges in social networks, and retailers as creators of knowledge. They conclude that a multitheoretical approach (in contrast to transaction cost theory alone) shows both that the disintermediation hypothesis was overstated and that cybermediaries can exist for many reasons. Bhargava et al. (2000) explored the aggregation benefits that consumers derive from having access to multiple providers through an intermediary. Their analysis is theoretical and economics-based. They concluded that when consumers are heterogeneous and differentiated in their willingness to pay for intermediation, the intermediary can offer two or more service levels at different price levels.
These theoretical propositions by Sarkar et al., Jin and Robey, and Bhargava et al. are consistent with empirical findings. Bailey and Bakos (1997) suggested that markets do not necessarily become disintermediated as they become facilitated by information technology. Thirteen case studies of firms participating in electronic commerce were explored, and evidence was found of new emerging roles for online intermediaries, including aggregating, matching sellers and buyers, providing trust, and supplying interorganizational market information. The authors discuss two specific examples to illustrate an unsuccessful strategy for electronic intermediation (BargainFinder) as well as a more successful one (Firefly).
Christensen and Tedlow (2000) categorized the Internet as a ‘disruptive’ technology, which enables innovative retailers to create new business models that significantly change the economics of the industry. They put this in an historical context by relating the Internet to three previous disruptive technologies in retailing: the department store, the mail order catalog, and the discount department store. They proposed that
the essential mission of retailing has always had four elements: getting the right product in the right place at the right price at the right time and the Internet has great potential for improving performance on various combinations of the first three of these. For information products and services, the Internet can also perform outstandingly on the fourth, time, dimension, but for physical products it does not. When shoppers need products immediately, they will head for their cars, not their computers. (Christensen and Tedlow, 2000: 42)
They further argued that the Internet is unsuited for products which require ‘touch and feel,’ not to mention ‘taste and smell.’
Based on their analysis of the three earlier disruptive technologies, Christensen and Tedlow noted that one pattern has been that generalist stores and catalogs dominate at the outset of the disruption but are then supplanted by specialists. A second pattern has been that the disruptive retailers initially sold easy-to-sell branded mass-market products and then moved up-scale with higher-margin, more complex products. They suggested that it is too soon to say whether the first of these two patterns will recur on the Internet—more likely, the pattern will vary between categories—but that there is some evidence that the second pattern is recurring, and probably much faster than with the previous disruptive technologies, since the Internet enables firms to swiftly achieve high market reach combined with high richness of content and range (also see Evans and Wurster, 1999). If this analysis is correct, we will see consumer e-commerce growth for complex, high-ticket items such as durables (but excluding those needing ‘touch and feel’) as well as for simple branded products such as books and music.
The corporate intranet represents another largely unresearched, new B2B channel (Barwise, 2001: 37). Increasingly, high-value corporate employees are using intranets for ordering office supplies, booking travel, etc. Some firms have extended this system to enable staff to make personal purchases (e.g., booking a leisure break at an attractive price that includes a corporate discount) through the intranet. This is seen as a tool for attracting and retaining key staff, as well as increasing their use of, and familiarity with, the intranet. For the supplier, it represents a low-cost, low-risk way of reaching money-rich, time-poor consumers.
Business-to-Business (B2B) Markets
Much of this previous research focuses on consumer services and it is these that have attracted the most media coverage. More important in business terms, at least initially, is the setting up of new B2B markets and exchanges or ‘e-hubs.’ Chircu and Kauffman (2000) describe a framework whereby a traditional intermediary is able to continue to compete by combining web technology with its existing specialist assets. The framework is based on literature from several disciplines and evidence from a study in the corporate travel industry. The results show that traditional travel firms have been able to avoid disintermediation and retain a highly profitable central role in this market.
Potentially more dramatic (in terms of changes in business processes) has been the emergence of entirely new B2B e-hubs. E-hubs can be defined as websites where industrial products and services can be bought from a wide range of suppliers. Ramsdell (2000) categorized B2B online markets into three kinds: ‘vertical’ markets, such as those for auto manufacturing or petroleum products; ‘horizontal’ markets, typically focusing on the supply of maintenance, repair and operations (MRO) products, such as safety supplies, hand tools, and janitorial services; and finally markets focusing on specific functions such as human resources. He concluded that these B2B markets would fundamentally change how firms and their suppliers interact, especially for MRO products and services. An alternative classification has been given by Kaplan and Sawhney (2000) based on what businesses buy (operating inputs versus manufacturing inputs) and how they buy (systematic sourcing versus spot sourcing). They give examples of each of the four types and also describe both forward and reverse aggregation models.
Since the dotcom meltdown, the number of pure-play B2B e-hubs has declined and many of those that survive are struggling. Currently, the growth is in less ambitious B2B online auctions, where individual firms use the web to support contract bids by their suppliers using technology provided by firms such as http://Freemarkets.com. There are, however, arguments that in the long term, the dominant e-market will be industry consortia delivering full supply chain integration (Bloch and Catfolis, 2001).
Online Marketing Strategy
Turning to implications for marketing strategy, below we first describe studies focused on general strategic implications of the Internet. We then turn to research on how firms should prepare themselves for the ‘new economy,’ and which business models they employ (or should employ) in online environments. We then briefly discuss research that has adopted an international perspective. Finally, we give some tentative conclusions. Even more than for the other sections, research described here is clearly ‘marked’ by the time at which it was conducted. Now that the Internet bubble has burst, we expect to see more research addressing issues related to the ‘new economy slowdown’ but, to date, such studies have not appeared in academic journals.
Strategy and the Internet
Many books and articles, and some research studies, have investigated the opportunities and challenges created by the Internet and the implications for marketing strategies. Baer (1998) is one of the few authors to take a long-term perspective. He described a century of failed visions and applications, drew some general lessons from past experience, documented why interactive services might now at last take off, and indicated some likely areas for growth. Also in 1998, the Journal of Business Research devoted a special issue to business in the new electronic environment (see Dholakia, 1998 for the introduction to this issue).
Evans and Wurster (1999) argued that electronic commerce was no longer about ‘grabbing land.’ Instead, they suggested that the battle for competitive advantage in this arena will be waged along three dimensions: reach, affiliation, and richness. Reach is about access and connection—how many customers a business can connect with and how many products it can offer to those customers. Richness is the depth and detail of information that the business can give the customer, as well as the depth and detail of information it collects about the customer. Affiliation reflects whose interests the business represents. This logic poses a challenge for incumbent product suppliers and retailers: they have to recognize that their value chain is being deconstructed.
Zettelmeyer (2000) offered another perspective on the implications of Internet growth for firms. He showed how pricing and communication strategies may be affected by the size of the Internet. Firms have incentives to facilitate consumer search on the Internet, but only as long as its reach is limited. As the Internet is used by more consumers, firms’ online pricing and communication strategies will mirror their offline strategies. According to Zettelmeyer, firms can increase their market power by strategically using information on multiple channels to achieve finer consumer segmentation.
Coltman et al. (2000) focused on the forces that determine the appropriateness of e-business to a firm. They sketched out the characteristics of organizations likely to survive in the new network economy. Three related questions guided their analysis: (1) Where is the revolution (or evolution) concentrated? (2) Why is the revolution (or evolution) occurring as it is? (3) Is it a revolution or natural evolution? They conclude that claims that e-business is driving revolutionary change are misleading and only partly correct. In contrast, Anderson (2000) proposed that e-business enables companies to transform not only their marketing operation, but also the entire way they do business, from procurement to communications to supply chain, massively improving their speed, global reach, efficiency, and cost structure. Cross (2000) also took into account the downside of these developments, which are forcing managers to rethink and reshape their business strategies, their use of technology, and their relations with suppliers and customers. In Cross’s view, the convergence of new technologies, hyper-competitive markets, and ‘heat-seeking’ financial and human capital that quickly flow to new and untested business models threatens a number of traditional business models and processes.
In a recent article entitled ‘Strategy and the Internet,’ Porter (2001) argued that, contrary to what many observers think, the Internet is not disruptive to most existing industries and companies, because it does not nullify important sources of competitive advantage—in fact, it often makes them even more valuable. Rather than rendering strategy obsolete, because the Internet tends to weaken industry profitability without providing proprietary operational advantages, it is more important than ever for companies to distinguish themselves through strategy. For traditional companies, he suggests strategy should be based on the view that the Internet complements rather than cannibalizes existing ways of doing business.
How Firms Should Prepare for the New Economy
Not surprisingly, much research in the strategy area has centered on the question of how firms should adapt to their rapidly changing environment and ‘get in shape’ for the new economy. For example, Voss (2000) described how firms can develop a systematic strategy for delivering service on the web, while Weiber and Kollmann (1998) evaluated the significance of virtual value chains in opening up possibilities in the so-called ‘marketplace’ and ‘marketspace’ (Rayport and Sviokla, 1994).
Most authors see becoming an e-business as an evolutionary journey for firms. For example, Earl (2000) identified six stages, each determining the course of the next: external communications, internal communications, e-commerce, e-business, e-enterprise, and transformation (drop the “e”). These correspond to six lessons representing an agenda for evolving e-business: (1) perpetual content management, (2) architectural integrity, (3) electronic channel strategy, (4) high-performance processes, (5) information literacy, and (6) continuous learning and change. Similarly, Albrinck et al. (2000) argued that virtually all companies pursue e-business opportunities in a consistent way, passing through four stages: grassroots, focal point, structure and deployment, and endgame. Venkatraman (2000) described five steps to a ‘dot-com strategy.’ In his view, vision, governance, resources, infrastructure, and alignment are the stepping stones to a successful web strategy.
Chavez et al. (2000) outlined a multidimensional framework to help managers decide how to structure their Internet businesses: whether to keep them integrated into the parent company, establish them as wholly-owned subsidiaries, or spin them off (wholly or partially). They argued that firms must weigh the tradeoffs between what they called the ‘three Cs’: control, currency and culture. Above all, the decision must be made in the context of a company’s total ‘digital agenda’: that is, as part of its overall strategy for creating and sustaining value in the new economy.
Dayal et al. (2000) specifically considered how firms can build digital brands. In their view, the ‘3 Ps’ of a physical brand in the consumer’s mind -its personality, presence, and performance—are also essential on the web. In addition, digital brand builders must manage the consumer’s online experience of the product, from first encounter through purchase to delivery and beyond (see pages 532-533, Section 2). As discussed below, their paper also analyzes the range of business models underlying digital brands.
More recently, Moss Kanter (2001) described ‘ten deadly mistakes of wanna-dots,’ i.e., established firms seeking to incorporate the Internet into their businesses. Based on empirical research in North America, Europe, and Asia, she found evidence for several common barriers to change in emerging e-businesses, including lack of commitment and lack of knowledge about how to change. She argued that ‘wanna-dots’ should go beyond ‘cosmetic changes’ and be prepared to undergo a ‘serious makeover.’
How can firms use the web to achieve strategic and marketing benefits? This is clearly a key strategic marketing question but, to date, research on online business models is limited. Perhaps this is because a proper examination of the question necessitates a thorough understanding of the behavior of consumers, firms and other players in the Internet arena, while the first empirical generalizations in these areas are only just starting to emerge.
Nevertheless, some researchers have started to work in this area, mostly focusing on describing and classifying the various business models that are being used. Ward Hanson’s Principles of Internet Marketing (2000) was the first advanced textbook on the topic. Hanson introduced a useful distinction between business models based on improvements in the product or service and those based directly on revenues. The first includes models focused on enhancement (e.g., brand building), efficiency (e.g., cost reduction), and/or effectiveness (e.g., information collection). The second includes models in which the provider pays (e.g., sponsorship or alliances) and those in which the user pays (e.g., product sales or subscriptions).
Ethiraj et al. (2000) examined the changes in opportunity space within firms’ value chains arising from online technologies, and the implications for competitive advantage. They identified four key components of the business model—scalability, complementary resources and capabilities, relation-specific assets, and knowledge-sharing routines -and discussed how and why these may be important drivers of competitive advantage in Internet-based business models. Also looking to identify what makes Internet businesses successful, Dayal et al. (2000) argued that there are six basic business models: retail, media, advisory, made-to-order manufacturing, do-it-yourself, and information services. They posited that the success of an Internet brand rests on the skill with which its business model combines two or more of these.
Building on the entrepreneurship and strategic management literatures, Amit and Zott (2000) examined the value creation potential of a sample of American and European e-commerce companies. They developed a model that enables an evaluation of the value-creation potential of e-commerce business models along four dimensions: novelty, lock-in, complementarities, and efficiency. Dutta et al. (1998) sought to understand how the Internet has transformed business models across different types of business. Specifically, they investigated how the marketing mix and customer relationships are being transformed across different sectors and regions. They concluded that few firms are rethinking how their business models are being transformed and, consequently, that most firms’ Internet presence is rather ineffective.
Focusing on business models specifically for online content services, Picard (2000) explored how such business models emerged, how new developments are affecting those models, and the implications for content producers. He divides the history of online content service providers into periods coinciding with four ‘abandoned’ business models (videotext, paid Internet, free web, and advertising push), one model in current use (portals and personal portals), and an emerging model (digital portals). The latter allows the combination of aspects of current content portals plus digitization of video and audio.
Werbach (2000) investigated another potential source of revenue for Internet firms that has received little attention: syndication. Syndication involves the sale of the same information good to many customers, who then integrate it with other offerings and redistribute it. It has its origins in the news and entertainment worlds but, Werbach argues, syndication is expanding to define the structure of e-business. As companies enter syndication networks, they will need to rethink their products, relationships, and core capabilities.
Most empirical research on revenue models has concerned one specific revenue source—Internet advertising (discussed mainly from an advertiser perspective, in Section 4).
For established businesses, Geyskens et al. (2001) studied the net impact of setting up an additional Internet channel on a firm’s stockmarket return, reflecting the change in expected future cashflows. They found that, on average, Internet-channel investments had been profitable by this measure. They also identified firm strategies and market characteristics that influence the direction and size of the stockmarket reaction. For example, they found that early followers had a competitive advantage against both innovators and later followers (as also argued by Porter, 2001).
Geyskens et al.’s empirical analysis revolves around the newspaper industry—a widely-cited example of an industry struggling to respond to the digital revolution. According to Baer (1998), when videotext was first invented, newspapers such as the LA Times saw it as simply a new publishing medium. They were reluctant to accept that the ‘content’ of most interest to consumers was largely other consumers they knew well (colleagues, friends, family), i.e., via email or chat. Today, virtually all newspaper groups have implemented online versions, partly to protect their classified advertising revenue, partly to protect subscription/cover price revenue, and potentially to open up new revenue sources. In recent years, several other researchers have studied the (online) news industry, including Dans (2000), Dans and Pauwels (2001), Elberse (1998), and Pauwels (2001).
To date, despite the fact that the Internet is likely to have far-reaching implications for international marketing, only a handful of studies have explored marketing and the Internet in an international context.
The question of how the Internet might revolutionize global marketing was raised by Quelch and Klein (1996). They discussed the opportunities and challenges that it offers to large and small companies worldwide. They examined the impact on global markets and new product development, the advantages of an intranet for large corporations, and the need for foreign government support and cooperation. More recently, Samiee (1998) studied two types of impediment to the Internet’s adoption and growth in international marketing: structural (nations’ differing infrastructures, languages, cultures, and legal frameworks) and functional (marketing program and process issues, including data management and customer discontent). His analysis suggested that the Internet will play a much greater role in B2B marketing across national boundaries than in international consumer marketing. For an exploratory study that examines how the Internet can help small and medium-sized firms (SMEs) reach a global market, see Lituchy and Rail (2000).
Developments in Europe versus those in the US have been studied by Cornet et al. (2000) and Hammond (2001). Cornet et al. (2000) discussed how Europe is ‘playing catch-up’ with the US in e-business. The European game, they argue, may well have a different outcome, as conditions specific to Europe give incumbents a better chance to win. Hammond (2001) surveyed a global panel of media experts after the bursting of the Internet bubble in 2000. She found that consumer usage of the web was expected to continue its growth much as previously predicted by the same panel. However, there were some regional differences. Experts from Europe, despite starting from a lower base in 2000, were more optimistic than those from North America for the five-year outlook for online purchasing (in their countries) of gambling, newspapers, and groceries. Eastern Europeans expected their countries to catch up with the West by 2005.
Within Europe, Rosen and Barwise (2000) compared business use of the web by corporates, and web startups in different countries. They confirmed the advanced development in the Nordic countries, but also found that French companies were in many respects as advanced as those in the UK, Germany, and Holland, despite the relatively low penetration of the web among consumers in France. Rosen and Barwise attribute this pattern to the earlier development of France Telecom’s proprietary Minitel system, which retarded consumer adoption of the web but accelerated French businesses’ development of online systems and data.
Using a series of exemplary cases, Berthon et al. (1999) illustrated how international firms can use the web to enhance the marketing of services across national boundaries. They argued that cyberspace ‘gives the marketer undreamed control over the previously capricious characteristics of services,’ and that innovative use of the web can address problems traditionally related to the marketing of international services (intangibility, simultaneity, heterogeneity, and perishability). In addition, they offered a ‘diagnostic checklist’ to evaluate the effectiveness of a firm’s website in providing such services.
Online Marketing Strategy: Tentative Conclusions
As yet, there has been little systematic research on the implications of the Internet for marketing strategy. What has been done is still fragmented. Opinions vary between those, such as Evans and Wurster (1999) and Anderson (2000), who believe that the Internet will transform strategy, and those, such as Coltman et al. (2000), Porter (2001), and ourselves, who argue that the fundamental tenets of strategy and marketing still apply.
There is some consensus on how existing firms should prepare for the new economy, with most writers agreeing that the change process involves a series of recognizable stages, although different writers use somewhat different labels.
Business models still represent a big gap. Hanson (2000) provides a useful and balanced classification, but it remains the case that both dotcoms and established firms are generally still struggling to find business models which generate significant direct revenues online (with some notable, widely cited exceptions such as AOL, Yahoo!, and Dell). The emphasis today is mainly on what Hanson calls product/service improvement models rather than direct revenue models. For instance, there is evidence that among the main B2C beneficiaries of the web have been established retailers using a ‘bricks-and-clicks’ strategy, whereby consumers browse online but purchase at the physical store (Vishwanath and Mulvin, 2001).
Finally, although a striking feature of the Internet is its ability to transcend geography, there has been little research on its impact on international marketing. The most cited paper to-date is Quelch and Klein’s (1996) early discussion of the opportunities and threats it raises for firms operating internationally.
Conclusions and Research Opportunities
Overall Evaluation: The Impact of the Internet on Marketing
Until the bubble burst in Spring 2000, expectations of the Internet had been so overhyped that the reality was bound to disappoint. The danger now, especially in a much less confident financial and geopolitical climate, is that sentiment swings too far the other way and the Internet’s significance is underestimated. We still believe that the Internet, and the digital revolution more generally, will have wide-ranging impacts on marketing as well as on many other aspects of society and business (Barwise and Hammond, 1998).
Contrary to some of the earlier hype, the Internet does not change the fundamental principles of marketing. Nor has its impact to-date (e.g., on consumer behavior, advertising, pricing, channels/intermediaries, strategy and globalization) been anything like as dramatic as predicted. However, we expect its impact to increase greatly over the next ten years for three reasons:
- Even in the US, Northern Europe and Australasia, the Internet is only a recent part of most consumers’ lives—and its penetration is still much lower than for traditional mass technologies (TV, radio, mail, telephone, automobile). In other countries, its penetration is even lower. Internet adoption and usage are still increasing among consumers and businesses, especially in countries with lower levels of penetration today. As it approaches universal adoption and usage, its role within marketing, especially international marketing, will continue to grow.
- Marketers are still learning how to use the Internet as a medium in its own right and especially in combination with other media (Flores, 2000; Silk et al., 2001). Despite the rhetoric about integrated marketing communications, ‘bricks-and-clicks’ strategies, ‘one-to-one marketing’ (Peppers and Rogers, 1993, 1997), and ‘permission marketing’ (Godin, 1999), it will take years for most firms to turn these words into really cost-effective reality. Many of the issues relate to technology, database management, organization, and policy (e.g., addressing security and privacy concerns), as well as creative execution (Barwise and Strong, 2002). All of this is happening—for instance, the Internet is steadily increasing its share of advertising in most countries—but it will take time.
- The technology is still developing fast. Current developments include broadband, mobile Internet, iDTV and virtual reality, wireless LANs (e.g., in the home and office), better intelligent agents, and more intuitive interfaces (Barwise, 2001; Barwise and Hammond, 1998; Maes, 1999). None of these technologies has significant penetration; none works very well yet; none has an agreed standard like the Internet Protocol; none is cheap enough yet to achieve mass market adoption. All of these developments seem likely to happen, but they will take several years. Over time, however, their combined effect on marketing practice will be great.
Online technology has the potential, depending on the context, to:
- Reduce customer search costs
- Allow low-cost customization of the marketing mix
- Support some market-related activities like auctions and brand communities in areas where they were not previously viable
- Give customers access to firms (and perhaps vice versa) any time, anywhere
- Abolish some types of intermediary and create other, new types
- Strongly reinforce globalization.
Which of these actually happens, and how much, will emerge over the next few years, but meanwhile, many specific insights have already emerged from the research to date.
Specific Insights from the Research to Date
- The Internet is already valued by a large minority of the population as an information-source, but it competes with other channels and has to offer clear benefits (Bellman et al., 1999; Morrison and Roberts, 1998). Convenience is a key benefit for such adopters; the development of recommendation systems and other agents will increase the Internet’s perceived utility and use for such people (Ariely, 2000; Häubl and Trifts, 2000).
- As the Internet matures we can expect it to become somewhat more entertainment-orientated in order to attract a broader user base. This will increase its revenue potential (Hammond et al., 2000; Novak et al., 2000).
- Brand equity can, at least to some extent, be reinforced through online communities (Muniz and O’Guinn, 2001). Although brand loyalty may not be higher in online markets (Fader and Hardie, 2001), where brand loyalty is increased, high market share brands benefit most (Danaher et al., 2000). Brands are especially important for new adopters (Ward and Lee, 2000). This partially supports the view that online retailer choice is mostly a low involvement process (Ilfield and Winer, 2001).
- The Internet is emerging as a significant advertising medium, although more slowly than the enthusiasts predicted. Most of the advertising revenue is captured by the biggest sites (AOL, MSN, Yahoo!, etc.), leaving the business plans of most ‘niche’ advertising-funded dotcoms in tatters.
- As an advertising medium, the Internet can be used in many ways—e.g., mass/targeted, push/pull, local/national/global (Silk et al., 2001). It now includes a wide range of formats, with many alternatives to traditional banner ads (Bellman and Rossiter, 2001; Flores, 2000) and email, probably the fastest-growing part of Internet advertising. Clickthrough rates are low and falling, but underestimate the total consumer response (Flores, 2000). Relevance and permission are becoming increasingly important (Barwise and Strong, 2002).
- Early predictions that the Internet would lead to ‘frictionless’ markets (Gates, 1995; Wigand and Benjamin, 1995) have so far proved wide of the mark. There is evidence that it has somewhat increased competition and reduced prices in some markets, but the impact on both price levels and price dispersion has been small (Smith et al., 2000).
- By the same token, brands continue to be just as important in the post-Internet world (Barwise, 1997; Dayal et al., 2000; Lal and Sarvary, 1999; Shankar et al., 1999), as does product differentiation (Lynch and Ariely, 2000).
- Economies of scale will continue to be extremely important. In addition to physical (e.g., fulfillment) and network externality reasons, even for pure information products the larger players will have more scope for bundling (Bakos and Brynjolfsson, 2000a, 2000b) and versioning (Shapiro and Varian, 1998a, 1998b).
- Online markets and auctions have had limited impact on consumer (B2C, C2C) markets despite the success of a few businesses, notably eBay. But they are developing fast in B2B markets (Bloch and Catfolis, 2001; Kaplan and Sawhney, 2000).
- The expected ‘disintermediation,’ as existing intermediaries are replaced by direct communications between primary suppliers and consumers (Benjamin and Wigand, 1995; Rayport and Sviokla, 1995; Shaffer and Zettelmeyer, 1999), has not happened. Instead, the initial evidence is that, as predicted by Sarkar et al. (1995), when intermediaries disappear they are replaced by new web-based ‘cybermediaries’ (Bailey & Bakos, 1997; Sarkar et al., 1998).
- There is also evidence that bricks-and-mortar intermediaries who have successfully combined online technology with their existing assets (a ‘bricks-and-clicks’ strategy) have gained competitive advantage (Chircu and Kauffman, 2000; Vishwanath and Mulvin, 2001).
- At this stage, there is little empirical research on the impact of the Internet on marketing strategy. There is wide agreement that firms should prepare for the new economy in stages (Albrinck et al., 2000; Earl, 2000; Venkatraman, 2000) and that full preparation requires a transformation going well beyond marketing (Anderson, 2000; Moss Kanter, 2001).
- There is still no consensus on the long-term impact on strategy, nor on viable business models. At this stage, most successful business use of the Internet has been as a supplement to existing activities rather than as a way of generating significant direct revenue (see the classification in Hanson, 2000).
- An early article by Quelch and Klein (1996) on how the Internet may revolutionize global marketing has not been followed up by large-scale empirical research. Given the inherently global nature of the Internet, this represents an opportunity.
Future Research Opportunities
The discussion in this chapter has covered a wide range of topics, but several key areas have emerged where further research can add to the Marketing discipline. We list these under three headings: first, the application of existing approaches to measure the impact of online media on substantive topics in marketing strategy and consumer behavior; second, the development of new theory; third, the emergence of new (or the effects on existing) market research methodologies.
The Impact of the Internet on Marketing Strategy and Consumer Behavior
Under this broad heading we may expect to see the following types of questions addressed:
- Does consumer behavior change when the Internet is used as a channel for commerce?
- Are consumers more or less brand loyal when they buy online? What factors determine customer loyalty in an online environment?
- How does the role of brands differ in online versus offline environments?
- What are the short- and long-term effects of promotions in an online shopping environment?
- Is online advertising more or less effective than offline advertising? In what respects and contexts/roles?
- How should we measure marketing performance online?
- What CRM strategies are effective online?
- How should online and offline channels be combined? Do online channels ‘cannibalize’ offline channels?
- To what extent does the Internet affect international marketing and diffusion processes?
- How does the Internet fit into everyday life? How has it changed everyday life?
- How will Internet marketing evolve into marketing using a range of converging digital media (broadband, mobile, interactive television)?
We would expect to see explicit use of existing theory to address these questions—theory not only from marketing but also from economics, psychology, and anthropology.
The Development of New Theory
In addition to the use of existing theory, the rise of the Internet creates the need for new theory in a number of radically new areas, particularly:
- Consumer-to-consumer interaction
- Agent-consumer interaction
- Agent-to-agent (or machine-to-machine) interaction.
Emerging Marketing Research Methodologies
As Rangaswamy and Gupta (1999) indicate, the Internet will influence not only which research issues we pursue, but also how we will explore those issues, and how we disseminate research results, insights, and techniques to a broad audience. This review is a case in point. We used the web extensively to find relevant papers from a wide range of sources. We are also using it to disseminate the results (athttp://www.marketingandtheinternet.com). As far as emerging market research methodologies are concerned, online real-time experiments promise to be an exciting area. The Internet generates huge quantities of unobtrusive data, which can be used to set up consumer behavior experiments that are more realistic and more complex than experiments in the offline world. Englis and Solomon (2000) review consumer research applications online, and describe one particular methodology for ‘visually-oriented consumer research.’ In addition, research into controversial, sensitive or ethical topics is challenging in a face-to-face (i.e., focus group) setting. Montoya-Weiss et al. (1998) suggest the advantages (and drawbacks) that electronic communication environments have for helping with communication apprehension (CA) problems in such research. Possible other applications include:
- Assessment of customer acquisition costs
- Measurement of responsiveness to advertising and promotions
- Investigation of price sensitivity
- Exploration of consumer-agent interaction.
We would like to thank everyone who contributed to this chapter, especially Tim Ambler, Kent Grayson, Bruce Hardie, Arvind Sahay, Craig Smith, and Naufel Vilcassim (all at London Business School), who provided detailed comments on an earlier draft, and many colleagues across the world who gave comments and suggestions about relevant literature.