Prestel's Lessons For Americans

E. Scott Maynes, Cornell University
Prestel, the British version of videotex, has been operating as a commercial information storage and retrieval system for several years. During 1981-82 the author undertook a critical review of Prestel from the consumer viewpoint. The fundamental question explored was the extent to which Prestel provided satisfactory answers to questions addressed to it by consumers, as compared with alternative information sources. This paper summarizes lessons" for Americans that emerged from the author's review of Prestel.
[ to cite ]:
E. Scott Maynes (1984) ,"Prestel's Lessons For Americans", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 520-524.

Advances in Consumer Research Volume 11, 1984      Pages 520-524

PRESTEL'S LESSONS FOR AMERICANS

E. Scott Maynes, Cornell University

[This paper represents a culling for the author's monograph (Maynes 1982) and is a revised version of a presentation made to the American Council on Consumer Interests in March, 1983.]

ABSTRACT -

Prestel, the British version of videotex, has been operating as a commercial information storage and retrieval system for several years. During 1981-82 the author undertook a critical review of Prestel from the consumer viewpoint. The fundamental question explored was the extent to which Prestel provided satisfactory answers to questions addressed to it by consumers, as compared with alternative information sources. This paper summarizes lessons" for Americans that emerged from the author's review of Prestel.

INTRODUCTION

The Central Issue: A Dream, A Nightmare, or a Mirage?

Prestel is the British proprietary name given to the information storage,'retrieval system known generically as "videotex" or "viewdata."

From the consumer viewpoint Prestel represents a fantastic dream--an information system that, suitably developed, would enable users greatly to improve their standard of living and their ability to choose effectively. By improving the functioning of markets, Prestel might confer similar if perhaps lesser benefits on non- users. Inappropriately developed, however Prestel could turn out LO be a nightmare, rather than a dream, for consumers. It might lead them to "bad buys" rather than "good buys" and rob them of their purchasing power, both ind visually and collectively.

Or Prestel could turn out to a mirage. Remember an invention called Picturephone? It was a device that enabled you to see the person on the other end of the telephone line. It worked, but no one wanted it!

This paper has three parts:

i. An Introduction to Prestel

ii. The Nature of the Author's Research

iii. Prestel's Lessons

AN INTRODUCTION TO PRESTEL

[For the official version of Prestel's progress, cf. (Hooper, 1981). The most convincing scenario of Prestel's future is that of (Campbell and Thomas 1981). For a general description of how Videotex works and of its implications for marketing, see (Talarzyk and Widing 1982). For an annotated review of the 20 or so active videotex projects in North America, see (Talarzyk and Widing 1983). For a consumer's view of the possibilities of Prestel, see (Mitchell 1981). Finally, for the inventor's view of Prestel's view see (Fedida and Malik 1579).]

The System Described

In essence, Prestel is an information system in which users seek answers to their questions by using their telephone line to call onto their home or office television screen items of information that are stored in a central computer. If Prestel can provide better answers more cheaply and more conveniently when competing sources of information -- periodicals, travel agents, yellow pages, and etc.,--then it will be a "winner".

Prestel is one of seven fully operational, commercial videotex systems in the world. [Prestel was first. The names and sites of commercial videotex systems are Bancshare, Columbus, Ohio; CompuService, Comp-U-Card, and Comp-U-Store all in Columbus, Ohio; Teleguide, Toronto; Grassroots, Manitoba: The Source, Readers Digest Association. See (Talarzyk and Widing 1983) for detailed descriptions of these systems.] In Britain in January. 1982 there were about 15,000 Prestel sets in use, of which about 2,000 were in residences. (This contrasts, vividly, with a 1978 Post Office forecast of 500,000 sets in place by the end of 1981, of which 95 percent would be sited in residences.) Subscribers had access to 180,000 pages of information supplied by 65% British and foreign information providers, available on a 24-hour basis with a high degree of reliability. For 200 ($300-350) one could buy an adaptor that would convert a black-and-whiLe television set into a Prestel receiver. Input devices, both numeric and alphanumeric, as well as printers were on sale.

Information retrieval had been available on Prestel from its beginning in September, 1978. In September, 1981 user-to-user messages were introduced, making possible the ordering of tickets or merchandise, entering quizzes, conducting public opinion polls, and the like. Early in 1¦82 Prestel introduced "Gateway," a service that enabled the user to conduct additional transactions and to have access to "outside" data bases.

Some Sample Questions

To convey the flavor of the Prestel system, four sample searches are summarized below. Details of each search are recapitulated in the Appendix of (Maynes 1982).

Exchange Rates. This search revealed Prestel at its best. The first step was to look up "Exchange Rates" in the Prestel Printed Index, i.e., the "Yellow Pages," then the appropriate page number was pressed in the control keypad. Thirteen seconds later (!) there appeared on the television screen the current rate of exchange of the pound against ten or so other currencies, updated 20 minutes ago. The charge for the search was $.15 (8P). A marvelous performance, cheaper and swifter than the telephone, swifter and more current than The Financial Times!

A Trip by Motorail. Motorail is a service by which the traveller and his car go on the same train. The purpose of this search was to ascertain departure times, fares, and availability for a Motorail trip from London to Perth. This search started with the entry, "Railways," in the Prestel Printed Index. It continued using the internal free-structure index that takes you from page to page through the hierarchically organized index and information files until you obtain the information desired. This internal cree-structure index, like a taxi-meter, incurs time charges as you use it, at a rate of $7-8 a hour during daytime hours and $1.60-$1.75 at other times. This search took 20 minutes, cost about $2.50, but yielded the desired information.

This search illustrates a limitation of the tree index: success depends on familiarity with the exact phrases used in the index. In dealing with a reservation clerk in a railroad station or on the telephone, you can ask about a service in which you and your car go on the same train." and expect the clerk to understand. Not so with this computer index which responds only to the word "Motorail."

Motor Insurance ("Automobile Insurance")

The object of this search was to ascertain (1) what motor insurance coverage .s desirable, (2! what was offered, (3) at what prices and from whom. The search involved three starts.

The first star: aborted almost from the outset. The page that was supposed to contain information about "Motor Insurance" according to the Prestel Printed Index pertained instead to Travel Insurance, not Motor Insurance. Thus the user was thwarted by an indexing error.

How to proceed? A second start was made by looking up "Insurance" in the Prestel Printed Index. Sure enough, this start led to a subheading "Motor Insurance" with three .r.format ion files. The first, offered by the Norwich Union Insurance Company consisted of typical public relations copy, telling of the company's history, size, and etc., but offering no concrete information helpful to a potential purchaser. A second file, offered by the Royal Automobile Club, for which the user paid a $0.03 "frame" or page charge, described "Third Party Coverage" well and "Comprehensive" unclearly. A final file under "Motor Insurance" was offered by the National Union of Teachers but was relevant only to teachers.

Thus ten minutes and a dollar's worth of search, yielded nothing. No useful information had been obtained and the user was irritated by the irrelevant public relations copy. This search illustrates a pervasive problem of Prestel: the user is often unable to discern in advance whether its files will provide the information he seeks.

New Car Purchase: What New Car Will Fit the User's Needs, Budget, and Offer Good Value for Money?

This was a search for advice on what new cars represent "good value for money" in the different size classifications--compact, mid-sized, etc. This is a type of information provided by Consumer Reports in the United States and by Motoring Which? in Britain.

Five minutes on Prestel employing every approach he could think of left the searcher stymied, unable to obtain the information he wanted and also unsure as to whether Prestel's files contained the desired information. Unfortunately, Motoring Which's files in Prestel are organized on the basis of particular models: the searcher must designate a particular model before he can obtain adequate information about it.

This search cost $1.00 and five minutes, but left the searcher frustrated and unhappy with Prestel. why hadn't they organized their files so as to be useful to ordinary inquiring consumers? Why hadn't Consumers' Association set themselves up to provide the advice that they offer so helpfully in Motoring Which?

Later the same searcher explored Prestel tiles relating to particular cars (e.g., a BL Mini) and found them most informative.

THE NATURE OF THE AUTHOR'S RESEARCH

Prestel's "Lessons" rest on two research projects on Prestel carried out by the author in 1¦81-82. The reader should know something of the nature of these two research projects. It should be noted that the projects focus exclusively on Prestel as an information storage and retrieval system.

The first was an assessment of the adequacy of the Prestel data base, i.e., the extent to which the Prestel data files provided satisfactory answers to the questions asked. The adequacy of the Prestel data base was rested by posing and seeking answers to a small set of questions that any consumer might address to Prestel. The questions and answers described above are representative.

This study was not a test of consumers' reaction to Prestel per se. The data base may be tested by a single individual. The validity of such a rest rests on its documentation: others may examine the documentation, reproduce for themselves the steps taken by the investigator and then judge whether they agree with the assessment.. On the other hand, a full-scale test of consumers' reactions to Prestel would require the exposure of a representative sample of users to the system under realistic conditions; it cannot be carried out by a single investigator.

This particular project by the author was conceived as a pilot study whose purpose was to explore the data base and to develop a methodology for assessing it, exhaustively. It became a substantive study when first the author and then the Steering Committee of the Prestel For People Project became convinced that a small-scale investigation--it dealt with only 16 questions--was itself persuasive and that further efforts would be wasted. [In his role as Consultant to the National Consumer Council, London the author was in effect acting as research agent for the Steering Committee of the Prestel For People Project. The Steering Committee consisted of U.K. government officials, all of whom were highly familiar and expert with respect to Prestel. They included Joan Macintosh, National Consumer Council, Chairman; Sir John Barran, Central Office of Information; Alex Eastabrook, National Association of Citizens' Advice Bureaux; Jeremy Mitchell, National Consumer Council: St. John Sandringhana, Consumers' Association Ann Smith, National Consumer Council; Robin Yeates, London and South Eastern Library Region; Norman Willis, Council for Educational Technology.]

The second study consisted of a critical review Of Prestel from the viewpoint of the consumer. In undertaking it, the author drew on the literature of Consumer Economics and Consumer Behavior as well as everything available on both Prestel and consumer information systems in general. The persuasiveness of this paper rests not on statistical tests or formal models, but rather on the cogency of the arguments and the evidence presented.

Readers wishing to examine these studies in detail should consult (Maynes 1982).

PRESTEL'S LESSONS

Why a Consumer/Household Rather Than A Business Information System?

Prestel (with apologies to Jean Baptiste Say) is a graphic example of "supply seeking to create its own demand." In a nutshell, the British Post Office "wondered" what it might do of a socially useful nature that would make commercial use of the immense excess telephone capacity that existed during non-business hours. Sam Fedida, the Manager of Computer Applications in the Research Establishment of the British Post Office provided an answer by inventing Prestel in 1972. In its first stage Prestel was devised as an information system that would permit consumers, especially during non-business hours [Reflecting excess capacity, Prestel's rates for nonbusiness hours are 22 to 25 percent as high as daytime rates.] to call up onto their home television screens answers to a wide variety of questions. The answers to the questions and the primary index that would enable the user to locate the question were stored in central computers. But telephone lines would be used to select and transmit answers, Thus at once vastly increasing the revenues, the economic importance, and the social usefulness of the telephone system operated by the British Post Office. [In mid-1982 the British Post Office performed mitosis, its electronic components--telephones, telegraph, radio and TV transmission as well as Prestel--becoming "British Telecom" while the traditional mail-related services were retained by "The Post Office."]

From this hope and demonstrations of the workability of the system it was only a short step to the prediction that 95 percent of Prestel receivers would be sited in homes. As noted earlier, the first three years' experience totally upturned this prediction: instead of 95 percent of Prestel receivers being located in residences, 95 percent were sited in businesses!

What went wrong? It is the author's belief that the promoters of Prestel w-ere so confident of their concept that they did not bother to test it. After demonstrating the feasibility of Prestel as a home information system, they assumed that consumers would want it. During his year in London, the author was unable to discover any tests in the public domain that were designed to ascertain whether Prestel would in fact meet consumers felt information needs. The author did encounter studies what explored such subsidiary questions as the optimal type faces and colors as well as alternative formatting of the tree structure index. But he was unable to locate any study (prior to those reported here! that was undertaken ascertain whether Prestel would in fact provide satisfactory answers to questions put to it by consumers.

Prestel's first lesson is clear: test, don't assume!

The Adequacy of the Prestel Data Base

Prestel tended to provide answers that were either "highly satisfactory" or "highly unsatisfactory" with few in between.

Examples of successful searches resulting in satisfactory answers were information about exchange rates, the advantages and disadvantages of inflation-indexed government bonds, the pros and cons of a particular model of used. cars, Motorail fares and rates, and prices for a wide variety of shavers (bur all from a single; large retail chain!. Unsuccessful searches included air schedules and fares from London to Copenhagen to Oslo and return, weather during the next week at a vacation site, stoves, bicycles (no information at c 11, despite the index heading), advice on value-for-money in a new car, prices of motor insurance. Problems that plagued users in this small study included: high costs, when known; subject matter entries with too little useful information; indexes that were time-consuming to use, costly, errorprone and frustrating; incomplete instructions to users; too few topics; and files not organ.zed to meet the user's needs.

Is the Prestel data base likely to be improved in the short term. It seers unlikely. Responsibility for the number and quality of information files rests with individual information providers and not with the government agency that "runs" the system in general. And here Prestel is in a "Catch 22" situation. Unsatisfactory answers to questions posed--if the author's experience is any guide--will turn away new subscribers. At the same time Prestel's current audience is too small to induce information providers Lo improve their files.

Again, the lessons are clear.

1. First, some organization must be assigned responsibility for maintaining the quality of files;

2. The monitoring of the quality of the files is a first-priority task that must be resolved at the outset lest an American Prestel not get off the ground.

Unsatisfactory Indexes

Indexing is the bane of any information system. So, too, wit h Prestel. This is ironic because the inventors of Prestel, in creating the "tree index," felt they had devised an easy-co-use index.

Prestel has three indexes, all unsatisfactory for reasons spelled out below.

Prestel's primary index is the internal tree index, organized hierarchically by subject matter. As an example, the sequence of Prestel pages consulted in the "Trip by Motorail" search was as follows: Railways, Stations from London, Type of Service, Route, Destination, Day or Night Service, Schedules, Fares, Availability of Space, Explanation of Standby Fares. From each page the User selects one entry which brings up another page with another set of entries, etc.

The tree index is unsatisfactory because:

1. Its use involves high money costs that increase with the length of a search, like a taxi meter;

2. Unlike a taxi meter, the rate at which the user is running up charges is not visible to the user;

3. It is frustrating and tedious to use;

4. Users are sometimes unable to find answers that do exist in the Prestel data base (Bush and Williams 1978), (Rigg and Sandringham 1977);

5. It fails to give an "early warning' that an answer is not to be found in the Prestel data base.

The printed subject matter index--essentially a "Yellow Pages"--is unsatisfactory because:

1. The term used and the thought structure embodied in this index are not those of ordinary consumers;

2. It is not redundant, failing to include all or even most of the words or phrases that different people use to denote the same topic;

3. There are two few cross references.

There is also a printed index of information providers, e.g., British Airways, National Westminster Bank, Central Office of Information, etc. It is also unsatisfactory because it is less important, we will not explain why here.

A good index must:

1. Maximize the probability that the user will obtain the information he wants. and

2. Minimize the probability of "wild goose chases"-the probability of undertaking a long search only to discover that there is no answer in the data base, and

3. Economize on the money, time and psychological effort required to achieve the first two goals.

The achievement of these objectives probably implies such features as redundancy in the words and phrases used to identify topics, extensive cross-listing, extensive and hierarchical use of subheadings, and a non-tree structure, unless it is computerized.

Prestel's experience should teach Americans that the development of an efficient index is a high-priority task. One possibility that merits exploration is the multiple key-word search. For example, suppose one wanted to obtain information for the following key words or phrases: restaurants, midtown New York City. medium-priced, Chinese. An index or computer program that selected all such restaurants, quickly and at low cost, would be most helpful. It seems likely that this idea could be generalized.

In considering the tree index, it should be noted that considerable research was undertaken to perfect it. Unfortunately, none of the research was directed toward ascertaining whether the tree index was (or could become, satisfactory. The defects cited above are serious enough to suggest that the tree index be either discarded or drastically revised.

The general lesson (again): test, do not assume.

The Matter of Success

From it s conception Prestel has been viewed as a "simple" system that could be used in the home by ordinary consumers operating the system themselves. However, the author's investigation uncovered a number of situations where ordinary consumers, using the system would have been thwarted, frustrated, or have conducted inefficient searches.

This raises the question of which serves the consumer better--(1) self-operation, (2) operation by an expert, or (3) some hybrid "guided access" operation in which ordinary consumers use Prestel but with expert guidance readily available.

It is obvious that self operation requires better training of users, better indexes, and a better quality data base. Only self operation (or a mainly self-reliant version of "guided operation") is compatible with home use. Expert operation adds to the cost of Prestel use since the services of the expert must be recovered.

Economies of Scale and Their Implication

For home users, Prestel embodies large economies of scale: the more answers the user obtains, the lower the cost per answer. Indeed, costs could become negative. This, in fact, occurs for subscribers of Consumer Reports or other product testing magazines when the savings that users obtain from product test information in the magazine exceed the cost of 2 subscription.

Economies of scale for home users have many implications for the developers of Prestel and similar systems. First, the more questions the system can answer, the more attractive the system will be to the user. Secondly. the more satisfactory each answer, the more attractive the system will be. Thirdly, the more components of the purchase process with which Prestel can deal--answering questions, ordering, paying, providing credit, sending messages, etc.--the more attractive the system will be.

Users of Prestel will accept or reject the system as a whole. This implies that everyone involved has an interest in everything that every other party does with respect to Prestel---British Telecom, equipment manufacturers, information providers whether public or profit-making. An example makes the point more cogently. A private retailer's excellent file on electric shavers should be regarded as a "plus" by everyone whether a consumer, government agency, British Telecom, or another private firm: it helps "sell" the user on Prestel.

Similarly, all concerned have an interest in technological extensions of Prestel such as gateways to alternative data files, alphabetic keyboards, copiers, videodiscs, compatible micro-computers, full interaction to make possible ordering, paying, financing, back-and-forth questioning.

Economies of scale have applicability to advertising, too. As long as Prestel is young and the number of questions and functions it can handle is limited, large advertising efforts will be required to induce people to use Prestel initially, and also to remind them to use Prestel when it is appropriate. As Prestel matures and provides more information and more functions more cheaply, then it will become what Sam Fedica, its inventor, terms the "obvious first port of call' (Fedica 1979). When this occurs, Prestel will become almost self-advertising. Indeed, advertising will focus exclusively on the recruitment of new users.

American developers of Prestel-like systems would be foolish f they did not recognize these implications of the economies of scale of Prestel for its development here.

SUMMARY

The research reported in this paper provides a partial and interim assessment of Prestel as an information storage and retrieval system. The paper summarizes some of the reasons why Prestel was a marketing and consumer research failure.

We should be deeply grateful to the British for their invention and investment in Prestel. Careful study of their successes, problems, and failures should enable American developers of similar systems to avoid the pitfalls that the British have encountered. [Ottobre (Ottobre 1983) in a recent Master's thesis at Cornell University developed a general model for identifying factors influencing consumers' acceptance of CompuServe's information system. In applying her model to a small, nonrepresentative sample of CompuServe trial customers, Ottobre found that "satisfaction was more important than "usage" in determining acceptance.]

REFERENCES

Bush, G. and E. Williams (1978), Viewdata: The Systematic Development and Testing of Post Office Routing Trees. Cambridge, U.W., Prestel Headquarters.

Campbell, J.A. and H.B. Thomas (1981), "The Videotex Marketplace--A Theory of Evolution," in Viewdata 81, Northwood Hills, Middlesex, United Kingdom. Online Conferences 439-453.

Fedida Sam and Rex Malik (1¦79), The Viewdata Revolution, Ne; York: John Wiley.

Hooper, Richard (1981), "Prestel--Two Years of Public Service," in Viewdata 81, Op. cit, 2-4.

Maynes, E. Scott (1982), Prestel In Use, A Consumer View, London: National Consumer Council.

Mitchell, Jeremy (1981), "Consumers and New Developments in Information Technology," in International Organization of Consumers Unions, Background Reader for the Tenth World Congress, The Hague: IOCU.

Ottobre, Angela (1983), An Analysis of Consumer Acceptance a Rejection of CompuServe Information Service, unpublished Master's Thesis, Department of Consumer Economics and Housing, Cornell University. Ithaca, N.Y.

Rigg, A. and St. John Sandringham (1977), Telewhich? Group Discussion Reports, London: Consumers Association.

Talarzyk, W. Wayne and Robert E. Widing II (1982), "Introduction to and Issues with Videotex: Implications for Marketing," working paper WP 82-1, College of Administrative Science, The Ohio State University. Columbus, Ohio.

Talarzyk, W. Wayne (1983), "Viewdata Project Reviews II ," working paper VP 83-17 College of Administrative Science. The Ohio State University, Columbus Ohio.

----------------------------------------