A framework for analysing the usage of mobile services Timo Smura, Antero Kivi and Juuso Toyli�� Abstract Purpose ��� Collecting and analysing data on mobile service usage is increasingly complex as usage diverges between different types of devices and networks. The purpose of this paper is to suggest and apply a holistic framework that helps in designing mobile service usage research as well as in communicating, positioning, and comparing research results. Design/methodology/approach ��� The framework was constructed based on longitudinal and cross-sectional mobile service usage measurements carried out in Finland annually in 2005-2008, covering 80-90 percent of all mobile users and service usage. Broad use of multiple data collection methods and measurement points enabled data and method triangulation, as well as analysis and comparison of their scopes and limitations. Findings ��� The paper suggests a holistic framework for analysing mobile services, relying on service science approach. For measurements and analysis, mobile services are decomposed into four technical components: devices, applications, networks, and content. The paper further presents classifications for each component and discusses their relationships with possible measurement points. The framework is applied to mobile browsing usage studies. Research limitations/implications ��� Future work includes adding an actors dimension to the framework in order to analyse their roles in the value networks providing mobile services. Extending the framework to Internet services more generally is also possible. Originality/value ��� The paper presents an original, broadly applicable framework for designing mobile service usage research, and communicating, positioning, and comparing research results. The framework helps academics and practitioners to design and to recognise the limitations of mobile service usage studies, and to avoid misinterpretations based on insufficient data. Keywords Mobile communications systems, Internet, Finland Paper type Research paper 1. Introduction The internet is of growing importance for businesses, enabling a virtual supply chain for marketing and delivering products and services, as well as in facilitating the production of digital services and content. Using internet as a distribution and marketing channel is widely expected to increase companies��� performance (Geyskens et al., 2002) as well as consumer surplus (Brynjolfsson et al., 2003) and it is increasingly important for competitiveness in most industries and markets today. Advances in mobile technologies are promising further benefits by decreasing the spatial and temporal constraints of service provisioning and use (Balasubramanian et al., 2002). In most developed countries, mobile phones have become an inseparable part of everyday life and a majority of people carry them all the time. In addition to complementing and expanding the use of various internet-based services, the evolution of mobile devices also enables entirely new types of services to be introduced, utilising, e.g. location and context-specific information and higher degree of personalization of the devices. Furthermore, aside from their role as a channel to services, mobile devices are also DOI 10.1108/14636690910970973 VOL. 11 NO. 4 2009, pp. 53-67, Q Emerald Group Publishing Limited, ISSN 1463-6697 j info j PAGE 53 Timo Smura and Antero Kivi are both Research Scientists based at TKK Helsinki University of Technology, Espoo, Finland. Juuso Toyli �� is a Professor based at the Turku School of Economics, Turku, Finland. Received 2 October 2008 Revised 6 March 2009 Accepted 10 March 2009

utilised for various other purposes, including, e.g. advertising (Leppaniemi �� and Karjaluoto, 2005) as well as collecting behavioural data from the users (Gonzalez �� et al., 2008 Eagle and Pentland, 2006). Because of the rapid pace of development, the need for measuring and analysing how people adopt and use the existing and emerging mobile services is evident, to support different stakeholders, including service providers, content creators, mobile operators, handset manufacturers, and regulators in their decision-making. At the same time, however, collecting and analysing the required information is becoming increasingly complex as usage diverges between different types of devices and networks. Statistics collected and disseminated by companies, policy-makers, consultants, and academics are often narrow-focused and miss a holistic view on service usage. Comparing the findings is difficult, and even the use of key terms such as ������mobile������ and ������service������ is often ambiguous and context-dependent. Mobile services are typically (and often implicitly) understood as services that make use of mobile devices and/or mobile networks. In reality, however, it is often difficult to draw a line between mobile and non-mobile networks, as, e.g. wireless local area networks (WLANs) offer more limited mobility compared to cellular networks. Furthermore, mobile networks are increasingly utilised by devices other than mobile phones, including less mobile devices such as laptop PCs or even home alarm systems. In the future of ������ubiquitous������ technologies and computing, the variety of wirelessly connected devices as well as the importance of machine-to-machine communications is expected to grow significantly, adding to this complexity (see, e.g. Uusitalo, 2006). Additional ambiguity is related to the term service having many incompatible definitions. This paper relies on a service science view. In general, Vargo and Lusch (2004) argue for a service-dominant logic to replace the traditional goods-dominant logic in marketing, and define service as the application of resources (competences, skills, and knowledge) for the benefit of another entity or the entity itself. Furthermore, service systems are defined as value-co-creation configurations of people, technology, other internal and external service systems, and shared information (Maglio and Spohrer, 2007 Spohrer et al., 2007). According to these definitions, service systems include businesses, government agencies as well as individual people, and both the client and provider of service are considered as service systems. A wide variety of mobile services exists, each fulfilling different types of needs (e.g. Anckar and D���Incau, 2002 Bouwman et al., 2008). Thus, organizing, structuring, and analysing mobile service usage data raises a need for classifying the services. Ideally, classification should be based on consistent use of relevant criteria, and should produce categories that are mutually exclusive and jointly exhaustive. Mobile service classifications have been proposed by many authors, for different purposes and using different criteria. Velez and Correia (2002) take a network viewpoint and classify mobile broadband services and applications based on the characteristics of the traffic produced by them (Interactive/Conversational, Interactive/Messaging, Interactive/Retrieval, Distributed/ Broadcast, Distributed/Cyclical). Holma et al. (2007) divide UMTS services to five main categories (person-to-person circuit switched services, person-to-person packet switched services, content-to-person services, business connectivity, and location services), and further to twelve subcategories. Pura and Heinonen (2007) take an end-user point-of-view to classification, identifying four dimensions of mobile services: type of consumption (hedonic vs. utilitarian use), temporal and spatial context, social setting, and relationship between the user and the service provider. The purpose of this paper is to construct a holistic framework that supports designing mobile service usage research as well as communicating, positioning, and comparing research results. The framework helps academics and practitioners to design and to recognise the limitations of mobile service usage studies, and to avoid misinterpretations based on insufficient data. By mapping research settings and results to the framework, it is possible to recognise areas where further data collection and analysis might be required. The PAGE 54 j info j VOL. 11 NO. 4 2009

framework also helps in selecting the most appropriate methods and measurement points for different research questions. The paper is structured as follows. First, we discuss the research process that resulted in the proposed framework and identify four relevant measurement points that can be used to collect data on mobile service usage. Then, we introduce our framework, comprising of the measurement points and technical components of mobile services, together with suitable classification criteria and resulting categories. We then proceed to discuss the relations between the measurement points and service components. Finally, after introducing the framework, we apply it to the analysis of mobile browsing studies, an important and timely topic area in the convergence point of mobile and Internet services. The paper is then closed with discussion and suggestions for further research. 2. Research process and methods The framework presented aggregates our experiences from a series of longitudinal and cross-sectional mobile service usage measurements carried out in Finland annually in 2005-2008, covering 80-90 per cent of all mobile users and service usage. Although relatively small and remotely located in Northern Europe, Finland is well-suited for studies concerning advanced mobile services for a number of reasons. With a population of 5.3 million, it ranks among the 12 richest countries in the world using GDP/capita as a measure.Finland is a highly industrialised open economy, and was No. 6 in World Economic Forum���s Global Competitiveness Index in 2007-2008 and 2008-2009 (www.weforum.org). It was also the first country in the world to launch GSM-based digital mobile communications networks in 1991, and led the global mobile service penetration statistics throughout the 1990s. Furthermore,the world���s largest mobile manufacturerNokia has its home and origin in Finland. Our mobile service usage measurements have taken place annually in 2005-2008. Descriptive results from these measurements have been published in Kivi (2008) and Verkasalo (2008). The measurements were carried out in close collaboration with the key players of the industry, including all three mobile network operators and Nokia. This wide support of the national mobile industry allowed us to collect data of comprehensive scope. Broad use of multiple data collection methods and measurement points has enabled data and method triangulation, as well as analysis and comparison of their scopes and limitations. Information on the usage of internet and mobile services can be collected from a range of different sources (see, e.g. Kivi, 2009). In our measurements, data were collected by: B capturing the IP traffic from mobile operator���s central network nodes B by utilizing the usage accounting systems of mobile operators B by monitoring end-users��� mobile handsets. Furthermore, the handset-based measurements were complemented with questionnaires filled by the participating users. To our knowledge, synchronised data of the same detail and scope have not been earlier available to researchers from any other mobile market. In general, mobile service usage data can be collected from four main sources, as illustrated in Figure 1. 1. End-users. Surveys are a commonly used data collection method for studying mobile user behaviour and service usage. Surveys are conducted on samples of real end-users, and can be implemented using various methods. Time series data can be produced by repeating a certain set of questions. Continuous panel studies, where the participating panellists register usage events manually to an online or paper diary often result in data of higher accuracy and granularity. 2. Usage monitoring systems. Usage monitoring includes both user monitoring as well as device monitoring systems. Device monitoring is common in studying PC and Internet usage, and is also used in TV audience measurements. The evolution of mobile phones towards computer-like devices has made it possible to conduct device monitoring also at mobile handsets (see, e.g. Verkasalo and Hammainen, �� �� 2007), regardless of whether any VOL. 11 NO. 4 2009 j info j PAGE 55