Ph.D thesis "The Viability of Multimedia Retrieval Systems for Marketing and Sales" by prof dr M. J. Hoogeveen   Chapter 1 Introduction

1.1. Multimedia Retrieval for Marketing and Sales

Today, the multimedia market is growing fast (Jeffcoate et al., 1993) and attracts global players from consumer electronics, information technology and telecommunication industries, like Apple, Microsoft, IBM, Philips, AT&T, Nintendo, and many others. These companies are putting considerable effort into conquering an anticipated market offering appealing, beautifully interfaced systems for business applications and home entertainment. Now, multimedia is booming, despite of all the warnings of the pessimists.

Multimedia has its roots in the home computing world in the eighties. Multimedia home computers were the first to be equipped with video and audio extensions to further improve their entertainment value. Their business value was limited to niche markets such as video editing. It was not until the offset of the nineties that the marketing potential of multimedia was widely recognised and multimedia was introduced into the world of the PC to revitalise the PC market. Since then developments have accelerated. Today, multimedia computing has become one of the major information technology trends. More and more information systems are being developed that are not only able to handle text, but also different information types like audio, stills, graphics and video. Multimedia systems are being developed for an astonishing, and still growing, number of application fields. One of the most promising application fields for multimedia, according to market research, is marketing and sales (Jeffcoate et al., 1993; Harlaar & Van Pelt, 1993). In the sphere of marketing and sales, multimedia systems are often used as gadgets to improve business presentations by sales people (Hooper Woolsey, 1991), as attractive points of information in galleries and points of sales in commercial environments (Josephson, 1991; Arbuthnot & Khalil, 1993), as sophisticated multimedia catalogues on CD-ROM or CD-i, or as sales support systems for sales employees. The value added of multimedia systems for marketing and sales is attributed, among others, to the communicatory power of moving pictures and audio, to an improvement in information and knowledge transfer using data visualisation techniques, and to the motivating effects of interactive gaming elements. To go further, it is claimed that multimedia systems help to raise sales figures and to improve the response rate to marketing actions, however, evidence for such claims is not given. The explanation for this lack of evidence is that it is probably not in the short term interest of information technology suppliers to evaluate their multimedia systems and, that evaluation requires considerable effort from the marketing and sales organisation involved.

Multimedia is also welcomed by educational researchers, who assume that congruent use of multiple information types increases learning effectiveness (Heller, 1990). Multimedia assisted instruction is used in the marketing and sales sphere to inform sales people about new product developments and to train business customers in the use of certain products or services.

Currently, an important development setting new frontiers in the world of multimedia is the evolution of multimedia systems from stand-alone systems to networked systems, which brings us in the worlds of (tele)communication and broadcasting. An extrapolation of this development is the electronic data highway (Reinhardt, 1994). This highway, it is envisioned, will give companies and citizens free access to multimedia services. Both governments and the business world have high expectations of the electronic data highway since it may offer the infrastructure for a virtual market (Hoogeveen & Van den Eijnden, 1994), which may stimulate the introduction of innumerable commercial activities, and may become the economic centre of gravity for the next century. In the sphere of marketing and sales one can see for example, the development of multimedia tele-malls which offer a subscriber the possibility of entering a virtual shop, to select products from the virtual shelves and to purchase them. Another example is the use of interactive tele-advertisement, already possible with interactive TV systems. Despite these high expectations, it must be realised that it will take many years before an electronic data highway is realised. It requires large investment in communication infrastructures to provide on and off ramps to the electronic data highway and the enthusiasm of investors is tempered by uncertainty about the viability of commercial multimedia services. Furthermore, one of the key technologies, namely Multimedia Retrieval (MR), is still under development.

MR is a key technology in a technical sense as it is at the heart of almost every multimedia system or multimedia service. Besides, it is also a key technology in a strategic and economical sense. Information and the ways information storage and access are organised are of tremendous strategic and economical value in a society which is becoming more and more an information society, in which software sales have a higher value than hardware sales, and in which copyright protected information has a relatively stable value in a world of decreasing hardware costs. This makes clear why sound MR systems (MRSs) for accessing and managing multimedia information sources are so vital.

It is not yet crystal clear what MRSs will be used to offer (on-line) access to and management of valuable multimedia information bases.

With the maturing of multimedia systems and development of multimedia services the limitations of more conventional retrieval technologies for managing and accessing databases have become evident. Therefore, Data Base Management Systems (DBMSs) (Date, 1986) which are commonly used for the retrieval of small alphanumeric records from large databases, and which form the core of most information systems, have recently been extended by the addition of facilities to handle multimedia database objects. In addition the need is felt, in some cases, to be able to include the document retrieval facilities offered by Information storage and Retrieval Systems (Salton, 1989), commonly used in archives and libraries. MR adds to all this new ways to retrieve data, for example by providing browsing mechanisms for video, by using pictorial search mechanisms or by using natural-language interfaces.

Yet, these extensions to DBMSs are sometimes not enough. In the situation where on-line multimedia services are offered to a large number of subscribers, performance becomes such a major bottleneck that usage of conventional DBMSs leads to unacceptable performance losses. Specially designed media servers with high processing power, a high input/output throughput and sophisticated compression functionality provide a possible solution for this case.

1.2. Insight into the viability

In the previous section I pointed out that marketing and sales (M&S) is one of the most promising application areas of multimedia. Next, I stated that MR is an important, fast developing, key technology at the heart of almost every multimedia system or multimedia service. A third point is made for the viability aspects of multimedia systems or services. These three points limit the scope of this thesis to the viability of Multimedia Retrieval Systems for Marketing and Sales (M&S). An MRS for M&S is a multimedia system with a clear retrieval component, that supports one, or more, M&S processes. In this context, M&S is defined as those business processes in an organisation directly related to the marketing and selling of goods and services, including support processes such as training sales staff and performing marketing research.

The fast developments within this research domain, particularly the advancements of MR technology, make it interesting to understand early experiences with developing MRSs for M&S. This will assist us to improve these systems and to improve the development of these systems. This bring us to the main research issue, viability, how viable are MRSs for M&S? The viability of MRSs for M&S are a major issue for every investor in multimedia systems for M&S, and insight into viability should guide the improvement and development of MRSs for M&S. Viability here refers to the state of being able to survive under business conditions. This implies that MRSs for M&S are technically feasible, are accepted by the market, by the users and by the principals. An example of a system that proved to be viable is the PC: they are technically feasible, and are accepted by the market. Examples of systems that did not prove to be viable are easy to find: aircraft that never left the ground, information systems that were build but never used, etc. Presumably, viability also has a time aspect: the higher the viability of an MRS, the longer the multimedia system will be able to survive under business conditions, and, once viable does not mean always viable, systems have a life cycle. The viability of MRSs for M&S appears to depend on the effectiveness or perceived effectiveness of an MRS in terms of meeting M&S business objectives. This may boil down to raising sales volumes, better promotion of product X, more effective training of sales people, or an improved corporate image. In general terms, these business objectives are related to improving the effectiveness, efficiency, and productivity of business processes, or are significantly meaningful in terms other than economical.

Given the above, the main research question of this thesis is formulated as follows:

What is the viability of Multimedia Retrieval Systems for Marketing and Sales?

Answering this question should lead to more insight into this viability, and the factors influencing the viability of MRSs for M&S. The effects of extending a system with MR facilities and the effect of this on system effectiveness in terms of meeting business objectives and subsequently viability need to be well understood. This insight may lead to useful guiding principles for the improvement and development of MRSs for M&S, in particular insight into realistic business objectives and business opportunities may prove worthwhile. Insight into the value added of MR in relation to system effectiveness is useful when considering the use of the more advanced fruits of retrieval technology, like full text retrieval and search profiles. This is important, as in current multimedia systems for M&S, these advanced retrieval facilities are often ignored by system developers. Insight into success/risk factors which strengthen or threaten the viability of multimedia systems, may also prove useful when developing MRSs, by helping to avoid costly mistakes.

1.3. Research approach

In this section the research approach is discussed, and is illustrated by the precedence chart given in figure 1. The main thread of the research process is shown as a bold line and starts with the research question followed by the development of a framework, the characterisation of current and innovative MRSs for M&S using the framework, the formulation of hypotheses and the testing of these hypotheses.

In general, several data sources are used to develop the framework, to describe and advance MRSs for M&S, to formulate hypotheses and to test hypotheses. As can be seen in figure 1, literature surveys, case research, and expert surveys are used. As a part of expert surveys, experts are confronted with demonstrators, scenario's and questionnaires to measure their responses.

With regard to case selection it is important to note that most of the cases selected are taken from work carried out at the Koninklijke PTT Nederland NV (KPN) and its subsidiary PTT Telecom BV. KPN is the principal for this research. There are two reasons for KPN's interest in the research subject discussed. First, M&S are major business processes within it's subsidiaries, for example, business sales and consumer sales within PTT Telecom. Second, KPN is a service provider, interested in offering MR services to its private and business customers. An example of a predecessor to such MR services is Videotex. Most of the Videotex applications available today have an M&S purpose, for example a wine shop selling wines via Videotex or the marketing of motorist services by a motorist association.

  Figure 1. Overview research approach

The fact that most of the cases are selected from one company, raises the problem of the generalisability of the findings, and this is somewhat reinforced by the fact that almost all of the cases are situated within The Netherlands. Thus, a cultural bias may also be introduced into the findings. I assume that the generalisability problem is softened by making use of international literature surveys, surveys explicitly including non-KPN firms, and the inclusion of some non-KPN cases. I also assume that the M&S activities of PTT Telecom, a large, international and heterogeneous company, are representative of the M&S activities of other firms in at least the western world. In other words, I assume that PTT Telecom's experiences with MRSs for M&S will not differ fundamentally from the experiences of other large companies in other (western) countries. An argument in support of this assumption is that PTT Telecom is part of a privatised company which uses the same resources as other international companies do; that is increasingly competing against other international companies in the same market places; and is becoming more and more involved in international partnerships. PTT Telecom uses the same MR technologies as other businesses in the world do, and similar ideas about organising business are accepted. An argument that weakens the generalisability assumption is that every company is unique in many respects, for example in its innovativeness, its financial resources, the size of its M&S activities, and the size of its bureaucracy which is an inhibiting factor.

What we can conclude from this is that we must be careful about generalising research results about viability aspects.

The four main research activities are discussed below, using four subquestions, derived from the main research question given in the previous section.

Development of a framework

A fixed framework is necessary to be able to characterise MRSs for M&S systematically. Such a framework consists of relevant characteristics of the research domain. Thus, to develop such a framework it is necessary to analyse what the essential characteristics of the research domain are. Hence, the first subquestion is:

1. What are the essential characteristics of MRSs for M&S?

In other words, how can MRSs for M&S be best characterised or what are the basic elements which characterise the business, technical and informational aspects of MRS for M&S? Several data sources were used to develop the framework.

• The literature was reviewed to obtain stable and representative characteristics to describe M&S, multimedia system and retrieval engine characteristics. The literature was surveyed with regard to current developments in M&S, multimedia systems (Hoogeveen, 1993a), retrieval systems (Hoogeveen & Van den Bempt, 1993) and standardisation of Multimedia DBMSs (Kwaaitaal, Hoogeveen & Van der Weide, 1994).

• Commercial Multimedia DBMSs were reviewed to get insight into the state-of-the-art technological developments with regard to the access and management of multimedia databases (De Ruyter, 1993). This shed light on the feasibility of the MR component of multimedia systems.

• In-depth case research was performed to identify practical characteristics of marketing and sales, and retrieval engines. This was a welcome addition to literature research. Case research is useful because it helps to determine the practical importance of certain theoretical aspects. PTT Telecom served as a case for the M&S characteristics. Participant observation in the OCTOPUS case proved to be useful in particular with regard to retrieval engine characteristics (Hoogeveen, Van der Meer & Sol, 1992b). The OCTOPUS experience gave an idea of which retrieval facilities are the most common, which facilities are used regularly by expert users, and which are only used in very special situations. Participant observation of several multimedia projects within PTT Research proved useful to obtain insight into multimedia system characteristics.

Characterisation of current and innovative MRSs for M&S

Before formulating ideas about the viability of MRSs for M&S, it is first necessary to characterise these systems; there is little value in reviewing nonviable system artefacts. Thus, the second subquestion is:

2. What potentially viable MRSs for M&S can we distinguish on the basis of practical examples or as extrapolations of developments in M&S, multimedia systems and retrieval engines?

The framework was used to systematise the characterisations of the systems, addressing both the technical and the business characteristics. The systematisation of system characterisations is necessary to reach an acceptable level of completeness, to make it more easy to compare systems with each other, and to help in identifying ways to improve or extend the systems. The value of the MRSs for M&S presented is that they can be (re)used as basic models for MRSs for M&S.

In subquestion 2, two ways to characterise the systems are mentioned, namely on the basis of practical examples and as extrapolations of developments in the research domain.

• The practical examples or cases were almost all, selected from within KPN. The use of practical examples guarantees that the systems based on them are at least feasible. The practical examples give an idea about related business objectives and opportunities, and success/risk factors. A limitation of practical examples is that they are often not the most innovative.

• Extrapolations of developments are reviewed on the basis of a survey of multimedia projects and are reflected in the framework for MRSs for M&S. The feasibility of MRSs for M&S based on extrapolations are demonstrated by experimental prototypes, several of which were developed within the context of my Ph.D. research project within PTT Research.

Formulation of hypotheses

Now, we get onto one of the key issues: the formulation of hypotheses about the viability, and the variables influencing the viability, of MRSs for M&S on the basis of early experiences with some of these systems. Subquestion 3 can be formulated as follows:

3. What hypotheses can be formulated about the viability of MRSs for M&S?

The hypotheses should make manifest the ideas, convictions, and experiences that MR developers share with regard to the (variables influencing the) viability of MRSs for M&S. Such a set of related hypotheses forms a theory. The term 'theory' is used here in the sense of grounded theory (Glaser & Strauss, 1967), i.e., a related set of ideas or hypotheses induced by empirical research data. The term 'grounded' stresses the point that the theory is closely related to a limited empirical research domain, in this case the viability of MRSs for M&S.

The formulation of hypotheses on the basis of early experiences with developing MRSs for M&S has three advantages. As said before, it makes clear what are the essential experiences with, and basic convictions about, MRSs for M&S. This may help us to understand somewhat better why and when MR systems are (believed to be) effective for M&S. For this reason it is important to explain the value added of MR. The second advantage of hypotheses formulation is that these hypotheses about the viability, and variables influencing the viability, of MRSs of M&S become testable. A third advantage is that the set of hypotheses, insofar confirmed, can be used cautiously, as guiding-principles when developing MRSs for M&S. Therefore, the hypotheses should make explicit typical business opportunities and typical success/risk factors for MRSs for M&S.

Hypotheses were formulated on the basis of:

• data from case research, namely early experiences with developing MRSs for M&S purposes;

• hypotheses and descriptions of experiments described in the literature;

• data from a survey of business opportunities and objectives, and success/risk factors for multimedia projects and multimedia systems.

Testing hypotheses

Hypotheses about the viability, and variables influencing the viability, of MRS for M&S must be operationalised and tested to draw conclusions about their validity. Subquestion 4 was, therefore, formulated as follows:

4. What support can be found for hypotheses about the viability of MRSs for M&S?

As true experimental conditions (Kidder, 1981) are not possible when studying in vivo business phenomenons it was necessary to adopt quasi-experimental methods and to look for convergent evidence, i.e., to look if evidence from different data sources pointed in the same direction. It is very important to note that the viability of MRSs is most often assessed retrospectively, but that the history of MRSs is so short that retrospective analysis is only possible for the first generation of MRSs. In several cases, the systems have to show their true worth yet. Despite these limitations there are ways to approach the viability problem, although the reliability and validity of the outcomes will not be 100%.

• Experts panels were used to validate hypotheses about developments within the experts fields of expertise, e.g., to test hypotheses about the respondent's perceptions of viability and related variables, like the value added of multimedia, and specific success/risk factors. An assumption underlying the use of an expert panel approach is that experts come to better judgements about developments within their fields of expertise than laymen. This assumption can, however, be disputed. A good example to the contrary is that most Eastern Europe experts, like everyone else, did not foresee the sudden collapse of the communist regimes. Another major point is: are the developments fundamentally unpredictable or not? I assume that developments within the research domain are predictable to some degree, and that in most cases experts predict better than laymen or that experts can better found their judgements than laymen.

• Several MRSs for M&S were evaluated. Cost Benefit Analysis was performed and Return On Investment was estimated to make a reasonable case for the economic viability of some of the implemented MRSs. The limitations of IT (information technology) evaluation methods such as Cost Benefit Analysis and Return On Investment are discussed by Farbey et al. (1992), who give an overview and classification of available IT evaluation methods. It can be concluded from their work that no completely satisfactory method exists to evaluate both quantitative and qualitative benefits, due to uncertainties and difficulties with quantifying, in monetary terms, certain qualitative benefits like 'improved image' or to give valid forecasts of quantitative future benefits like 'expected extra revenues'. Likewise, De Lone and McLean (1992) observe in their review of dependent variables for Information Systems Success that quantitative Cost Benefit Analyses are often found lacking due to the difficulty of quantifying 'intangible benefits'. Despite these limitations, I believe that the outcome of a Cost Benefit Analysis is a valuable indication of the viability of a multimedia system.

• Market research is reviewed to look for convergent evidence with regard to viability, and to obtain a concurrent validity estimate for some expert panel estimates.

1.4. Outline of the thesis

The outline of this thesis follows the activity structure and order of subquestions presented in the previous section. In this chapter the research subject, the research question, and the research approach followed are introduced and explained.

In chapter 2 the M&S business aspects, multimedia system characteristics and retrieval subsystem characteristics that are stable enough to systematically characterise MRSs for M&S are analysed. These characteristics taken together form the framework. With regard to the M&S business level, a typology is proposed, formulated in generic terms but discussed in M&S specific terms. At the system level the multimedia processing elements and information types involved are discussed. At the subsystem level the type of retrieval engine and the essential retrieval facilities are discussed. Finally, the characteristics of the framework are summarised visually.

The framework presented in chapter 2 is used in chapter 3 to characterise potentially viable MRSs for M&S. The MRSs range from systems based on case research, such as a Marketing Communication Archive, a Tele Sales Assistant system, a Multimedia Business Catalogue for business sales assistance, a Multimedia Promotion System at a trade fair, and Multimedia Assisted Instruction for self-help training on the job, to innovative systems based on extrapolations of current technological developments, such as a Marketing Documentation Archive and the Virtual Market. To conclude chapter 3, an overall system is presented to put all the MRSs together and to show the interconnections between these systems, working together in one imaginary M&S organisation. Next, the MRS related business objectives and noticed success/risk factors for developing MRSs are summarised.

This information about business objectives and success/risk factors is used as input for the formulation of hypotheses in chapter 4. Chapter 4 presents hypotheses about the viability, and the variables influencing the viability, of MRSs for M&S. One of these variables is the value added of MR technology. This value added is delineated and explained by discussing experimental research findings described in literature. Other variables are the success/risk factors mentioned before. Hypothesised success/risk factors are listed and discussed on the basis of a survey of multimedia projects. These success/risk factors are grouped into project management success/risk factors and multimedia system success/risk factors. The business objectives and the success/risk factors are related to the MRSs presented in chapter 3 to give insight into MRS related differences. Chapter 4 ends with a discussion about the operationalisation and measurement of the viability of MRSs for M&S and a summary of testable hypotheses.

The selected hypotheses are tested in chapter 5, which has two parts. The first part contains the results of expert assessments. The second part contains two Cost Benefit Analyses and computations of Return On Investment. The first part starts with the results of a survey of multimedia projects on the value added of multimedia and the relevance of success/risk factors. Then we proceed with a general expert assessment of the viability of seven clusters of multimedia teleservices. Next, the results of two more specific evaluations by an M&S panel are given for a Multimedia Business Catalogue and a promotional system. Thereafter, the results of an interesting survey of potential investors in multimedia teleshop services are given. The second part of this chapter contains a retrospective Cost Benefit Analysis and ROI computation of the IECT photo archive, and the results of a prospective Cost Benefit Analysis and Return On Investment computation of an MRS for tele-ordering by Top 1000 accounts are presented.

Chapter 6 returns to the main research question presented in the first chapter. An answer is given on the main research question with reference to the developments noted in the research domain in chapter 2, early experiences with developing MRSs for M&S presented in chapter 3, the hypotheses formulated in chapter 4 and the test results presented in chapter 5. Next, the consequences of the findings for the further improvement of the MRSs, and the consequences for the development of these systems are discussed. The discussion about the generalisability of the research findings is important. Further, conclusions are drawn with regard to the research approach: what is the usability of the framework presented, the theory presented, and the research methods used. Chapter 6 is concluded by looking forward and formulating recommendations for further research on the viability of MRSs for M&S.

© 1995-2002 Martijn Hoogeveen