=Paper=
{{Paper
|id=None
|storemode=property
|title=Decision-Support for Service Bundling
|pdfUrl=https://ceur-ws.org/Vol-593/Paper04.pdf
|volume=Vol-593
}}
==Decision-Support for Service Bundling==
https://ceur-ws.org/Vol-593/Paper04.pdf
Decision-Support for Service Bundling
Thomas Kohlborn
Faculty of Science and Technology, Queensland University of Technology,
126 Margaret Street, 4000 Brisbane, Australia
t.kohlborn@qut.edu.au
Abstract. Offering service bundles to the market is a promising option for
service providers to strengthen their competitive advantages, cope with
dynamic market conditions and deal with heterogeneous consumer demand.
Although the expected positive effects of bundling strategies and pricing
considerations for bundles are covered well by the available literature, limited
guidance can be found regarding the identification of potential bundle
candidates and the actual process of bundling. The proposed research aims at
filling this gap by offering a service bundling method complemented by a
proof-of-concept prototype, which extends the existing knowledge base in the
multidisciplinary research area of Information Systems and Service Science as
well as providing an organisation with a structured approach for bundling
services.
Keywords: Service, service-orientation, bundling
1 Introduction
The interest in service-orientation has increased over the last years due to new
technological developments [1] and novel approaches for organizational management
[2] since services have become focal units for the cost-effective creation of customer
value and innovation. The multidisciplinary nature of service-oriented concepts has
led to the emergence of Service Science as a new academic discipline [3]. Business
Service Management (BSM) can be positioned as the business discipline within
Service Science dedicated to the holistic management of services in an organization to
ensure alignment between the needs of the customer and the objectives of the
organization [4].
Business Service Management is a research project as part of the Smart Services
Cooperative Research Centre (CRC) research initiative.1 An essential component of
the strategic side of BSM is Service Portfolio Management (SPM) [5], which is one of
the work packages within the BSM project and is led by Prof. Michael Rosemann,
who is also the principle supervisor of the candidate’s research. 2
1 Please refer to http://smartservicescrc.com.au
2 Associate Supervisors are Dr. Axel Korthaus and Dr. Erwin Fielt. Both are postdoctoral
research fellows at the Queensland University of Technology.
In focus of SPM is the service portfolio, which comprises a well-defined set of
services. Offering innovative service bundles to the market is a promising option for
service providers to strengthen their competitive advantages, cope with dynamic
market conditions and heterogeneous consumer demand [e.g. 6]. Service bundles are
composed of at least two services that can be integrated to a certain extend to create a
new packaged service offering. Therefore, a key task within SPM is service bundling
that deals with the challenge of identifying services within the portfolio that can and
should be used to create service [5]. A demanding managerial task exists in
comprehending potential service candidates offered internally and externally, and
being able to identify service bundles that lead to new efficient and strategically-
aligned packages.
Literature lacks approaches that facilitate the creation of adequate service bundles.
Despite the fact that companies across all industry sectors with increased market
pressures are challenged by the issue of service bundling [7], only little guidance has
been provided so far for the identification of potential bundle candidates and for the
actual process of bundling to answer the questions: “How can services be identified
that should be bundled?” and “how can the act of service bundling be effectively and
efficiently supported?”.
The research related to the candidate’s studies tries to answer these questions by
developing a service bundling method. In particular, it will focus on the perspective of
a service provider. Existing methods for service bundling usually use a given
customer demand to drive the creation of service bundles [7-9]. While these methods
are useful for situations where customer demand is well known and understood, poor
performance can be expected when demand is hard to capture or anticipate. As
customer-driven service bundles typically relate to an outside-in perspective, the
induction of new, innovative bundles to the market to trigger demand is not yet
sufficiently covered. The proposed method will fill this gap and provide an inside-out
perspective on service bundling. The main contribution of the research is therefore a
structured guideline to facilitate the composition of bundles in practice. Designed as
an innovative artefact it extends the knowledge base of service management, while
facilitating a multi-disciplinary approach honouring the importance of business and IT
alignment. As a proof of concept, the method will be complemented by a software
prototype.
The remainder of this paper is structured as follows. Based on the problem
description that has been provided in this section, we first present related work and
alternative approaches in this area of research, before the foundations of our approach
will be detailed further. Subsequently, a detailed research design that aims at
providing a structured guideline to answer the stated research questions will be
described. Finally, the current status of the research is presented before the paper ends
with a conclusion and directions for further research.
2 Related Work and Alternative Approaches
The objective of this research is to provide a service bundling method. A review of
the academic knowledge base yields various possible approaches that can be utilized
to identify service bundles as pointed out in [10].
For example, the area of artificial intelligence (AI) research offers techniques that
can potentially support the design of solutions to the service bundle identification
problem. Particularly, machine learning solutions are conceivable that can “learn”
from existing successful service bundles to identify or propose new service bundles
[e.g. 11]. A general problem of machine learning is that it usually does not yield
absolute guarantees of the performance of algorithms. Moreover, in spite of many
successes, AI research in general has been the target of fundamental criticism [e.g.
12]. To the best of the authors’ knowledge, comprehensive AI approaches to identify
and analyze new service bundles are not existent in the academic knowledge base.
Business Intelligence (BI) employs systems that “combine data gathering, data
storage, and knowledge management with analytical tools to present complex and
competitive information to planners and decision makers” [13]. Hence BI is used to
analyze existing data to support future decisions. Within BI, the area of Association
Rule Mining can be employed to identify bundle candidates. This mining approach
analyses basket data type transactions, for example receipts from a supermarket, to
identify items that are frequently bought together within one transaction [14]. The
identified so called frequent item sets are used for recommender systems to offer
customers related products, hence enabling cross-selling potentials (e.g.
Amazon.com).
The ideas of Semantic Web approaches can be utilized as well. These approaches
generally require three sorts of machine-understandable information: “ontologies to
define vocabulary, data about observations of the world, and theories that make
predictions on such data” [15]. An ontology specifies an explicit, simplified view of
the world [16]. Using these artifacts, it is possible to model the relations between
services and reason about them. This has been done for web services in the Web
Service Modeling Ontology [17] and for real-world services using the OBELIX
(Ontology-Based Electronic Integration of Complex Products and Value Chains)
service ontology [7].
A different semantic approach utilizes Latent Semantic Indexing (LSI), also called
Latent Semantic Analysis. Latent semantic indexing is an “information retrieval
technique based on the spectral analysis of the term-document matrix” [18]. This is
done through the creation of vector spaces using a mathematical technique called
singular value decomposition. The created vector spaces can be queried for the
semantic distance (usually expressed as a vector) between two terms or service
descriptions, and the so found semantic relationships can be utilized to identify
clusters or bundles of services. LSI requires no formal ontology that classifies the
different elements of a service [19] and uses unstructured text documents as the only
source of information. Several limitations to LSI restrict its usage, the main drawback
being that the underlying vector space model “is unable to convey any relationship
[...] existing between the terms” [20].
All of the introduced identification approaches can be expected to provide
interesting results for a bundling method. As motivated in the next section, the
proposed method is based on service descriptions. Thus, a semantic approach has
been chosen. Instead of employing a formal ontology, the notion of relationships is
used to discover semantics between services. This ensures the simplicity of the
method and its utility across a range of different service descriptions.
The single work that specifically targets service bundling is from Baida (please
refer to [8]). The author used an ontology-based approach “to facilitate the
automation of the service bundling task”. The created ontology includes the notion of
resources as prerequisites or outcomes of service elements and so-called functions and
relationships that define dependencies between two service elements (e.g. enhancing,
excluding, and substituting). Using a given customer demand by expressing required
resources, the method can create service bundles that satisfy the demand and adhere
to the predefined set of dependencies between services. The author named his
approach ”Serviguration” to express his view of service bundling as a configuration
task. A detailed discussion of the differences and distinguishing characteristics of the
“Serviguration” approach and the proposed approach will follow in the next section.
3 Conceptual Framework for a Service Bundling Method
Preliminary Considerations
The purpose of the proposed method is the identification of possible service
bundles. It is designed to support the process of bundle creation in its early stages.
The method therefore focuses on limiting the solution space of possible bundles,
using indicators that express some form of bundling motivation as described in [10].
It is important to point out that this method is not supposed to omit the evaluation
of bundles by a domain expert. It has to be acknowledged that the domain expert is
still needed to evaluate the overall feasibility of bundles, since this requires complex
analysis, often utilizing tacit knowledge across a range of different disciplines (e.g.
economy, marketing, legal). Rather, the aim of this method is to limit the scope of the
necessary evaluation for the domain expert.
This approach is particularly useful when a large number of services are available,
which is a common scenario particularly in business networks or service ecosystems
[21]. Since a human expert would be overwhelmed by the task, the proposed method
could be applied in an automated way using a corresponding support tool in order to
quickly constrain the solution space of possible bundles. Consequently, the domain
expert can focus on evaluating only the short-listed bundles that somehow indicate a
bundling opportunity.
[8] relies on a given customer demand to drive the creation of service bundles.
While this approach can be useful for situations where customer demand is well
known and understood, poor performance can be expected when demand is hard to
capture or anticipate. Furthermore, the economically desirable situation where
customer demand is induced by a new service offering is not supported at all. Our
proposed method explicitly targets the latter case by focusing on the creation of new
and innovative service bundles. Therefore, customer demand is not utilized to reason
about the suitability of potential bundles in this method. Instead, the driving source of
this method is a repository of services that are available for bundling. Depending on
the given context, this repository might consist of the services of a single provider, a
provider network or even contain all available services in a service ecosystem.
The bundling method created by [8] identified six distinct relationships that define
dependencies between two services: core/enhancing, core/supporting, bundled,
substitute, excluding, optional bundle. The (manual) evaluation of all services
regarding these relationships is a prerequisite for the actual bundling process, as the
feasibility of bundles is determined by the existing relationships. This evaluation is a
time consuming task, one that becomes practically impossible to handle for a large set
of services.
The proposed approach is therefore based on a service description which does not
necessitate the step of explicating relationships between services. Instead, this method
uses commonalities of attributes that indicate such a relationship. As long as services
are consistently described and attributes relevant for this bundling approach are
present, the proposed method can be employed. The following section will explain the
term relationship as used in this method.
Leveraging Relationships between Services
[22] found that functionally complementary components in a bundle lead to high
intentions to purchase compared to bundles in which no complementary components
are present. The authors state that, “as the relationship among the components
increased from ”not at all related” through ”somewhat related” to ”very related”,
intention to purchase also increased”.
This method builds upon these findings and the conjecture that other
commonalities or relationships between services can also indicate potentially useful
bundles. For this method the term relationship is defined as a connection, whose
existence can be evaluated by a logic expression. A relationship builds upon attributes
from services’ descriptions. Every relationship refers to previously specified attributes
(e.g. location of a hotel, destination of a flight) and evaluates them using a given logic
(e.g. distance between destination airport and location of the hotel). This evaluation
can be realized ranging from simple value comparisons of single attributes to complex
algorithms using multiple attributes. Figure 1 illustrates the coherence between the
mentioned terms using UML.
Figure 1: Concept of a Relationship
Relationships can display varying degrees of strength. For example, the distance
between arrival airport and hotel determines the strength of this relationship. It
depends on the concrete scenario and type of involved services as to whether a certain
distance translates into a strong or weak relationship. Therefore the domain expert can
configure the logic, where this is applicable. A first set of empirically derived
relationships are presented in [10].
4 Detailing the Research Methodology
The overall research can be positioned in the area of Information Systems and
Service Science research focusing on design science [23], as the service bundling
method can be regarded as an innovative artifact to solve a contemporary problem. As
a proof of concept, the method will also be supported by a software prototype, which
represents another artifact. The research will be classified according to the design
science approach with a strong action research flavor. Within this approach, different
methods will be applied to satisfy the relevance and rigor criteria and align with the
guidelines postulated by [24]. The research methodology can be divided into five
stages, which will be shortly described in the following.
During the preparation stage a thorough literature review will be conducted to
synthesize existing knowledge and position the proposed research within the overall
body of knowledge. Thus, initial, preliminary answers will be found for research
questions. Hereby, the literature review will provide initial answers regarding the
nature and characteristics of services/SPM, the different types of bundling, the
existing service description languages and service bundling/development
methodologies. These insights will be used to articulate a preliminary draft of a
service bundling method.
The second stage, analysis stage, aims at deriving empirical insights into the
foundations of bundling by conducting a content analysis of existing bundles and
services. The scope of the analysis will include traditional bundles found in the
business domain as well as novel, integrative bundles found in the IT domain. The
outcome of this stage will be a set of factors that potentially contribute to the
composition and in particular support reasoning about the feasibility of bundles. The
content analysis will comprise the development of a questionnaire to identify initial
relationships between services in a bundle that can be used to reason about the
suitability of service bundles. Rigorous design science research must also take into
considerations existing theories and frameworks. As part of this stage, existing
theories will be analyzed to support a theory-driven design of the bundling method.
Once the related literature review and content/theory analysis have been
conducted, it is envisaged to conduct at least one case study to complement the
theoretical findings with empirical data about the current and desired situation as part
of the exploration stage. The exploratory case studies are of an observable nature. The
main objective is to find out about current SPM practices. As such we want to analyze
the current way of describing services, practices of bundling services and portfolio
management as well as requirements for a decision-support tool and method for
service bundling as part of SPM. The requirements will be used at a later stage to
design and validate the main artifact.
The fourth stage, design stage, is focused on designing the service bundling
method (and its tool-support). Hence, this stage can be regarded as the core of design
science research [23-25]. Once the final set of dimensions and requirements have
been empirically identified, a gap analysis needs to be conducted to analyze the
deficiencies of existing frameworks/methodologies and service description languages.
The gap analysis will provide important insights of possible extensions of existing
work as well as additional requirements for the service bundling method, which will
be developed in this stage. The design phase of the artifact will be nourished by the
results of the previous stages. Hence, insights gained through the literature review
will be incorporated as well as insights gained through the case studies. Furthermore,
empirical and theoretical insights gained in stage two as well as the requirements
gathered in stage three will provide the basis for the specification of scope and
features of the method and tool. The outcome of this stage is a tool-supported service
bundling method that satisfies the requirements of the case study partners (relevance)
and has been built upon existing knowledge and findings (rigor).
The evaluation stage focuses on evaluating the developed bundling method.
According to the guidelines postulated by [24] and the requirements of design science
by [25], an artifact needs to prove its utility and validity in a real world scenario. As
the developed artifact of this study is a tool-supported service bundling method, a
valid research method that aligns with the objectives of the study and the interest of
the candidate is action research. Hereby, existing methods and expertise will be
applied from the knowledge base to evaluate the tool-supported bundling method. The
final outcome of this stage will be an evaluated tool-supported service bundling
method.
5 Current Status
As pointed out previously, the first stage comprises activities to arrive at a
synthesized view on the existing knowledge base related to services / SPM, different
types of service bundling and a first draft of the bundling method. The results of this
stage have been published in [4, 5] and [26] respectively.
In regard to stage two, we derived a first working set of generic relationships based
on the proposed content analysis and subsequent empirical studies. Results have been
synthesized and are published in [10]. An analysis of potentially applicable theories is
still ongoing.
As pointed out in the last section, we aim at conducting at least one exploratory
case study to gain empirical insights into SPM and bundling practices. As part of this
stage, we were able to commence a comprehensive case study with a government
authority. Hereby, it is envisaged to bundle services that can be accessed by potential
consumers, namely public citizens, by utilizing the online channel. Hence, our
research in that area touches upon current advances in the area of e-government as it
aims at identifying ways to enhance the consumer satisfaction through the
specification of service bundles that are presented on the government’s webpage. The
bundles comprise services that are offered by multiple departments to ease access and
enhance the overall consumer satisfaction. By doing this, consumers no longer need
to consult multiple departments and their respective websites as all services of
potential interest are comprised within specific bundles. Furthermore, citizens do not
need to know the internal structure of governments anymore in order to find their
services of interest.
As part of this case study we will be able to accompany this specific government
on its way to implement a customer-centric one-stop portal that comprises the
presentation of service bundles. Currently we are observing market studies that aim at
analyzing the potential benefits of offering services in bundles as part of a one-stop
portal strategy. Furthermore, we are conducting several interviews with multiple other
governments that already implemented such a one-stop portal based on service
bundling activities in order to gain additional empirical insights. Finally, we will take
part in all activities that are directly conducted by our case study partner in order to
achieve their objectives successfully, such as conducting usability testing sessions
with a representative sample of the targeted user group. This case study provides us
with an unique opportunity to gather empirical insights into the bundling process and
helps us to answer our research questions. First findings look very promising and will
be in a more advanced stage in the near future.
Regarding stages four and five, we first need to successfully finish the other stages
before we can actively target the activities in these stages.
6 Conclusion
This paper describes a research project proposing a novel approach to identifying
service bundle candidates. Because of its potential to combine innovation with cost-
effective re-use of existing services, we envision that service bundling will become as
important as new service development as, for example, can be seen in the growing
attention for mash-ups. However, while the process of new service development has
been extensively researched and conceptualized, the process of finding suitable
service bundling candidates is still ill-defined.
The described research project proposes a method that facilitates the creation of
bundles by providing organizations with systematic and practical guidelines. The
method is a contribution to design science research in the field of Information
Systems and Service Science. It represents an innovative artifact that extends the
academic knowledge base related to service management. The developed method
builds on service bundling concepts from both the marketing and the technological
literature, thereby addressing the increased need for business-IT alignment. As such,
it also is an example of a multi-disciplinary approach that builds on existing research
in different areas and extends this research in new directions.
Based on the descriptions and explanations in the previous sections, multiple
directions for further research can be identified. First, the “service bundling” task
needs to be positioned as part of a management discipline. First insights suggest to
position service bundling as a key task of service portfolio management, but further
research needs to be conducted. Second, research in the area of service descriptions
has to be conducted to develop a universal service description language that is
applicable across industries and covers business as well as software services.
Alternatively, extant service description languages need to be analyzed to determine
in how far they accommodate the identified relationships and provide possibilities to
be extended. Third, strategies and rationales of service bundling need to be analyzed
further, to provide valuable insights for the internal and external validation of initially
identified bundles.
Acknowledgements: This research was carried out as part of the activities of, and
funded by, the Smart Services Cooperative Research Centre (CRC) through the
Australian Government’s CRC Programme (Department of Innovation, Industry,
Science and Research).
References
1. Papazoglou, M.P.: Web Services: Principles and Technology. Prentice Hall, Edinburgh
Gate, Harlow (2008)
2. Schroth, C.: The Service-Oriented Enterprise. Journal of Enterprise Architecture 3, 73--80
(2007)
3. Chesbrough, H., Spohrer, J.: A Research Manifesto for Services Science. Communications
of the ACM 49, 35--40 (2006)
4. Rosemann, M., Fielt, E., Kohlborn, T., Korthaus, A.: Business Service Management In:
Smart Services CRC White Paper Series pp. 1--14. Faculty of Science and Technology,
Queensland University of Technology, Brisbane, Australia (2009)
5. Kohlborn, T., Fielt, E., Korthaus, A., Rosemann, M.: Towards a Service Portfolio
Management Framework. In: Proceedings of the 20th Australasian Conference on
Information Systems. Melbourne, Australia (2009)
6. Lawless, M.W.: Commodity Bundling for Competitive Advantage: Strategic Implications.
Journal of Management Studies 28, 267--280 (1991)
7. Akkermans, H., Baida, Z., Gordijn, J., Peiia, N., Altuna, A., Laresgoiti, I.: Value Webs:
Using Ontologies to Bundle Real-World Services. IEEE Int. Systems 19, 57--66 (2004)
8. Baida, Z.: Software-Aided Service Bundling. Intelligent Methods & Tools for Graphical
Service Modeling In: Dutch Graduate School for Information and Knowledge Systems.
Vrije Universiteit, Amsterdam (2006)
9. Gordijn, J., de Kinderen, S., Wieringa, R.: Value-Driven Service Matching. In: Proceedings
of the 16th IEEE International Requirements Conference, pp. 67--70. IEEE Computer
Society Washington, DC, USA (2008)
10.Kohlborn, T., Luebeck, C., Korthaus, A., Fielt, E., Rosemann, M., Riedl, C., Krcmar, H.:
How Relationships Can Be Utilized for Service Bundling. In: Proceedings of the Americas
Conference on Information Systems. Lima, Peru (2010)
11.Russell, S., Norvig, P.: Artificial Intelligence: A Modern Approach. 2 edn. Prentice Hall, NJ
(2002)
12.Dreyfus, H.L.: What Computers Can't Do: The Limits of Artificial Intelligence. 2 edn.
Harper and Row, New York (1979)
13.Negash, S., Gray, P.: Business Intelligence. In: Proceedings of the Americas Conference on
Information Systems. pp. 3190--3199 (2003)
14.Agrawal, R., Imielinski, T., Swami, A.: Mining Association Rules between Sets of Items in
Large Databases. In: Proceedings of the ACM SIGMOD Conference on Management of
Data. pp. 207--216 (1993)
15.Poole, D., Smyth, C., Sharma, R.: Semantic Science: Ontologies, Data and Probabilistic
Theories. In: da Costa, P.C.G., d’Amato, C., Fanizzi, N., Laskey, K.B., J., L.K.,
Lukasiewicz, T., Nickles, M., Pool, M. (eds.): Uncertainty Reasoning for the Semantic Web
I. Lnai/Lncs, Vol. 5327/2008. pp. 26--40 Springer, Berlin, Heidelberg (2008)
16.Gruber, T.R.: A Translation Approach to Portable Ontology Specifications. Knowledge
Acquisition 5, 199--220 (1993)
17.Roman, D., Keller, U., Lausen, H., de Bruijn, J., Lara, R., Stollberg, M., Polleres, A., Feier,
C., Bussler, C., Fensel, D.: Web Service Modeling Ontology. Applied Ontology 1, 77--106
(2005)
18.Papadimitrou, C.H., Raghavan, P., Tamaki, H., Vempala, S.: Latent Semantic Indexing: A
Probabilistic Analysis. Journal of Computer and System Sciences 61, 217--235 (2000)
19.Prokopp, C.: An Experimental Approach to Unsupervised Semantic Information
Categorization by Selective Clustering and Tessellation of Associations. University of
Queensland, Brisbane (2006)
20.Chu, H.: Information Representation and Retrieval in the Digital Age. Information Today,
Medford, NJ (2007)
21.Riedl, C., Boehmann, T., Rosemann, M., Krcmar, H.: Quality Management in Service
Ecosystems. Information Systems and E-Business Management 7, 199--221 (2009)
22.Herrmann, A., Huber, F., Coulter, R.H.: Product and Service Bundling Decisions and Their
Effects on Purchase Intention. In: Fuerderer, R., Herrmann, A., Wuebker, G. (eds.): Optimal
Bundling: Marketing Strategies for Improving Economic Performance. pp. 253--268
Springer (1999)
23.Hevner, A.R.: A Three Cycle View of Design Science Research. Scandinavian Journal of
Information Systems 19, 87--92 (2007)
24.Hevner, A.R., March, S.T., Park, J., Ram, S.: Design Science in Information Systems
Research. MIS Quarterly 28, 75--105 (2004)
25.March, S.T., Smith, D.: Design and Natural Science Research on Information Technology.
Decision Support Systems 15, 251--266 (1995)
26.Kohlborn, T., Luebeck, C., Korthaus, A., Fielt, E., Rosemann, M., Riedl, C., Krcmar, H.:
Conceptualizing a Bottom-up Approach to Service Bundling. In: Proceedings of the 22nd
International Conference on Advanced Information Systems Engineering (CAiSE'10).
Hammamet, Tunisia (2010)