Reviews the functions and nature of shopping bots, the intelligent agents designed to assist consumers in product identification and comparison over the WWW. Product search and identification is an important stage in e-shopping. At this early stage of development, the analysis of the different characteristics of shopping bots such as is offered by this article, offers a platform from which it is possible to develop a more sophisticated understanding of the nature and potential role of shopping bots in e-retailing. Depending upon the features and information offered by the shopping bot, the bot may be viewed as an intelligent shopper acting on behalf of the customer, or as a department store, in which producers contract with shopping bot providers to ensure that their products are drawn to consumers attention.
Search engines are useful in e-shopping, but shopping bots are specifically designed to support comparison shopping across a number of sites. The stages in the use of a shopping bot are outlined. Search facilities available in shopping bots can be categorised into those based on classified categories, simple keyword searching, and parameter searching. The range of help offered in different bots is also reviewed. The following criteria for the evaluation of shopping bots are proposed: coverage of e-commerce sites; product range covered; evaluative information offered; geographical coverage; search facilities, and technical features. A key research agenda relates to the way in which consumers use shopping bots in e-shopping, and specifically whether they feel that they can rely on the information proffered.
Keywords: Shopping bots; e-shopping; intelligent agents.
Product searching during e-shopping
There is increasing interest in e-shopping and e-retailing, and a general recognition that e-retailing will establish itself as an alternative channel alongside traditional shopping channels. (e.g. De Kare Silver (1998). There has been extensive discussion of the challenges associated with e-retailing, in areas such as security of payments (e.g., Mitchell 1995, Booker 1995, Lynch and Lundquist 1996). Some have sought to explore the nature of the shopping experience, with a view to seeking to understand those products and audiences most suited to this new channel. (e.g. Rowley 1998, Resnick 1995, Grain 1996, Macmillan 1997, Chesborough and Teece 1996). Others, in recognition that shopping in a consumerist society is a major arena for social interaction, have recognised the potential for e-shopping to re-define societal relationships and communities. Virtual communities which take little cognisance or cultural of national boundaries are a real possibility. (e.g. Dertouzos 1997, Markham 1998, Tapscott 1998).
Consumer behaviour in traditional shopping contexts has received much attention, and a number of models of the buying process have been developed. The first stage in each of these models is generally identified to be that of information search. This stage is recognised to be an important phase during which promotional messages should reach the intending consumer, but little work has been conducted on the different ways in which the information search is performed by the user. This is perhaps because the process is somewhat ad-hoc. Like many other stages of the buying process information seeking becomes more structured and constrained in the e-shopping environment. In particular, the ability to collect product information and make comparisons between the different product offerings from different providers, possibly across national and currency boundaries is often viewed as one of the main competitive challenges of e-shopping. The Internet lowers the cost of searches for alternative and substitute products in a commodity like bargaining atmosphere, thereby encouraging greater price competition for relatively generic products. Thus consumers approaches to product searching are a key factor in successful business, and any approaches or tools which can enhance product visibility and ease of location in this market are key in defining and maintaining competitive advantage.
This article focuses on a significant innovation in the tools available for searching e-commerce offerings, the shopping bot. Shopping bots are tools that help e-shoppers to identify, locate and compare products available from e-retailers. At this early stage of development, the analysis of the different characteristics of shopping bots such as is offered by this article, offers a platform from which it is possible to develop a more sophisticated understanding of the nature and potential role of shopping bots in e-retailing.
Shopping bots are ostensibly a tool for the consumer. As an intelligent agent the consumer would expect such a tool to offer unbiased recommendations to a range of quality products, or, to act as an expert shopper. The term intelligent agent is also often taken to imply that the software tool, in this case often loaded on the consumer’s machine client will ‘learn about the consumer’s preferences. So, for example, after the shopper has instructed the agent to perform a number of searches, the bot will be able to recognise the shoppers brand preferences, or inclinations in relation to price. The extent to which the bot acts as an expert shopper depends upon the characteristics of shopping bots as discussed in this article. An alternative perspective is that of the shopping bot as a virtual department store, or shopping mall. Here the shopping bot can be viewed as an alternative means of virtual retailers designing their product offering, or influencing the product portfolio to which customer’s are directed. In this model, the e-retailers to which a shopping bot provides access have commercial relationships with the provider of the shopping bot. These relationships can take a number of alternative forms. They may be financial, such as commission agreements, or they may be strategic, where the two companies are part of the same group, or they may be marketing based, with mutual promotion agreements in the e-commerce arena.
Shopping bots are an addition to the existing range of tools available for searching the Internet. All such tools are used in a public access environment and therefore pose challenges in respect of both the user profile and the task (Rowley and Slack 1998) The user profile has two aspects, which make system design more demanding:
· Users have a wide range of different educational backgrounds and levels of experience with the system. Users range from being subject domain novices and computer novices all the way to subject experts and computer experts. The degree of knowledgeability of the computer user and the domain experience should be reflected in the design of the user interface prompts alerts and help facilities.
· A large proportion of the population are naive and new users who need to be able to adapt quickly to different systems. Many users are also subject novices and their system use is constrained by their inability to appreciate what the system can be expected to contain.
The task is ill defined and there is an element of uncertainty in both:
· What the user is likely to retrieve and accept as output from the process
· The search strategies which will prove the most effective.
In the Internet environment, the remoteness between information provider and information user is especially acute. Here, for example:
In addition, systems need to be able to accommodate the spectrum of different search strategies that might be adopted by a consumer. The spectrum ranges from a search strategy designed to locate a specific item of information, (sometimes called directed or purposeful searching), through to the almost aimless or general browsing, over the Web and through other sources (such as magazines) for ‘something interesting’. Many searches can be placed at some point in the middle of this spectrum; often the user is refining not only the search strategy, but also their information requirements as the search proceeds, so that, a search that may start with browsing may eventually have a very focused intended outcome. Alternatively, the Web search that starts with a very targeted objective, may open up other experiences, access to other sources, and suggest other lines of investigation or action that had not occurred to the user at the start of the search. Breitenbach and Van Doren (1998) in quoting Lewis and Lewis (1997) identify five categories of Web visitors: directed information seekers, undirected information seekers (browsers), bargain hunters (browsers of a type), entertainment seekers, and directed buyers directed searchers with a buying intent).
Most e-shopping applications operate within the context of the Web. This means that there are two approaches to searching: those offered by browsers, which exploit hyperlinks between documents or sites, and search engines, which perform searches on the basis of words or phrases, through the use of a large index of Web resources.
Browsers rely upon the network of hyperlinks that are embedded in sites. Access to web sites or documents is via the Uniform Resource Locator (URL). These addresses link the user to the host computer and their individual files; these are then displayed on the user’s personal workstation.
A search engine is a retrieval mechanism that performs the basic retrieval task, the acceptance of a query, a comparison of the query with each of the records in a database, and the production of a retrieved set as output. The primary application of such search engines is to provide access to the resources that are available on the WWW, and stored on many different servers. Most search engines are free, with their financial support coming from advertising revenue and through sales of the underlying technology. Search engines can be located on a remote server on the Web, or located on a home PC or internal network. Increasingly, search engines are becoming more than a Web index, and are adding content to their sites, in the form of additional services. Some believe that they are fast becoming information providers or ‘hosts’ in their own right. (Poulter 1997)
Each of the records contained in the database maintained by a global Web search service is created automatically by a program called a spider, robot, Web wanderer or Web crawler. Each time a spider is run, it is initially issued with the URL’s of a small seed set of target Web pages. It retrieves and downloads copies of the targets of those links and then activates every link contained in those pages, and so on, until is has downloaded copies of every single page that it can find. Retailers may adopt specific strategies in order to ensure that their pages are visited by such spiders, and that they can be accessed via a search engine (Chowdhury 1999). Hidden comments and irrelevant meta tags are a hazard; since these are at the discretion of the message sender, some provides of goods and services massage metadata to draw attention to the products (Laursen 1998).
There are a number of different types of search engines, of which shopping bots are one. The others are:
Shopping Bots
Shopping Bots, are
specialised search bots, which are designed to locate and compare products.
They take a query, visit e-shops or the sites of e-merchants, that may have
the product sought, bring back the results, and present them in a consolidated
and compact format that allows comparison shopping at a glance. Most shopping
bots claim to eliminate the searching necessary to identify the right product
at the best price. Searching is on the basis of full text and/or product categories.
Coverage varies both with respect to product range, sites and virtual retailers
or catalogues covered, and frequency of update to data accessed. Bots use a
variety of algorithms to perform searches with keywords. Many also provide access
to an order form. Examples are listed in Figure 1. Two useful sources of information
on shopping bots are BotSpot (Http://botspot.com),
and SmartBots.com (Http://www.smartbots.com).
These sites maintain an updated list of shopping bots with comments on product
range and related information.
Figure 1 : Some Shopping Bots
Name of Shopping Bot |
Product Range Covered |
BargainBot |
Books |
Bargain Finder Agent |
Music, CD’s |
Bid Smart |
Auctions |
*Bizrate |
Comprehensive |
*Bottom Dollar |
Eight product categories |
*Buyer’s Index |
Comprehensive |
*Cadabra |
Computer hardware and software and electronic goods |
*Compare-It |
Comprehensive |
*CompareNet |
Comprehensive |
Computer Shopper |
Computer and related products |
*Consumer World |
Comprehensive(consumer related resource) |
*E-Compare |
Comprehensive |
*eSmarts |
Various categories |
Gift finder |
Gifts |
Jango |
Comprehensive |
MySimon |
Comprehensive |
NetMarket |
Comprehensive |
Planet Retail |
Comprehensive |
Price Scan |
Computer hardware and software |
*PriceSearch |
Computers and associated products |
Shopping explorer |
Computer hardware and software |
*StoreSearch |
Comprehensive list of product categories, anchor stores, select stores and major brands |
Premium shopping directory |
Malls and stores. |
Product Searching using shopping bots.
This article is based on a study of a number of shopping bots. The shopping bots used are asterisked in Figure 1. Searches were performed in these shopping bots with a view to identifying:
Search topics used varied between shopping bots, since product ranges vary, and the extent of specification of the product that is possible in some bots is more extensive than in others. Typical search topics used included: flowers, running shoes, and digital cameras. Since many of the shopping bots used were based in the United States one or two other search topics which were initially investigated, were later discarded, because they could not be relied upon to produce an appropriate response. Some bots catered for language variation, either through deliberate cross-referencing or by accident of the variability of natural language, but others did not. Topics where such a problem was evident included football boots, and filing cabinets.
On the basis of
the test searches conducted this investigation it was possible to identify the
stages in product searching as shown in Figure 2.
Figure 2 contexutalises the specification of the search topic in the wider search
strategy, which also includes choices about the display of product details.
For those bots that also provide information and advice about product categories,
consumers may choose to access this early in the search process. Similarly consumers
who feel that they are unsuccessful with the search process and are unsure as
to how to proceed may access search help at any time during this process. In
addition, the consumer may return to earlier stages in the search process at
any one of a number of points in the search. Particularly if the consumer is
uncertain as to the product that is required (as perhaps with uncertainties
as to price that the consumer is prepared to pay) they may pass through several
of these steps a number of times, in a ‘browsing’ search mode. Further, should
the consumer be engaged in purchasing more than one product the search process
is likely to be repeated.
Analysis of the search facilities revealed three different categories of search facilities. Some shopping bots offered all of these, whilst others only offered the first two: classified product categories, and simple keyword search. Each of these categories is described below, and examples of each are provided:
Category Search Many bots give a number of categories, sometimes described as departments, on their home page. Storesearch, for instance, invites shoppers to select a department; the next screen displays a series of product categories in that department; choosing one of these categories will display a number of merchants. Other bots use a similar category based search to locate a list of recommended products. The Compare-It home page shows a list of linked bots. One of these is Flowerscompare; a scrollable window on this page shows a number of categories concerning types of flower arrangements; choosing one of these causes a list of suggested products form a number of different merchants to be displayed.
Category search is a simple menu based search, in which there are typically only two or three levels in the menu hierarchy. Such search approaches are recommended for novice users, since the categories prompt users. Each category is likely to contain a group of related products or merchants. Categories are helpful for browsing, with a view to learning more about the product range available before seeking more specific product information. As with any classified scheme or scheme based on categories the potential for problems lies with the users ability to choose the correct category for the product that they are seeking. For example, are golf shoes to be found under golfing equipment or sportswear?
Simple keyword search is another common option, which is often offered on the home page alongside category search. Sometimes keyword searches are offered alone, but in other interfaces they are available alongside category searches. The Consumer World Page incorporates some search tips. These hint at the problems that can arise with the superficially simple process of entering a keyword. These include potential variations on word form (plurals and singular), different terms for the same concept (e.g. Autos or Cars), and, searching for multiple word terms (e.g. credit cards). Searchers may also need to hit the correct level of specificity and specification. Specification is concerned with providing sufficient, but not too much, information to retrieve sites. For example, although a searcher may be able to specify both the title and author of a book, if the form in which these are specified does not match those in the database, the item may not be retrieved. A lower level of specification based, say, only on the author’s name might be more successful in locating the book. Specificity is concerned with being precise about the product that is being sought. Large databases will generate hundreds, if not thousands of hits in response to an imprecise search. Brand names can be a useful search term in this context. For example, if the purchaser requires running shoes, this more specific search term will provide a more focussed list of products than the terms shoes, or sports shoes. Keyword searching is particularly useful is searching for products with unique names.
Parametric searching, allows the searcher to be more specific. Parametric bots (Reda 1997) search using a series of parameters. Customers specify a range of choices by checking off boxes or selecting options in drop down boxes. When the customer clicks on the search button, the search is executed. Typically the parameters or characteristics to be specified (apart from price) are product specific. Parametric searching for products with complex technical specifications (such as digital cameras) assumes that the customers have product knowledge. Although customers need not necessarily specify all of the options offered, they need some understanding of the technical specification of the product that they are seeking to buy. A number of the bots which provide access to such products also offer product advice and information, as discussed below.
Help and tips are offered to support users with the search process. Simple keyword searching is often accompanied by search tips. Larger databases that offer advanced search options or parametric searching may offer more elaborate help. Buyers Index, a service which accesses over 11,000 Web sites and mail order catalogues, offers answers to the following questions about searching:
Criteria for the Evaluation of Shopping Bots
The focus in the previous section has rested with the product search process and those facilities designed to support such a process. The effectiveness of a shopping bot is also determined by a number of other factors. Search facilities and strategies are important in the identification of items within a ‘database’ but, as with any shopping situation, the product range, which is available to be accessed, is equally important. In the context of the buying decision, and other dimensions of the shopping experience, other information and advice and support facilities may assist the shopper. Shopping bots differ from one another in respect of a range of characteristics. These include:
Coverage of e-retailers. The extent of coverage of different sites is variable between shopping bots; a good range is important if individual consumers are to have a range of product options. Clearly, the product range available through any one bot is determined by the sites that are accessed. Bots can be selective in the products that they list from a specific site; this selection might be performed on the basis of product category or price range. This dependence on the product range of other sites can mean that the shopping bot does not have a product portfolio which has been designed specifically to meet the needs of its anticipated customer group. The product range may have idiosyncrasies, with some products being available from many more e-retailers than others; true comparative data may only be offered in specified product categories and not across the entire product range. Other aspects of coverage include:
Product range covered. As Figure 1 shows most of shopping bots specialise in specific product ranges. The product category which has attracted the most attention to date are information products, such as computer hardware and software, CD’s and books. At the other end of the range, few bots cover groceries; this may be related to the nature of the buying decision process associated with the product or the extent to which the product is being marketed in the e-commerce arena. A typical list of product categories for a relatively comprehensive shopping bot includes books, compact discs, auction items, health products, office supplies, toys and children’s presents, sporting goods, golf equipment, cook war, movies and video, electronics, fragrances, and computer software (as offered by E-Compare). Shopping bots may be most useful for those products which normally provoke extensive decision making (Brassington and Pettit 1997). Or, alternatively, shopping bots may be most effective for products that are normally bought by brand and specification, such as digital cameras, golf clubs or running shoes. It may be difficult to access a full range of such products through traditional retailers, or by visiting individual sites. For the present, the availability of shopping bots to cover some product categories is limited. There is work to do in identifying those product categories for which shopping bots can be most helpful to consumers. Further development of niche shopping bots to support purchases of specialised products, as, for instance, specialised products for those with restricted mobility, offers interesting possibilities.
Compare It describes itself as a shopping portal. It comprises a number of sites each covering a product category. Currently available sites include those covering books, software, CD’s, insurance, shoes, and colleges and universities. The coverage of the sites currently in beta version offers a telling insight into the potential for e-services, and e-retailing. These include: airplane tickets, autos, churches, contact lenses, cosmetics, credit cards, dating and romance, jobs, legal assistance, medical and dental, and real estate.
Geographical coverage. Many of the existing shopping bots primarily give access to products available in the United States. For European or UK consumers this will pose a number of challenges, associated with delivery, payment processes and currency. Electronically delivered information products can be delivered globally with little additional inconvenience, and suppliers of such products are likely to be accustomed to managing currency issues and payment arrangements. In other product categories, support for these aspects of e-retailing varies between retailers. Buyer’s Index allows users to limit their searching by categories, which include ship-to regions, and ship-from regions; this makes sense in a global marketplace. In general, a wider range of shopping bots will need to develop, which address themselves to specific geographical markets, where this is appropriate and necessary.
Evaluation, and Authority. Detailed product specifications, coupled with price can assist consumers in the identification of products. However, consumers also benefit from advice and evaluation. This evaluation can pertain to both e-retailers and specific products.
In the context of e-retailers, shoppers are concerned that they will receive a reliable service. The Consumer’s Association (UK) has just launched an on-line shoppers hallmark to boost confidence in buying over the internet. Under the scheme, UK traders which have been vetted by CA lawyers and which promise to abide by a special code of practice, can display the Which? Web Trader logo. The scheme means that the trader has undertaken to display prices clearly, with no hidden extras, to keep to a delivery deadline (or offer a refund), and to deal properly with any complaints. Traders must also clearly display their full contact details and their web sites must be secure. CA will do random checks on companies, and also investigate any complaints about them. Online traders pay nothing to sign up. Since this scheme is operated by an independent consumer organisation it may have more credibility than those, including the providers of shopping bot, who have a commercial interest in relation to e-shopping transactions. Another type of consumer service is provided by 20-20Consumer, an independent consumer comparison shopping service.
Some shopping bot providers also evaluate sites or merchants before they are included. For example, Bizrate offers both a merchant guide and a product guide. All merchants have been rated by customers and/or staff, and ratings are published. Merchants are rated on:
In other contexts, there may be a range of different types of strategic or financial arrangements between shopping bots and the e-retailers whose products are covered. There is significant scope for influencing and shaping purchase decisions. Ethical marketing takes on a new meaning in this context.
To return to products, some shopping bots offer various types of product details and evaluation. For example, Consumer World offers product comparisons, reviews and pricing services. This advice generally falls into one of the following three categories:
ConsumerREVIEW, and Open Discussion list for Online Shoppers offer access to consumer generated product reviews. Such services offer a unique opportunity for consumers to increase the impact of ‘word-of-mouth’ in marketing communications, and thereby to enhance the reputation of some products, whilst damaging that of others.
A key issue with all types of advice is their authority. Can and should, and indeed, to what extent do, consumer’s rely on such advice and use it to inform their decision making processes? Another issue is that of liability, should information or advice prove to be misleading. Indeed, there may be a wider issue of liability associated with any of the recommendations offered by a shopping bot.
The need for evaluation of retailers and products varies with the product category. Some commentators have warned of information overload.
Search facilities. The search facilities available with different shopping bots vary as has been discussed in the previous section. Optimally such facilities need to be appropriate for the size of the ‘product database’, the accuracy of that database, and the precision necessary in specifying search strategy.
Userfriendliness of the human computer interaction, deriving from factors such as the readability of screen displays, ease of navigation between screens, and the way in which the stages in the product search are presented to the user.
Intelligent agent functions. Some shopping bots learn from the search process, or make ‘judgements’ on the product range that might be of interest to a specific consumer, on the basis of a consumer profile that is provided by the consumer. For example Lifestyle Finder uses a range of indicators to perform a personal life style analysis and uses this as a basis for recommending specific products. MySimon is an intelligent bot that can imitate human navigational behaviour and can be taught to shop at thousands of merchants in hundreds of product categories. Tete-a-tete is a project within MIT Media Lab’s Agent-mediated Electronic Commerce initiative. It engages consumer owned shopping agents and merchant owned sales agents in integrative negotiations over the full value of each product offering to maximise their owner’s individual needs. These intelligent agent functions take the concept of the shopping bot beyond that of a purpose built search engine, to a tool for creating a personalised, or individualised ‘virtual store’. As in a conventional store lifestyle analysis is used to define the range of products on offer; the difference is that this analysis need not any longer pertain to a group of consumers, but can be personalised. On the other hand, some consumer’s may opt to replicate the real world shopping experience in the virtual world, and may prefer group profiles, and associated product recommendations designed to create a ‘lifestyle’ which is consistent with that of the reference group to which they aspire. Intelligent agents could also be trained to reflect lifestyle scenarios and to continuously update the product range associated with those scenarios.
Future Research and Development Agendas
Shopping bots are a useful innovation, and reduce some of the complexities associated with product searching in e-shopping. However, the interface associated with each shopping bot is different, and whilst interface design is often superficially user-friendly, there is a high cognitive load associated with performing a successful search using just one of these search tools. Some offer advanced search features which are likely to be an additional hurdle to the novice user since they may require that users have both a familiarity with search strategies and product specifications for a given product category. For some product categories, where non-routine purchases are the norm, it is common for consumers to learn a lot about a type of product during the purchase process. Shopping bots need to provide this advice and information, and consumers need to be convinced of the authority of the information that is provided. In reality the consumer also has to make a choice about the search tool that is used, and this choice will determine search outcomes in ways that consumers are unlikely to recognise. There is real scope for information overload.
Understanding of consumer behaviour in this environment is limited since, e-shopping is only just beginning to penetrate mass markets. As activity in this market grows, researchers will be interested to develop an understanding of the factors that affect consumer behaviour in general, and the role of shopping bots more specifically. Retailers will share an interest in these issues, but their perspective will be more applied, with a focus on strategies for optimising the visibility of their product offerings, and encouraging customers to buying decisions that lead to the purchase of their products. As ever, consumers will make a range of different types of buying decisions, and it is to be expected, for example, that information gathering habits will be different for different types of purchase decisions. Similarly, there will be a range of other factors, some associated with the buying decision, and others associated with the searcher’s experience and competence in performing searches that will influence search activities. One piece of recent work which examined tourist search behaviour (Fodness and Murray 1999) showed that tourist information strategies were found to be the result of a dynamic process in which travellers used various types of information sources to respond to internal and external contingencies in vacation planning. Individual tourist characteristics are also important, as are behavioural search outcomes.
Against the context of more general agendas associated with developing an understanding of product searching, information behaviour and consumer decision making during e-shopping, there are a number of areas for development of shopping bots. These include
In conclusion, shopping bots are an interesting innovation, which will prove attractive to shoppers if they are proven to support them in product searching in e-shopping. Further development of shopping bots, coupled with an intelligent evaluation of consumer responses to those developments may allow a more sophisticated understanding of the way in which consumers use information in the buying process, which may have application beyond the realm of e-commerce.
References
Blakeman, K (1997) Intelligent search agents: search tools of the future? Business Information Searcher, 7 (1), 16-18.
Bradshaw, R (1997) Introducing ADAM: a gateway to Internet resources in Art, Design, Architecture and Media. Program, 31 (3), July, 251-267.
Brassington, F and Pettitt, S (1997) Principles of marketing. London: Pitman Publishing.
Breitenbach, C S and Van Doren, D C (1998) Value added marketing in the digital domain: enhancing the utility of the Internet. Journal of Consumer Marketing, 15 (6), 558-575.
Cheeseborough, H and Teece, D (1996) When is virtual virtuous? Harvard Business Review, Jan- Feb.
Chowdhury, G G (1999) The Internet and information retrieval research: a brief review. Journal of Documentation, 55 (2), 209-225.
Clarke, S J and Willett(1997) P Estimating the recall performance of Web search engines. Aslib Proceedings, 49 (7), 184-189.
Dong, X and Su, L T (1997) Search engines on the world wide web and information retrieval from the Internet: a review and evaluation. On-line and CD-ROM Review, 21 (2), 67-81
Ellis, D, Ford, N and Furner, J (1998) In search of the unknown user: indexing, hypertext and the World Wide Web. Journal of Documentation 54 (1), 28-47.
De Kare Silver, M (1998) e-schock: the electronic shopping revolution: strategies for retailers and manufacturers. Basingstoke: Macmillan
Dertouzos, M L (1997) What will be: how the new world of information will change our lives. Harper Edge.
Fodness, D and Murray, B (1999) A Model of tourist information search behaviour. Journal of Travel Research, 37 (3), 220-230.
Grain, A (1996) Will multiples miss the home shopping boat? Grocer, 20 July, 218, 7263, 4.
Laursen, J V (1998). Somebody wants to get in touch with you: search engine persuasion. Database, 21 (1), 42-46.
Lynch, D and Lundquist, L (1996) Digital money. Wiley
MacMillan, G (1997) Is Internet shopping the panacea for all consumer woes? Campaign, 2 May, 33.
Markham, J (1998) The future of shopping. Macmillan
Mitchell, R (1995) Safe passage in cyberspace. Business Week, 3416, 20 March,33.
Poulter, A (1997) The design of the World Wide Web search engines: a critical review. Program 31 (2), 131-145.
Reda, S (1997) Improved search function seen as key to Internet shopping. Stores, 79 (8), 60-67.
Resnick, R (1995) Business is good, not. Internet World, June, 71-73.
Rowley, J (1998) Internet food retailing: the UK in context. British Food Journal 100 (2), 85-95.
Rowley, J and Slack, F(1998) Designing public access systems. Aldershot: Gower.
Tapscott, D (1998) Growing up digital. McGraw-Hill.