The numbers of papers, studies and articles from the scientific community are still few. However, starting with the field of consumer behaviour, a few important articles should be mentioned. Firstly the articles by and in Journal of Marketing, secondly the articles by , and which have a managerial focus. In the newer journals mentioned above, the most relevant articles are: , , , , They discuss various aspects of web, consumers, market mechanisms and market structures.
How do auctions fit into this picture? When discussing products and structures of EC, writes; ‘Market-based co-ordination can be classified into four categories (as in Garbade 1992): direct-search markets (where the future partners seek out one another), brokered markets (with the brokers assuming the search function) dealers market (with the dealers holding inventories against which they buy and sell), and the auction markets.’ Zwass distinguishes between hierarchies of individual firms and open marketplaces. Market-based co-ordination belongs to electronic market places with multiple buyers and suppliers, not to electronic hierarchies which are long-lasting supplier-customer relationships between firms.
This is one point of departure. Another is to look at the attention auctions on the Internet is given in recent magazines, reports and articles. The company Jupiter Communication (http://www.jup.com/newsletter/) wrote in one of their newsletters (January 1997); ‘The auction concept is perfectly suited to the Web, and one that will likely become commonplace as the medium matures.’ Business Week September 29, 1997 has a mini-portrait of Jerry Kaplan, the co-founder and chief executive of Onsale, the ‘leading Web auctioneer’ (Moran, 1997, p. 58). The sale is expected to exceed $80 million in 1997 and Kaplan’s shares of Onsale is worth about $141 million (!) even though the surplus (profit) has only been in the range of $50000 to $140000 since the start in May, 1995.
Thus, it appears that auctions on the Internet are worthwhile studying. The purpose of this paper is to describe a prototype of an online auction and what we have learned from testing it with a group of students. The paper is organised in the following way: 1) examples of electronic auctions and Internet-auctions like Onsale.com 2) to describe the prototype developed by NR, 3) the test of the prototype and the web-questionnaire to collect consumer data, and 4) a short discussion with concluding remarks.
This auction-project was done in co-operation with one
of our partners in the research program ‘Open Networks as the Future Marketplace‘
Auction Web by eBay (http://cayman.ebay2.com/aw/index.html)
According to their press release of November 4. 1996, eBay announced ‘the world’s first service enabling consumers and businesses to bid for airline seats, hotel rooms, cruise berths, vacation packages, and other travel-related products over the World Wide Web.’ A year later they claim to be ‘the largest and longest-running person-to-person auction on the Web.’ More than 3 million items have been advertised for sale since the start.
eBay acts solely as a listing-agent. Auctions typically last 3 - 7 days from the placement of an advertisement, based on the duration the seller selects at the time the advertisement is placed. The sellers and bidders must fill in a registration form with e-mail address, name and a complete address. Then a password is sent by e-mail and has to be used in a reply and confirmation sequences. The bidder uses a ‘bid-button’ to send to eBay information about the maximum amount he is willing to pay for the item. This amount is kept secret from other participants and eBay is bidding on behalf of the person as other bidders increase their bids. eBay automatically terminates the bidding at the end of the auction, and notifies the seller and the highest bidder via e-mail. Then, the buyer and seller should contact each other within three business days of the end of the auction.
Onsale is specialising in computer equipment and has two auction formats called yankee auction and standard auction. At the yankee auction a number of identical items are offered for sale at the same time. When the auction closes, the highest bidders win the available inventory at their actual bid prices. Prior to closing, the currently successful bidders are listed in priority order, including their price, quantity, and time. The yankee auction closes at the posted closing time or five minutes after the last bid is received, whichever is later. Therefore, according to Onsale.com the auctions may remain open after the posted closing time during what is called the "Going, Going, Gone" period.
All standard auctions bids are 'proxy' bids where Onsale keep bidders actual bid secret and make lower bids as necessary on bidders behalf. If there are no other bidders, Onsale places a public bid in the bidders name at the minimum bid price. This means the bidder never pay more than a little above what someone else is willing to pay, and so are free to place the highest bid privately right from the start according to Onsale. The standard auction closes at the posted closing time or five minutes after the last bid is received, whichever is later. Therefore, the auctions may remain open after the posted closing time during this "Going, Going, Gone" period.
Travelfact auction (http://www.travelfacts.com/tfacts/htm/bidgen.htm )
This auction offers airline tickets, hotel-rooms, cruises and travel items. The bidder must first register on the site and will then be sent a user-ID and a password via e-mail. To place a bid the bidder must enter the user-ID and password along with an amount. Minimum bids are specified. Auctions usually last 7 - 14 days and the highest bidder is notified by e-mail.
Most often, e-mails are used for the bidding - cf. the
Internet Auction List. There are some auction-sites with ‘live auctions’
according to the advertisements. However, we haven’t been able to try any
of these (credit-card numbers are required.)
As a part of the test we developed a web-questionnaire that the participants were asked to complete. We received 14 answers. Two of them were from the scientists and some of the questionnaires were only partly filled out. One of the purposes of presenting the answers is to show to which extent the participants used the fixed alternatives when answering. This is in other words a pre-test of the questionnaire. Secondly, the answers can give an indication of the relevance of the given questions. Should this concept be extended to a larger audience and be used for investigating consumer behaviour, it is important that the good and well-considered questions are asked.
We had enough participants to create traffic on the auctions, but the sample has obvious weaknesses with regard to consumer behaviour. The participants are not real customers, they are invited to participate and they are few.
The participants received the URL to the auction and web-questionnaire 45 minutes before the auction started. It means that they didn’t have much time to think about if they liked it or not before taking part in the test. Furthermore, to participate the students had to send an email and received an id-number in response. This number was the registration number. Some had problems logging on to the auction, but were assisted via e-mails. These problems were mainly due to incorrect configuration and/or too old versions of the browsers.
The auction consists of three parts from the view of a potential bidder. The first web-page is the introduction and log-on page. When registered the bidders can continue to what is called the ‘show-room,’ and finally to the ‘auction-room’ In the show-room the participants can choose destinations and flights in addition to information about when the next auctions will start. In the auction-room the bidder had to write the amount he or she was willing to pay and press the send-button if interested in the ticket. (cf. English auction) Otherwise one could simply observe the bidding, included the number of active bidders. (i.e. bidders present in the ‘room’ that had actually made bids)
Within the test-period of 1 hour and 15 minutes eight auctions were run. The test-participants were not using real money. They were given the instruction only to bid a little bit higher then the former bidder in the first auctions. Then they could behave as they pleased, but were asked to image a real situation when bidding. As a result, the bidding-process went quite fast.
Even though this is test and not a real auction, it is important to look at the characteristics of the participants. For an airline there are some critical questions with regard to an Internet-auction; for instance will it be used by a certain segment of our customers, is it likely that it will attract new customers, is it easy to use and is it good enough (and stable) enough technically?
It is not unlikely that a number of students could be an interesting market segment for such an auction. A high percentage of them have access to the net, they have some flexibility with regard to when they can travel, and the price of the ticket is important - most will look for a good bargain. Moreover, they are often experienced web-users. Thus, even though our sample is a convenience sample and a very small sample, it consists of young students very familiar with the web. It should be added that half of the test-group consisted of computer science students. They were chosen because we expected that they would be willing to give constructive feedback about the auction-model. In comparison with other web-users in the student community, they are too experienced. However, quite a few of them might be willing to test new commercial services on the net.
The test was held in mid April 1997. In Norway, as in the rest of the world, EC is still in its infancy. When doing interviews related to consumer behaviour background information might be useful to understand future behaviour. Of the test-participants eight had previously bought something on the Internet and six hadn’t. Six said they would participate to save time, five to save money, two the options (number of alternatives) are better and three used the open alternative answering-box writing; ‘It is easier to get information’, ‘I don’t have to go anywhere - I don’t have to send snail mail’ and ‘It was the only possibility.’ We also included a question about payment methods, and five answered that they had sent their credit-card number though mail - not encrypted, two answered credit-card number - encrypted.
In addition to being a technical test, this was also a test of the web-questionnaire. The ideal situation is that the test-participants are similar to (have the same characteristics as) the potential real customers. Then it is important that the test-participants are using the fixed alternatives. If so, the instrument, in this case the web-questionnaire is functioning satisfactorily and it could be recommended for use with real auction-participants. We will argue that the test-participants are an interesting group in this context. However, there are a few inconsistencies in their answers.
The answers of the test-participants can only serve as an indication of the potential of an Internet-auction. For instance, if nearly all of them had been negative to the use of an auction and to the functionality of the prototype, it would have been a warning-signal - it would have been risky to pursuing this idea further. However, even if all the test-participants are very positive, definite conclusions should not be drawn.
The next question is about future behaviour. Even though somebody answers ‘I will take part in an auction’ in the future, it does not mean that he or she actually will. ‘Do you think you will take part in an (weekly) auction in 1997 if an airline launched an auction like this one?’ 8 answered ‘I am not sure, but it is an alternative I will consider’, 2 answered ‘Yes, I think so’ and 4 the ‘other-alternative’
‘If participating in such an auction in the future, what would you emphasise?
(1 = first priority, 2= second priority.......)’
The alternative ‘The possibility of getting a good bargain’ was rated by 8 as their first priority, and by one as second priority. For ‘An Internet-auction is a practical (convenient) way of buying ticket’, the numbers for the priorities were 1, 7, 2 and 1.
We expected that price should be the most important parameter to the test-participants. Price is a salient attribute in general. However, it can be assumed that quite often it is a set of variables that influence on consumer behaviour.
The next two questions are related to the Norwegian context. It is important to tailor the questions to what will be the alternatives for the auction for the potential customers. The first one is about standby-tickets for young people. Ten answered yes, a standby-ticket is a relevant alternative to me, and two answered no.
‘A standby-ticket or the ‘auction-ticket’ - which one would you prefer?’
The auction-ticket (round-trip) even if it costs 50 NOK ($7) more: 3
The auction-ticket (round-trip) even if it costs 100 NOK ($14) more: 4
The auction-ticket (round-trip) even if it costs 200 NOK ($14) more: 0
Two standby-tickets even if they cost 50 NOK ($7) more: 2
Two standby-tickets even if they cost 100 NOK ($14) more: 0
Two standby-tickets even if they cost 200 NOK ($28) more: 1
In Norway a standby-ticket (one way) normally costs 400 NOK ($60) and a lot of young people such as students are travelling this way. In a real market situation we hypothesise that quite a few will prefer the traditional standby-ticket. However, after taking part in this test and answering the questionnaire, we expected some answers in favour of the auction-ticket alternative. From the seller’s point of view it is very useful to know not only that price is regarded as an important variable for the market segment, but also the relevant limits/thresholds. Whether there should be a maximum auction-price is a complicated question. For instance, Cathey Pacific experienced that the successful bidders paid more than normal price (full-price) for the auction-tickets (Los Angeles - Hong Kong) on their first auctions (in 1996) - then it dropped. In general it is difficult to interpret answers from respondents about their price-sensitivity. How many of them will give honest answers and not tactical ones to influence the price policy, is hard to know.
It is important for the auctioneer that the bidders are serious. The test-participants had to give away their email-address to get to what was called the auction rooms. It is not obvious that it is necessary to include a question about giving away personal information to the auctioneer? It is hard to image that a company is willing to let anybody participate in an auction without giving up some anonymity. To the question: ‘To what extent does it matter to you having to give away your email-address to get to the auction-room?’, ten answered it is not important and four answered yes it is to some extent. This question might be superfluous if it is necessary for the auction house to have the email-address from the bidders in order to deliver the goods. (cf. 2.2)
It was easy to find the relevant flights: 3
It was quite easy to find the relevant flights: 7
It was somewhat difficult to find the relevant flights: 1
It was difficult to find the relevant flights: 2
Even though the sample is very small, this is a useful piece of information. This part of the auction functioned reasonably well, but can be improved.
‘The ‘auction-room’ - how did the bidding function?’
It was easy to follow the bidding: 5
It was quite easy to take part in the bidding: 1
It was somewhat difficult to take part in the bidding: 5
It was difficult to take part in the bidding: 3
These answers reflect something very critical for the bidders in an online auction. In a normal auction-room in an auction-house all the bidders receive the same information at the same time. When using the Internet this is not the case due to minor transmission delays. However, if the bidders can observe that their bids are shown one or two seconds later, they might accept this delay as ‘the name of the game.’ In this test, the log showed in one incident that the bids were received and accepted with an interval of 2 seconds. When not using real money it is easy to respond very fast and raise the amount all the time. We cannot make any conclusion regarding whether the speed of the bidding was unrealistic. The bidding could have been more smooth if a the participants could choose between the standard bid-button and another bid-by-a-fixed-increase-button. And based on the feedback in the test this seems to be necessary. In a real auction, it is easy to get an intuitive feel for the atmosphere of the situation, such as the number of persons in the room, the number of bidders, the bidding frequency, etc. All this factors enables a potential bidder to get a good feeling as to how his chances of placing a winning bid will be. This same feeling is hard to convey electronically. We have tried to give some of the same information by presenting some of these figures as numbers in the user interface, but this will probably not provide the same sense of ‘community’ as a real auction.
To the question; ‘When being the successful bidder, the auctioneer will send you a confirmation. When would you like to confirm to the auctioneer that you accept that you have bought the ticket?’ One answered within 5 minutes, two answered within 10 minutes, three within 30 minutes, four within 1 hour and three within 2 hours.
The seller wants to be certain that a winner will actually pay for his ticket. At the moment, when the payment mechanisms are not well established, one alternative is to mail a reference number, and have the winner to pay for his ticket at the airport. This is the solution we have suggested in our prototype.
On the question; ‘As the successful bidder, what is the acceptable and unacceptable response time (for your answer/confirmation)?’ the result was as follows: four answered it is acceptable having to answer within 30 minutes, seven answered within 1 hour, and three within 2 hours. Moreover, one answered it is unacceptable having to answer at once, six within 30 minutes, one within 1 hour and four within 2 hours. Three questions were used in the test on this issue. Most likely, it is enough to ask about what is the unacceptable response time and how soon he would prefer to receive the confirmation. It seems that the ‘within 2 hours’ alternative was unnecessary. Since the bidders are present when the auction closes, the confirmation sequence should be executed in relatively short time after the auction.
To summarise, it seems that the web-questionnaire did
function satisfactory. After the test we had some e-mails from some of
the participants enabling us to identify some minor weaknesses in the questionnaire.
However, some improvements have been made. We think web-questionnaire can
be a useful instrument to monitor consumer behaviour in this context.
As part of the project, we were interested in investigating how the new and emerging technologies of the web would be suitable for implementation of the auction prototype. In particular, we wanted to find out whether the highly interactive nature of an auction could be realised with these technologies, or if there were scalability or other constraining issues that would limit the usefulness of such an application.
The auction system was implemented using a common 3-tier architecture.
The middle tier is the co-ordinator of the auction. It communicates with the database, inserting new flights for sale into the auction room. In addition it handles connection requests from bidders, registering them and their subsequent bids into the database. It is also responsible for broadcasting events to all participants, such as information about a new highest bid. This application is implemented as a standalone, multi-threaded Java application, with one thread pr connected participant. It communicated with the database through the Java Database Connectivity Library (JDBC), and with the participants through a socket-based TCP protocol.
The client tier is a standard Java applet. Participants initiate a TCP connection to the auction engine by registering some information about themselves. All subsequent interaction with the auction is then performed via a proprietary "auction protocol" on top of this TCP connection. Later versions of the auction protocol should probably be based on a standard wire-protocol, such as Remote Method Invocation (RMI). RMI was not available in browsers at the time of implementation, and was thus not a feasible option.
We found few scalability problems with our group of participants.
In internal tests, we have simulated a higher number of participants, with
few signs of degradation. We feel it is fair to say that the potential
success of interactive web-based auction systems is probably not hindered
by technical issues. It would be relatively easy to scale the solution
by adding more instances of the auction engine in the middle tier. In real
life, there is a physical limit as to how many people can be present in
a room at the same time. It will probably be wise to enforce a similar
upper limit on electronic auctions as well, to make sure that the performance
of the system does not drop below an acceptable level as a result of attracting
a too big audience.
An auction-prototype and a test such as the one described can give useful information to a company interested in an on-line auction as a sales tool. However, a complex set of factors will probably influence whether it will be successful or not in the market. An auction has to attract enough bidders (a critical mass) otherwise it might die slowly. If the company has a popular web-site, it is easier to attract customers or potential bidders to an auction. Moreover, the company has to be aware of false bidders. Thus, a trade-off between a low threshold for participation and the risk of sabotage has to be made.
An airline ticket can be view as a commodity, particular in a domestic market. Therefore, from a company’s point of view, an on-line auction should be integrated with yield management and sales automation systems. If the airline can release the number of tickets and set the minimum price automatically, the operating costs can be small. For an airline, a number of tickets are picked up and bought at the airport. As a result the distribution of the tickets and the payment is not a difficult to arrange. Many flights have a number of empty seats. For a carrier a low profit on these seats is better than nothing. If new customers can be attracted due to the low prices without cannibalising existing market segments, on-line auctions as a sales tool could be a viable supplement.
The possibility of making a good bargain has always been a way of attracting potential customers. In addition, some people find satisfaction in being able to out-bid others like Kaplan, Onsale’s owner; ‘I love to out-bid people. It’s entertaining and I love a good bargain.’ (Moran, 1997) An auction could be a part of a company effort to built a community around their web-site. Armstrong and Hagel (1996, p.136) distinguish between communities of transaction, interest, fantasy and relationships. When discussing the differences between these communities they write; ‘Travellers might be invited to join game hosted by an airline running a special deal (community of fantasy)’
Based on our experience, we expect to see an increasing
number of auctions on the web. It is also likely that new concepts and
auction-models will be tested and utilised. If the prototype technically
is functioning well, the next step can be either to introduce it directly
into market or to continue the testing. The first alternative is more risky,
but since there are few (real) on-line auctions today, the advantage of
being early must not be under-estimated. The second alternative is less
risky. However, it is important that the test-environment is very similar
to a market situation. The stock price of the company Onsale shows that
there are investors expecting a profitable future for such a company.
Alba, J., Lynch, J., Weitz, B., Janiszewski, C., Lutz, R., Sawyer, A., & Wood, S. (1997). Interactive Home Shopping: Consumer, Retailer, and Manufacturer Incentives to Participate in Electronic Marketplaces. Journal of Marketing, 61(July), 38-53.
Armstrong, A., & Hagel III, J. (1996). The Real Value of ON-LINE Communities. Harvard Business Review, May-June, 134-141.
Deighton, J. (1996). The Future of Interactive Marketing. Harvard Business Review(November-December), 151-162.
Garcia, D. L. (1997). Network Commerce: Public Policy Issues in a Deregulated Communication Environment. The Information Society, 13, 17-31.
Gogan, J. L. (1997). The Web's Impact on Selling Techniques: Historical Perspective and Early Observations. International Journal of electronic Commerce, 1(2), 89-108.
Hoffman, D. L., & Novak, T. P. (1996). Marketing in Hypermedia Computer-Mediated Environments: Conceptual Foundations. Journal of Marketing, 60(July), 50-68.
Hoffman, D. L., Novak, T. P., & Chatterjee, P. (1995). Commercial Scenarios for the Web: Opportunities and Challanges. Journl of Computer Mediated Communication, 1(3).
Jarvenpaa, S. L., & Todd, P. A. (1996-97). Consumer Reactions to Electronic Shopping on the World Wide Web. International Journal of Electronic Commerce, 1(2), 59-88.
Lee, H. G., & Clark, T. H. (1996). Impacts of the Electronic Marketplace on Transaction Cost and Market Structure. International Journal of Electronic Commerce, 1(1), 127-149.
Moran, Susan (1997) Going...Going...Downloaded. How Onsale, the leading web auctioneer outsmarts rivals. Business Week, September 29, p.58
Neo, B. S. (1992). The implementation of an electronic market for pig trading in Singapore. Journal of Strategic Information Systems, 1(5), 278-288.
Quelch, J. A., & Klein, L. R. (1996). The Internet and International Marketing. Sloan Management Review(Spring), 60-75.
Reck, M. (1994, ). Types of Electronic Auction. Paper presented at the Conference ‘Information and Communications Technologies in Tourism’, ENTER'94 Innsbruck.
Wigand, R. T. (1997). Electronic Commerce: Definition, Theory, and Context. The Information Society, 13, 1-16.
Wigand, R. T., & Benjamin, R. I. (1996). Electronic Commerce: Effects on Electronic Markets. Journal of Computer-Mediated Communication, 1(3), http://shum.cc.huji.ac.il/jcmc/vol1/issue3/wigand.html.
Zwass, V. (1996). Electronic Commerce: Structures and Issues. International
Journal of Electronic Commerce, 1(1, Fall), 3-23.