New Service Development in Financial Service Institutions
The influence of deviations from structured NSD stages
on firms’ innovation performance
By
A.E.M. Rademaker
Master Thesis MSc in BA, specialization: Strategic Innovation Management
University of Groningen Faculty of Economics and Business
First Supervisor: Dr. R. A. van der Eijk Second Supervisor: Dr. W. Biemans
19-6-2014 A.E.M. Rademaker Hoendiepskade 26a 9718BH Groningen The Netherlands a.e.m.rademaker@gmail.com Student number: s2586304 Telephone number: +31626875731 Abstract
Financial service institutions face a dynamic business environment as a result of deregulation and redrawn boundaries. They need to innovate and a key success factor in developing new financial services the proficiency of the activities in the new development (NSD) process. In this study, the innovation process is critically analysed from two perspectives: the Process Theory, wherein the formal process in which it was originally created needs certain constant conditions, and the Configuration Theory, which argues that deviations from such an ideal process are possible with respect to NSD stages. The results indicate that following the activities from the idea generation stage is beneficial for both exploitative and explorative performance, while following the activities from business analysis and marketing strategy stage and technical development stage are beneficial in case of explorative innovations. Furthermore, following the testing activities accordingly is beneficial in terms of exploitative innovations. Additionally, it has been confirmed that merging the activities from the final two NSD stages positively influences firms’ explorative new service innovation performance. Overall, it is found that the more deviations from the NSD process negatively influences firms’ innovation performance.
Keywords: New Service Development (NSD), Financial service institutions, Innovation performance,
Configuration theory
Table of Contents
List of figures ... i
List of tables ... i
1. Introduction ... 1
1.1 Problem definition ... 1
1.2 Purpose and significance study ... 2
1.3 Research question ... 2
1.4 Research scope and domain ... 2
1.5 Research outline ... 3
1.6 Summary chapter ... 3
2. Theoretical background ... 3
2.1 Financial service innovations ... 4
2.2 New service development process ... 4
2.3 New service innovation performance ... 6
2.4 Summary chapter ... 7
3 Hypotheses and conceptual model ... 7
3.1 A formal execution of NSD activities ... 7
3.2 Deviations from the NSD process ... 9
5.1 Data reduction results ... 17
5.2 Descriptives and correlations ... 18
5.3 Hypotheses testing ... 20 5.4 Focus-group interview ... 24 5.5 Summary chapter ... 25 6. Discussion ... 25 6.1 Summary chapter ... 27 7. Conclusion ... 27
7.1 Implications for management ... 28
7.2 Limitations ... 28
7.3 Future research ... 29
References ... 30
List of figures
Figure 1 New service development process ... 5
Figure 2 Conceptual model on firm level ... 10
List of tables Table 1 Theoretical concepts and explanations ... 3
Table 2 Sample statistics ... 12
Table 3 Overview of variables ... 16
Table 4 Cronbach Alpha for dependent and independent variables ... 17
Table 5 Descriptive statistics ... 19
Table 6 Correlation statistics ... 19
Table 7 Regression analysis Hypothesis 1 and 2 ... 21
Table 8 Theory and expert profile ... 22
Table 9 Regression analysis Hypothesis 3a - EDE... 23
Table 10 Regression analysis Hypothesis 3b - TDE ... 24
Table 11 Hypothesis confirmation and rejection ... 25
Table 12 Focus-group interview ... 42
Table 13 Rotated component factor analysis for dependent variable ... 46
Table 14 Pattern factor analysis for independent variable ... 46
Table 15 Factor analysis control variable (formality) ... 46
Table 16 Product newness as control variable ... 47
Table 17 General information sample - Finance (n=49) ... 48
Table 18 General information sample - Fintech (n=35) ... 48
Table 19 General information sample - Insurance (n=30) ... 48
Table 20 Regression analysis prior PCA Hypothesis 1 and 2 - Finance ... 49
Table 21 Regression analysis prior PCA Hypothesis 3 - Finance ... 49
Table 22 Regression analysis prior PCA Hypothesis 1 and 2 - Fintech ... 50
Table 23 Regression analysis prior PCA Hypothesis 3 - Fintech ... 50
Table 24 Regression analysis prior PCA Hypothesis 1 and 2 - Insurance ... 51
1. Introduction
“The best way to predict the future is to create it” (Drucker, 1991).
In today’s fast changing world services play an ever increasing role. Its’ share of GDP in the world is over 70% and rising (Cusumano, 2010). It covers the value added in wholesale and retail trade, transport, government, financial, professional and personal services, healthcare and real estate services (World Bank, 2012). Moreover, service innovation is increasingly recognised as the driving force behind firm performance (Mention & Torkkeli, 2012) and firms are focusing on creating value. In order to do so the new product development (NPD) process is often being used (Patterson, 2009). This is a process for success, survival and renewal of firms (Brown & Eisenhardt, 1995) and has been studied in-depth to establish best practices (Cooper, Edgett & Kleinschmidt, 2004; Cui, Chan & Calantone, 2014; Nash, 1937). However, just following the studied best practices might not bring the innovation needed to survive in today’s dynamic service business environment. Besides, this mature literature field about the innovative NPD process is often simultaneously used for product and service innovations (Ndubisi & Agarwal, 2014). Even with the differences between the innovation process for products and services, the boundaries between manufacturing and service firms are breaking down and both start to look alike. Traditional manufacturing firms have reinvented themselves as service providers in order to survive, which is also known as servitization (Neely, 2008). Furthermore, companies with high product innovations face direct increasing profitability from services supporting those products (Eggert, Hogreve, Ulaga & Muenkhoff, 2011). As a result, New Service Development (NSD) has become popular among both scholars and practitioners (Alam, 2006).
1.1 Problem definition
important than the individual ingredients? Or more specifically, will the innovation outcome differ when certain NSD activities are not performed strictly?
1.2 Purpose and significance study
The purpose and significance of this study are twofold. Firstly, despite the ever increasing importance of services in the world, the literature field of NSD remains scarce and is simultaneously used for product and service innovations (Jiménez-Zarco, Martínez-Ruiz & González-Benito, 2006; Ndubisi & Agarwal, 2014). Interestingly, according to Papastathopoulou and Hultink (2012), one of the six main NSD researched topics is ‘process and stages’ and this specific topic has a decreasing amount of articles published in the business literature (from 30% in 1982-1995, to 18.4% in 2002-2008) while services play an ever increasing role in today’s businesses (World Bank, 2012). Secondly, together with the two earlier mentioned contrasting theories, Process theory and Configuration Theory, the study at hand contributes by investigating how they can be applied to practice. Despite the existing NSD literature streams, evidence on the theoretical explanations of deviations from the structured NSD processes are still inclusive. Overall, theory and practice seem to be conflicting and Hauser, Tellis and Griffin (2006) suggest that more research is needed to better understand NSD in practice.
1.3 Research question
The proposed study provides empirical evidence to demonstrate the relationship between a firm’s formal NSD process and its new service innovation performance by answering the following research question: “What is the influence of deviations from the formal NSD process on financial firms’ new
service innovation performance?”
The following sub-questions have been developed in order to answer the main research question: - What is the importance of service innovations in Dutch financial institutions?
- What are pros and cons from following a formal execution of the NSD process?
- What is the influence of customer- and employee interaction in the final two NSD stages on explorative innovations?
1.4 Research scope and domain
formal NSD context by acknowledging the deviations of the structured process on firms’ new service innovation performance.
1.5 Research outline
The remainder of this paper is structured as follows. First, a literature review is conducted to highlight relevant theoretical concepts. Subsequently, the hypotheses are developed in line with the earlier mentioned conflicting theories and a conceptual model is created. Next, the research approach is described extensively and research findings are documented. Furthermore, the findings are discussed and analysed in-depth by using a focus-group interview. Finally, conclusions are drawn, implications for theory, management and policy makers established, and limitations stated.
1.6 Summary chapter
The rise of financial services offered together with the theoretical interest in the innovation process increased the popularity of the NSD process among both scholars and practitioners. However, the literature field of NSD remains scarce and the exact execution of this process, structured or flexible, has been argued by different literature streams namely: Process Theory and Configuration Theory. The study at hand explores the influence of deviations from the formal NSD process on Dutch financial firms’ new service innovation performance.
2. Theoretical background
The literature review at hand provides an overview of the different theoretical constructs that firms experience related to service innovations. The constructs and theory choices made are summarised in Table 1.
Table 1
Theoretical concepts and explanations
Concept Explanation
Financial Service Innovation Largely identified are the factors contributing to the creation of new services: A market-driven approach wherein the needs and preferences of the end-users are very important and a major source for economic growth and employment (Scarbrough & Lannon, 1989; Sinha, Maheshwari & Kedia 2013).
New Service Development Process (NSD)
Activities executed and decisions made to generate ideas, develop the concept, analyse the opportunity, test and launch the service (Cooper, Easingwood, Edgett, Kleinschmidt, & Storey, 1994).
New Service Innovation Performance The performance outcome of a new service is the result of the development process followed. This is influenced by the innovativeness of the new service (Avlontinit et al. 2001; de Brentani, 2001; Droege, Hildebrand & Heras Forcada 2010).
2.1 Financial service innovations
This research is set in the financial service industry, since these firms face time-to-market issues in their innovation process (Deloitte, 2015; Lievens & Moeanaert, 2000). This dynamic environment arose as a result of deregulation and redrawn boundaries for financial institutions (Blazevic & Lievens, 2004). Financial service firms increasingly try to innovate by the emergence of online banking and e-business like crowdfunding and data mining (Alam, 2006). However, the number of growth initiatives usually grows faster than the capacity of the firm to bring them to market (Day & Moorman, 2010, p.107). Opportunities for further acquisitions are narrowing and financial firms need to find ways to grow organically and thereby innovate in galore (Thomke, 2003). Hence, service innovation is a topic of great interest and a major source for economic growth and employment since financial institutions are constantly seeking for new markets and new financial services (Sinha et al., 2013). Berger and Nakata (2013) describe that research on financial service innovation largely identified the factors contributing to the creation of new services. Respectively, this process entails a market-driven approach wherein the needs and preferences of the end-users are very important in order to offer innovations.
Within the financial service industry, not all large banks are able to provide satisfactory innovative solutions for the market and that is one of the reasons why collaboration between banks and other industries are stimulated (Banken, 2015). The exponential growth of new market players resulted in a situation in which it is more difficult to judge companies on scale alone. Instead, “regulators should engage with more parties to learn about new technologies and how that will affect the markets” (Lee, 2015, p.1). According to Scarbrough and Lannon (1989), the redesign of bank branches is heavily dependent on Information Technology (IT). Moreover, a study of Accenture showed that 277 million euros was made available for financial institutions to invest in new innovations from IT software firms (Finance Innovation, 2015). This explains why financial technology (‘fintech’) firms, are often incorporated on behalf of the financial institution (Emerce, 2015). Additionally, fintech firms deliver not only to banks, but in some cases they also offer services for customers to bypass the bank. Therefore they can be seen as an opportunity as well as a threat from the perspectives of banks (Finance innovation, 2015).
2.2 New service development process
Figure 1
New service development process
innovation. Additionally, innovation in services is less disciplined than in the manufacturing sector (Chesbrough, 2005) since innovative and entrepreneurial initiatives cannot be planned precisely in advance (Moorman & Miner, 1998). That might be the reason why innovation processes are often perceived as ad-hoc and hardly ever repeated and formalised into the standard service offering of a firm (Droege, Hildebrand & Heras Forcada, 2010). Ad-hoc innovations are defined as “the interactive (social) construction of a solution to a particular problem put forward by a client” (de Vries, 2006, p. 1039).
A key success factor in developing new financial services is the proficiency of the activities in the new development process (de Brentani, 1993). A systematic search for growth initiatives that potentially can enter the market can be done with a structured stage-gate model which allows managers to continue or terminate kill projects in time (Schilling, 2010). This stage-gate model, initiated by Cooper (1990), provides clear stages for the innovation project which can be helpful when uncertainty is involved. Uncertainty can be defined by the extent to which future states of the environment can be anticipated or accurately predicted (Tushman & Anderson, 1986). The success of a certain service, in contrast to products, largely depends on what customer thinks (subjective assessment) since customers are not simply the ‘open wallets’ at the end of the supply chain (Frei, 2006). Firms must spend time and effort to discover and monitor customers’ needs (Bowen & Ford, 2002) which forces them to integrate the customer in the NSD process (Alam, 2006). Research strands indicate the importance of NSD (Alam, 2006) and many authors (e.g. de Brentani, 1989; de Brentani & Ragot, 1996; Froehle et al., 2000) address the importance of following structured stages in order to be successful. This might be the reason why a NSD process moves systematically through a set of stages. Although different approaches have been proposed, they are in fact all variants on a linear theme and they all follow certain steps (Bitner et al., 2008). This study uses a NSD process from Avlonitis et al. (2001) visualised in figure 1 and briefly described below.
The first stage, idea generation, has been characterised by low levels of formalisation, unstructured processes and uncertainty (Khurana & Rosenthal, 1997). Idea quality or idea creativity refers to ideas that are both novel and useful (Rietzschel, Nijstad, & Stroebe, 2010) but next to the idea generation, the ideas should also be screened with the objective to narrow the amount of ideas. Managers are generally not sure of the best way of generating new ideas and follow a ‘try it and see’ approach. However, this stage dictates all further stages of an innovation process and is therefore very crucial (Alam, 2006). Concerning the second stage, business analysis and marketing strategy, the business attractiveness of the remaining ideas have to be examined (Papastathopoulou, et al., 2001). Specifically, it is based on the analysis of market conditions needed to develop the new service which is important since it is the
foundation for successful innovation efforts (Sunbo, 1997). Furthermore, it incorporates an analysis of firm specific conditions and requires that firms continuously collect information to create new value. The third stage, the technical development, relates to the design and the development process. Respectively, the firm “develops and tests the core service, delivery system, associated marketing program operationally and trains operational and frontline personnel” (Melton & Hartline, 2010, p.412). The first three stages of the NSD process are also known as the ‘fuzzy front-end’ stages of the NSD model and have been studied by for example Murphy and Kumar (1997), Khurana and Rosenthal (1997) and Alam (2006). The final two stages are known as the “extent to which firms undertake prelaunch service testing, personnel training and an internal and external promotional program during new service launch” (de Brentani, 2001, p.176). The testing stage, referring to testing of operational aspects, concerns in-house and within-market testing. According to Kahn (2005) it can be used to estimate the potential sales value by exposing the service to customers or personnel. In preparation of the final stage, service launch, extensive testing among customers and employees would help managers to determine acceptable levels of service quality (Melton & Hartline, 2010). The final stage refers to the full-scale introduction of the service to the market and the evaluation of its performance (Avlonitis et al., 2001, p.326). However, identifying the targets for the new service and properly aligning the final offering and related communication strategies to address those targets are critical tasks during this stage. This is especially the case for more explorative innovations (Danneels, 2002) and Xie, Bagozzi and Troye (2008) conclude that the positioning and promoting of a new offering for different categories of prospects is more challenging in service contexts than in product contexts because of the highly experiential nature of service outcomes.
2.3 New service innovation performance
Explorative innovation also involves related terms such as radical innovation, breakthrough innovation, really new innovation, and new-to-the-world innovation such as fundamentally new loan structures and contingent contracts not previously provided by organisational units (Uzzi & Lancaster, 2003). This type of innovation has the potential to set up a new market, create first-mover barriers and provide lead-time advantages. However, very few firm develop services that create a new market or reshape a market completely. De Jong and Vermeulen (2003) argue that radical innovations are led by customer input. This is especially relevant for financial service firms, since they are characterised by a continuous interaction with their environment (Lievens, 2000).
Exploitative innovation captures the extent to which units build on existing knowledge and meet the needs of existing customers and is mainly based on increasing efficiency (Uzzi & Lancaster 2003). It involves less market uncertainty, fewer risks and is easier to support using firms’ existing resources. According to Berry, Shankar, Parish, Cadwallader and Dotzel (2006) most firms make exploitative innovations to their services and it had been argued that those innovations are often treated as an operational activity which does not have to be managed according a rigorous NSD (Oke, 2007). Overall, exploitative innovation involves related terms as incremental innovation, new-to-the-firm innovation and line- or brand extensions.
2.4 Summary chapter
The dynamic service environment forces Dutch financial service firms to innovate and a key success factor in developing new financial services is the mastery of the activities in the new development process. This NSD process moves systematically through a set of stages: idea generation, business analysis and marketing strategy, technical development, testing and service launch. Following this process in a structured manner can be very beneficial for financial firms, especially when uncertainty is involved. The performance outcomes are influenced by the innovativeness of the new service and two types are recognised in this study: explorative and exploitative.
3 Hypotheses and conceptual model
3.1 A formal execution of NSD activities
Vermeulen, 2003). The formal process in which it was originally created (e.g. formal NSD process) needs certain constant conditions in order to duplicate an outcome (e.g. service innovation performance). The Process Theory enhances organisational effectiveness and explains that there is a logical variable form: If not-X (necessary conditions), then not-Y (outcome). This means that it cannot be extended to “if more X, then more Y” (Soh, & Markus, 1995, p. 31). In line with the reasoning of the process theory (Mohr, 1993) and innovation management theories like the market-based view (Mintzberg et al. 1998), constant necessary conditions must be reached to keep the process exactly the same as it was originally created to obtain performance. Hence, Edvardsson et al. (2013) show that the number of failures in NSD can be reduced by focussing on following a formal NSD strategy. This would explain why several authors found that a structured execution of NSD models is beneficial, hence deviations of this original process will lead to worse results. Consequently, it is hypothesised:
H1a; Following the activities from the idea generation stage positively influences firms’ new service
innovation performance.
H1b; Following the activities from the business analysis and marketing strategy stage positively
influences firms’ new service innovation performance.
H1c; Following the activities from the technical development stage positively influences firms’ new
service innovation performance.
H1d; Following the activities from the testing stage is positively influences firms’ new service
innovation performance.
H1e; Following the activities from the launching stage positively influences firms’ new service
innovation performance.
3.1.1 Execution of testing and service launch activities
As described in Chapter 2.2, the success of an innovation largely depends on what the customer thinks and firms integrate customers in the NSD process (Bowen & Ford, 2002). More specifically, during the final two stages of the NSD process, the customer (and employee) contact is used to estimate potential sales and align the final offering. This process cannot be planned precisely in advance especially in the case of more explorative innovations since these type of innovations have the potential to set up a new market. Prior research has produced inconsistency related to the role of customer involvement in explorative innovations. On one hand side, radical innovations are led by customer input (de Jong & Vermeulen, 2003) and this will positively affect customer perceptions of the service quality. Next, frontline employees can improve customer perceptions like ease of use and brand image of the focal firm. Thereby it can improve the firms’ image, exceed their sales objectives and lead to greater speed-to-market. (Melton & Hartline, 2010).
of that market is impossible (Danneels, 2002). According to Danneels (2002), it involves identifying potential customers, constructing relationships and building communication channels which is very risky and demands more financial and human resources. Next to these major market investments and unknown outcomes, the investigation and the involvement of such customers can be signalling for competitors who might use this information into their own decision calculus as they determine when, whether and how to respond (Aboulnasr, Narasimhan, Blair & Chandy, 2008). Therefore, in the case of explorative innovations, it seems beneficial to combine the activities of the final two NSD stages to prevent potential customers to create first-mover barriers and have lead-time advantages (Uzzi & Lancaster, 2003). Consequently, is it hypothesised:
H2 Merging the activities from the final two NSD stages is positively associated with firms’
explorative new service innovation performance
3.2 Deviations from the NSD process
Despite the line of reasoning of the Process Theory, the innovation process of services are often perceived as ad-hoc (Droege et al., 2010). More specifically, the various entrepreneurial initiatives cannot always be planned in advance as described by Moorman and Miner (1998). The degree of changes within financial service innovations are less a matter of a logical development process with pre-defined stages (Scarbrough & Lannon, 1989). Hypothesis 2 explains such disruption of the process and captured the interaction concerning the final two NSD stages with explorative innovations. The configuration theory captures the interaction and deviation among all NSD stages. More specifically, it provides accurate and useful details about individual structure and process variables (Miller, 1981). It is a “conditional association of the independent variables with one dependent variable” (Meyer, Tsui & Hinings, 1993: p. 1177).
Control Variables - Industry - Firm age - Firm Formality - Firm size - Firm Innovation Experience - Product Newness H1a + H1b + H1c + H1d + H1e + H3ab - Figure 2
Conceptual model on firm level
H2
-deviation from the entire process, the larger the influence might be towards performance1. Hence it is
hypothesised:
H3a: The greater the deviation from an expert profile with respect to all NSD activities, the larger
the negative effect on a firm's new service innovation performance.
H3b: The greater the deviation from a theoretical profile with respect to all NSD activities, the
larger the negative effect on a firm's new service innovation performance.
3.3 Summary chapter
Innovation is currently high on the strategic agendas of many financial services firms. Following the NSD process in a structured manner enhances organisational effectiveness and reduces the number of NSD failures which has been confirmed by the Process Theory. Nevertheless, other literature streams argue that firms cannot always plan each activity of the NSD process in advance, especially in the case of more explorative innovations and emerging customer needs. Moreover, since each firm is unique and customers interact throughout the process (Dobni, 2006), a more flexible execution of the whole NSD has been argued by the Configuration Theory where individual activities does not have to represent the whole (Drazin & Van de Ven, 1985).
1 This research study is based on an assumption that there are two ideal profiles of following NSD processes to ‘deviate from’
within a given environment. This does not imply that there are only two successful strategies which is a useful line for future inquiry.
New Service Innovation Performance
4. Methodology
This study follows a theory testing approach since several theoretical gaps are identified and there is sufficient literature available to derive hypotheses. The established research question required an in-depth understanding of Dutch financial institutions and without any existing database the researcher was required to collect the data herself by means of a survey. The overall research strategy, the type and number of data required for each stage together with the method of analysis is described below.
4.1 Literature review
In order to describe the business phenomenon, high quality journals were investigated to identify theoretical gaps in support of the research interest. Hence, based on peer-reviewed articles from the Business Source Premier database, articles in the literature streams of service innovations, NSD, NPD (since people use terms like ‘financial products’ instead of ‘financial services’), financial industries, service innovation performance, types of innovation, Process Theory and the Configuration Theory were reviewed to identify the important constructs and generate a conceptual model and hypotheses
.
Furthermore, related theories were used to find suitable variables to be tested which are described in the method of analysis section in Chapter 4.5.4.2 Sample creation
via online innovative websites (Finno, 2015; Financeinnovation, 2015; Hollandfintech, 2015). The total sample consisted of 150 Dutch financial firms which participated in a NSD process from 2010 to 2015. Other details (e.g. size and age) of the selected firms were obtained from LexisNexis, LinkedIn and firm specific websites. Moreover, recent innovations (2010-2015) of the 150 selected firms were collected via firm’s websites to personalise the pre-notification per firm/respondent.
Additionally, one person per firm who was responsible for the NSD process was selected to participate in the study. These individuals were identified via the ‘function option’ on LinkedIn and contacted by an e-mail to state their agreements to participate. A personalised pre-notification letter was mailed to the potential respondents explaining the objectives of the study and soliciting cooperation.
4.3 Data collection
After the total sample of 150 firms was created, a test was executed among two firms. This pre-test consisted of an actual meeting wherein the respondent read the questions out loud and gave insights in their thinking and line of reasoning. By doing this, the researcher was able to modify questions improving the respondents’ understanding (e.g. translate the questions into Dutch). The responses of the two firms were excluded for further analysis by the researcher. The optimised questionnaire was then directed towards the NSD manager, project/service manager or service delivery manager responsible for NSD activities. The questionnaire was sent to the general manager or (research) director if the former function did not exist within the particular firm. Once the contact person agreed on participation, the link to the survey and the specific requirements to fill out the survey (via Google Spreadsheet Document Function) were sent by e-mail.
Subsequently, multiple reminders by e-mail and phone followed and 114 persons of the total sample (n=150) replied to the message and agreed to participate – anonymously - . This resulted in a response rate of 76% and a distribution of 49 participants in financial institutions, 35 participants in fintech and 30 in insurances (Table 2). The high standard deviations (σ) show a relative high variation in the sample statistics. The total sample comprised of 95 men and 19 women. The respondents held functions like service/product/marketing manager (40.3%), NSD/innovation manager (31.6%) or were director/owner/board member (28.1%).
Table 2 Sample statistics
The survey began without telling the respondents about the specific purpose of this study and the questions in the survey were asked in a random order. Furthermore, no specific NSD stages were mentioned so that respondents did not know that they were in a specific stage. This was mainly done
Industry Amount of
respondents
Avg. Age Avg. Size Avg. Innovation
Experience
because a quantitative approach has the possible weakness of people given socially desirable answers (e.g. all activities ranked at 7=‘always’). To help avoid common method bias, the survey was instructed to include different types of responses. For instance, response types included yes/no answers, Likert scales and multiple-choice answers. As mentioned before, all questions were translated to Dutch to cater to the native language of the country of research, facilitating an easy understanding of all respondents (Appendix E).
4.4 Methods of analysis
The statistical program SPSS was used to execute analyses as described below. Prior to testing the actual hypotheses, a variety of tests was undertaken to secure the validity of the results. Firstly, a factor analysis which is a class of multivariate procedures aiming to identify the underlying structure in the data matrix (Hair, Anderson, Tatham & Black, 1995), was conducted. Factor analysis is often used in data reduction to identify a small number of factors that explain most of the variance observed in a much larger number of manifest variables. It can be done with a large sample size and a database that assumes reliable correlations, like this study’s database. In specific, the principal component factor analysis (PCA) is conducted since it uses the total variance of the variables in the computation process and derives factors that contain only the shared variance. Due to the existing large database, both analyses were set on a maximum iterations for convergence of 30, which specifies the number of times that SPSS will search for an optimal solution. Moreover, they were sorted by size, meaning that the variables were listed according to the size of their factor loadings. Additionally, it suppressed small numbers (<0.4) for interpretation purposes based on Stevens (2002) because loadings above 0.4 represent substantive values. Then the variables can be constructed and after the reliability is set, a multiple linear regression allows determining the overall fit (variance explained) of the model and the relative contribution of each of the predictors to the total variance explained. The multiple independent variables related to a dependent variable, including the control variables and the hypothesised interaction between the latter two NSD stages, can be examined and hypothesis one and two can be tested.
independent variable towards the dependent variable is calculated and deviation from such an ‘ideal (theoretical and expert)’ profile reflects a unit of misalignment and should have corresponding negative relationships with performance. Common approaches to the interaction test of ‘fit’ consist of a series of two-way analyses of variance, multiple regression analysis with interaction terms, ANOVA or subgroup analysis (Venkatraman & Prescott, 1990). This misalignment is calculated by the following equation (Venkatraman & Prescott, 1990):
MISALIGN =
Where = the score for the formal NSD activity in the study sample for the jth variable.
= the mean score for the calibration sample (theory profile: each stage on 7, expert profile: average score per stage of experts) along the jth variable.
= standardised beta weight of the OLS regression for the jth variable in the environment.
= 1n where n is the number of NSD activities that are significant related to firms’ new service innovation performance in that environment.
4.5 Measures
4.5.1 Dependent variable
4.5.2 Independent variables
The independent variable is the formal NSD process and its stages: idea generation activities, business analysis and marketing strategy activities, technical development activities, testing activities and service launch activities (IG, BAMS, TD, T and SL). These stages have already been reported in previous investigations (Calantone & Cooper, 1979; Urban & Hauser, 1993) and are translated into Dutch to be more appealing for the respondents. The literature review of Avlonitis, et al. (2001) provided 29 specific development activities suiting the five stages and is therefore used as the baseline for this study. The respective activities are used as a scale, acknowledging that they are an outcome of their study and not used as a scale before (each activity is scored on a 1-7 Likert scale).
4.5.3 Control variables
In order to assess and clarify the relationship between the other variables, the following control variables are identified. First, the size of the firm is assessed with an open-ended question about the total number of employees (Doty et al., 1993). Second, experience in the field of financial services is measured by the age (years of existing) of the firm in line with Delery and Doty (1996). Moreover, the innovation experience of the firm in the financial service industry will be measured by the amount of years the firm is actively working with innovation (innovation experience). Fourth, there is a control for product newness since the innovation process differs per project. Since this study uses three sectors to measure the Dutch financial industry, there will be a control for industry. Finally, it is controlled for
formalisation like Jansen et al. (2006) did in their study concerning financial service innovation
(Desphandé & Zaltman, 1982). Formalisation is the degree to which rules, procedures, instructions, and communications are formalised or written down (Khandwalla, 1977). It is a commonly criticised term that can either be positively related to new financial services or cause delays and an overload of information (Papastathopoulou et al., 2001).
4.5 Focus Group
In addition to the quantitative research approach, a focus group is set up to gather qualitative data performed in a group setting (Fontana & Frey, 1994). According to Lederman (1990) this methodology is often used to confirm the data gathered by another approach, like the quantitative survey in this study. This approach encourages discussion among group members, and this interaction tends to have a synergy effect to stimulate ideas (Morgan, 1988). Furthermore, it is said to be more revealing than the sum of the individual surveys (Lederman, 1990) which strengthens this study. According to Hartman (2004) six steps should be undertaken to execute focus groups which are summarised in Appendix D.
4.6 Controllability, validity and reliability
independent variables is secured by pre-testing the survey and checking for socially desirable responses. A Likert scale is used for both independent as dependent variables since it is a direct measure of attitudes and designed to circumvent the problem of obtaining meaningful quantitative answers and generate statistical measurements (Likert, 1932). Having seven points tends to be a good balance between enough points of discrimination without having too many response options (Nunnally, 1978). Moreover, since this study deals with constructs that can only indirectly be observed in reality, more than one measurement instrument is needed to show that the outcomes share a common variance. In order to provide reliability, standardisation of data collection is done wherein the results should be independent of the tools used (use of tools: survey). Furthermore, when all data was obtained, a focus-group interview was conducted to validate the results in line with Hitt and Middlemist (1978).
4.7 Summary chapter
A final sample of 114 Dutch financial firms was created after a personalised pre-notification letter. Since no existing data suited the research purpose of this study, a database had to be created and detailed descriptions of this process were provided to secure the controllability. Moreover, a variety of tests was undertaken to secure the validity of the results and validated scales were used to create an online standardised survey and improve the reliability of the study. An overview of all variables used is provided in Table 3.
Table 3
Overview of variables
Variable Measure operationalisation Scale Appendix
Dependent Service Innovation Performance
An explorative and exploitative 1-7 Likert scale (1: never, 7: always) used from Jansen et al., (2006) and financial indicators (% new sales, % new profit).
Ordinal A Independent Idea Generation (IG) 1-7 Likert scale (1: never, 7: always) of Avlonitis et al.
(2001) with six questions.
Ordinal B Business Analysis
and Marketing Strategy (BAMS)
1-7 Likert scale (1: never, 7: always) of Avlonitis et al. (2001) with eight questions.
Ordinal B Technological
Development (TD)
1-7 Likert scale (1: never, 7: always) of Avlonitis et al. (2001) with five questions.
Ordinal B Testing (T) 1-7 Likert scale (1: never, 7: always) of Avlonitis et al.
(2001) with three questions.
Ordinal B Service Launch (SL) 1-7 Likert scale (1: never, 7: always) of Avlonitis et al.
(2001) with seven questions.
Ordinal B Control Size (S) Open-ended question about the number of employees (Doty
et al. 1993).
Ratio Age (A) Open-ended question about the number of years from the
founding date (Delery & Doty, 1996).
Ratio Innovation
Experience (IE)
Open-ended question about the innovation experience of the firm in number of years.
Ratio Formalisation (F) A 1-7 Likert scale (1: never, 7: always) of Desphandé &
Zaltman (1982) with 5 questions.
Ordinal Product Newness
(PN)
Three questions concerning the type of product working on and/or launching more often (incremental/radical)
5. Results
5.1 Data reduction results
Prior to any in-depth analysis, a factor analysis was conducted to reduce the current dataset2.
Appendix F shows the outcomes for the independent, dependent and control variables provided by the Principal Component Analysis (PCA) which is widely used for data reduction. In case of the dependent variable, the orthogonal rotation ‘varimax’ with two factors to extract (EXLOR & EXPLOIT) was selected, because it minimises the number of variables that have high loadings (high correlations) for each factor. Only one item, EXPLOR3, was deleted due to the 0.4 suppression limit, as can be seen in Appendix F. The oblique rotation, ‘direct oblimin’ was chosen for the independent variables with five fixed numbers of factors, namely all five NSD stages (IG, BAMS, TD and SL). From the total 29 NSD activities, 12 activities remained, suggesting that the PCA extracted 17 activities. Reasons include the 0.4 suppression or the deletion of items which were not a linear combination of original variables. Concerning one control variable, the formality scales from Desphandé and Zaltman (1982), the PCA had to be executed with one factor to extract. Firstly, one code (F3) had to be reversed, since it was
negatively worded. By reversing this code, a high value indicates the same type of response on every item. This time, with no rotation selected, all factors and factors suppressed by the <0.4 limit and resulted that no items need to be deleted.
When the PCA was conducted, new variables were constructed with the remaining components of the dataset. More specifically, means of the remaining components per stage were calculated and a Cronbach Alpha was run for all new variables (except for T1 since this stage had only one question left after PCA). The Cronbach Alpha is an estimate of the internal consistency associated with the scores that can be derived from a scale (Cronbach, 1951). According to Cronbach (1951), the reliability is stated to be important because in the absence of it, it is impossible to have any validity associated with the scores of the newly reduced scale. Nunnally (1978) argues that a Cronbach Alpha of 0.7 is ideal, which indicates that 70% of the variance in the score is reliable and 30% is an error variance.
Table 4
Cronbach Alpha for dependent and independent variables
Variables EXPLOR EXPLOIT IG BAMS TD SL
Cronbach Alpha
0.745 0.701 0.497 0.639 0.716 0.598
2 The objective of a factor analysis - PCA - is to identify a set of variables such that each variable, called a principal component,
As can be seen from Table 4 three variables do not comply with the 0.7 standard set by Nunnally (1978)3. Nevertheless, no drastic conclusions can be made about internal consistency solely based on
the level of alpha due to multiple reasons. On one hand side, and in line with the reasoning of Eunseong and Seonghoon (2015, p. 217), “the advised levels of alpha are neither the result of empirical research nor the consequence of clear logical reasoning; instead, they were derived from Nunnally’s personal intuition” (Churchill & Peter, 1984; Peterson, 1994). On the other hand, efforts to increase the Cronbach Alpha above a certain level may harm reliability and validity. Sacrificing the diversity of items to increase alpha hinders content validity (Raykov, 2007) especially in this case since 17 items were already extracted. Acknowledging the limitations of the Cronbach Alpha in this study, further analyses were conducted without significantly harming the validity of the results.
5.2 Descriptives and correlations
Table 5 provides the descriptive statistics for the dataset, after reductions from the PCA. The descriptive statistics of several variables are worth mentioning. First, concerning the dependent variables, respondents scored higher on exploitative compared to explorative performance measurements. This can be also be confirmed by general questions concerning their most common type of innovation launched (Appendix G). Only three banks and four insurance firms launch more radical innovations compared to incremental innovation, whereas the fintech firms launch more radical innovation. This is in line with Berry et al. (2006) who propose that most firms make incremental innovation to their services. Second, as can be seen from the independent variables, the front-end stages (first three) of the NSD process scored higher compared to the last two stages. The more mature deliberated development of the front-end stages compared to the latter two stages has also been confirmed in the literature (e.g. Alam, 2006; Khurana & Rosenthal, 1997; Murphy & Kumar, 1997). Finally, standard deviations of the control variables are extremely high, which shows the enormous variations in firms’ specific characteristic
3 An additional Cronbach Alpha analysis has been conducted for all variables prior PCA (incl. testing stage). Similar results
Table 5 Descriptive statistics
Table 6 Correlation statistics
Variable Mean Std. Deviation Minimum Maximum
1. Exploration 4,7237 1,14547 1,83 6,67 2. Exploitation 5,6053 ,77883 3,00 6,71 3. Idea Generation 5,1374 1,02555 2,33 7,00 4. Business Analysis & Marketing
The correlation matrix is provided in Table 6. Six variables have a negative Pearson r in correlation with explorative performance, one with exploitative performance measurements. This means that if the dependent variable (EXPLOR/EXPLOIT) increases, all other variables will decrease. Looking at the Sig (2-tailed) measurements, all independent variables (except for SL) in relation to exploration have a significant correlations (P<0.1). Again increases or decreases in one variable relate to increases or decreases in the other variable. An interesting observation in the correlation matrix is the significance of the interaction-effect (T*SL).
5.3 Hypotheses testing
To test the hypotheses, multiple logistic regression analyses were performed. These regression analyses were conducted with the constructed variables after PCA and excluded the industries (finance, fintech and insurance) to improve the analyses’ outcomes. Moreover, the control variable innovation experience (IE) showed high correlations with age and was excluded as well. By excluding this control variable, age became positively significant. The product newness was excluded since this scale was found to be very subjective and respondents did not answered these questions in a consequent manner (Appendix G). Moreover, the dependent variables incorporates the different types of newness. And therefore each analysis is conducted twice. Both models (EXPLOR & EXPLOIT) showed interesting results (Table 7).
As can be seen in Table 7, model 2 and 5 the model explains 25.9% (EXPLOR) and 13.9% (EXPLOIT) of the total variability in the dependent variable based on the adjusted R-square. Moreover, the ANOVA (F-value) shows that these models are significant (P<0.05). The table visualises the coefficient results concerning the positive relationships of all independent variables on the dependent variable (EXPLOR), while the control variables shows a negative relationship. However, only IG (P<0.1), BAMS (P<0.05) and TD (P<0.1) are significant. A similar situation can be seen from model 5, whereas only IG (P<0.05) and T (P<0.1) have positive and significant effects on the dependent variable (EXPLOIT). Additionally, the Tolerance (>,10) and VIF values (<10), which are an examination for multicollinearity issues, are examined and are in good condition in both models. Model 1 and 4, visualising the control variables does not show and significant relationship. Model 3, including the significant interaction-effect does show a negative significant relationship of one control variable, Age (P<0.1).
Concerning Hypothesis 1 it is predicted that following the activities from the NSD process positively influences firms’ new service innovation performance. In specific, all five stages of this NSD process are incorporated (H1a t/m H1e). As hypothesised in H1a, the first stage, idea generation, positively and
analysis and marketing strategy positively influences firms’ new service innovation performance. This hypothesis can also only be partially supported, since the model is positively significant for EXPLOR (P<0.01) with a B value of ,316. Hypothesis 1c, concerning the technical development, can also be
partially supported since the significance of the B value ,167 is only found for EXPLOR (P<0.1). As hypothesised in H1d, the testing stage positively and significantly influences firms’ new service
innovation performance. This hypothesis is also partially supported, since a significant (P<0.1) relationship has only been found for EXPLOIT. Finally, it is hypothesised that the activities of service launch are positively influencing firms’ new service innovation performance. Next to the non-significant relationship between both EXPLOR and EXPLOIT, the model shows a negative relationship and this hypothesis could not be supported.
As can be seen in Table 7, the interaction-effect shows a significant (P<0.1) relationship with firms’ explorative new service innovation performance. Moreover, the correlations in Table 6 were significant, except for the three industries and innovation experience (industry and innovation experiences were excluded when performing the regression analysis as could be seen in Table 7). Therefore, Hypothesis 2 can be confirmed. Moreover, as could be seen in model 3, one control variable –age- shows small negative relationship (P<0.1). This can be confirmed by the type of innovations firms most often pursue (Appendix G).
Table 7
Regression analysis Hypothesis 1 and 2
Model 1 EXPLOR β Model 2 EXPLOR β Model 3 EXPLOR β Model 4 EXPLOIT β Model 5 EXPLOIT β Model 6 EXPLOIT β Constant 5,232*** 2,012** -,997 5,434*** 3,760*** 4,768** Control Age Size Formality -,089 -,003 8,775E-6 -,004 1,179E-5 -,157 -,004* 8,115E-6 -,121 ,019 ,0024 7,023E-6 ,001 5,30E-6 -,082 ,000 3,396E-6 -,078 Independent variables IG BAMS TD T SL ,194* ,316** ,167* ,078 -,083 ,870 1,329** ,766 ,177 -1,143** ,154*** ,111 ,027 ,070* ,62 ,628 -,202 -,339 ,337 -,351 Interaction effects4 T*SL ,063* ,004 Model summary F R² Adjusted R² N 1,694 ,045 ,018 114 5,897*** ,312 ,259 114 3,716*** ,416 ,304 114 ,882 ,024 -,003 114 3,260** ,200 ,139 114 2,532** ,327 ,198 114 *P<0.1, **P<0.05, ***P<0.001
4 Addition analyses has been conducted to find other interaction effects among all independent and control variables.
In
order to test the third hypothesis, an expert value as well as a theoretical value has been established in order to measure the deviations from that profile. The theoretical value, in line with the Process Theory, results in an ‘ideal’ theoretical profile if all survey questions are graded with a 7 on the 1-7 Likert scale. The ‘ideal’ profile in case of the expert value was calculated by the % new sales, a variable that was provided by 55 of the 114 respondents. These 55 inputs indicated an average of 25.11 and a standard deviation of 25.679. Hence, all participants that had ≥ 50 % new sales were included in the ‘expert’ sample (n=10). Based on these ten experts, the mean of all NSD stages were collected and resulted in the ‘ideal’ expert value, as visualised in Table 8.Table 8
Theory and expert profile
Stages IG BAMS TD T SL
‘Ideal’ Theoretically 7 7 7 7 7
‘Ideal’ Experts 4.6667 5.6667 4.8333 4.7000 4.2000
The original dataset, excluding the experts (n=10), resulted in a new database of 104 inputs. New variables were computed for all IV’s with deviations from expert values using the mean of experts’ outcomes. Subsequently, as soon as all individual stages were added together, a new intertwined variable arose: EDE5. Deviations from experts (EDE) and a linear regression with these variables can be found
in Table 9. The table shows lower correlations with the dependent variable (EXPLOR/EXPLOIR) compared to the one used in hypothesis 1 (Table 7). Looking at the adjusted R-square, only 4.5% of the variability is explained by EXPLOR, while -0.8% is explained by EXPLOIT. Moreover, the ANOVA results show that only one model, EXPLOR, is significant with F-value 2,2205 (P<0.05).
Concerning hypothesis H3a, it was hypothesised the greater the deviation from an expert profile with
respect to all NSD activities, the larger the negative effect on a firm's new service innovation performance. EDE shows a negative relationship for EXPLOR as can be seen from Table 9 in model 2. Specifically, concerning the explorative performance the B value of -.043 is significant (P<0.05). However, a positive but not significant relationship is found concerning firms’ exploitative new service innovation performance This is in line with the F-values provided in model 2 and 5, which are only significant in the case of explorative innovation (P<0.05). Therefore, this hypothesis can only be partially supported. An additional analysis showed the possible interaction-effects of the independent variable and the three control variables as can be seen in model 3 and 6. However, no significant relationships could be found concerning these interaction-effects. Just like model 1 and 4 in Table 7, no significant relationship were found among the control- and dependent variables.
5 EDE was calculated by: ((IG-4.6667)*(IG-4.6667)) + ((BAMS-5.6667)*(BAMS-5.6667)) + ((TD-4.8333)*(TD-4.8333)) +
Table 9
Regression analysis Hypothesis 3a - EDE
Model 1 EXPLOR β Model 2 EXPLOR β Model 3 EXPLOR Model 4 EXPLOIT β Model 5 EXPLOIT β Model 6 EXPLOIT Constant 5.267*** 5,638*** 5,525*** 5,622*** 5,588*** 5,783*** Control Age Size Formality -,002 9,510E-6 -,107 -,001 8,021E6 -,1188 -,004 -1,772E-5 -,049 ,002 8,272E-6 -,027 ,002 8,406E-6 -,026 ,002 -1,077E-5 -,058 Independent variables EDE -,043** -,025 ,003 -,025 Interaction effects6 EDE*Age EDE*Size EDE*Formality ,000 2,878E-6 -,005 -5,085E-5 2,042E-6 ,006 Model Summary F R² Adjusted R² N 1,476 ,042 ,014 104 2,205** ,082 ,045 104 1,621 ,098 ,037 104 1,049 ,031 ,001 104 ,799 ,031 -,008 104 ,599 ,038 -,026 104 *P<0.1, **P<0.05, ***P<0.001
In order to test the theoretical deviations (TDE), the complete database (n=114) was used including the earlier extracted experts (n=10). Again new variables were computed for all IV’s, using the theoretical values. All individual stages were added together and a new intertwined variable arose: TDE7. The regression results in Table 10 show the respective relationships. Prior to analysing those
models, the model summary is examined. The second model has an adjusted R square of .031 which indicates that 3.1% of the variability is explained by EXPLOR. Another positive result of 11.3% of the variability is explained by EXPLOIT. The ANOVA results (F-values) show that both models are significant (P<0.05).
Hypothesis H3b predicted that the greater the deviation from a theory profile with respect to all NSD
activities, the larger the negative effect on a firm's new service innovation performance. As can be seen from Table 10, the B value of TDE is -,025 in model 2 for the explorative dependent variable. The significance levels show highly trustworthy results (P<0.001). Model 5 shows the significant B value of the exploitative dependent variable, again with a significant relationship (P<0.001). These findings suits the significant levels of the earlier mentioned F-values. Additionally, there has been checked for the Tolerance (>.10) and VIF values (<10) which are in good condition in both models. Thereby, this hypothesis is supported. Additional analyses showed the possible interactions-effects of the independent variable (TDE) with the three used control variables. However, no significant results are obtained as can be seen in model 3 and 6 and thereby no further conclusions can be made. Furthermore, just as in Table 7 and 9 no control variable has been found significant in Table 10.
6 Addition analyses has been conducted to find other interaction effects among all independent and control variables.
Concerning EXPLOR three interaction effects were found: TD*Size (P<0.1), T*Size (P<0.1),, and BAMS*Size(P<0.1). Concerning EXPLOIT only one interaction effect was found: IG*SL(P<0.05)
Table 10
Regression analysis Hypothesis 3b - TDE
Model 1 EXPLOR β Model 2 EXPLOR β Model 3 EXPLOR Model 4 EXPLOIT β Model 5 EXPLOIT β Model 6 EXPLOIT Constant 5,232*** 6,763*** 5,801*** 5,434*** 6,271*** 6,275*** Control Age Size Formality -,089 -,003 8,775E-6 -,003 5,304E-6 -,267 -,003 ,000 -,070 ,019 ,0024 7,023E-6 ,002 5,124E-6 -,078 ,000 4,109E-5 0,072 Independent variables TDE -,025*** -,001 -,013*** -,013 Interaction effects8 TDE*Age TDE*Size TDE*Formality -3,875E-5 -6,490E-6 -,005 5,857E-5 -2,259E-6 ,000 Model Summary F R² Adjusted R² N 1,694 ,045 ,018 114 7,981*** ,082 ,031 114 6,249*** ,294 ,247 114 ,882 ,024 -,003 114 4,657** ,031 ,113 114 2,664** ,151 ,094 114 *P<0.1, **P<0.05, ***P<0.001
5.4 Focus-group interview
Executed at one of the unique conference centres of the Regardz Business Hotel located in Amersfoort, the focus-group interview was held on the 10th of June (2015) with participants from the
three selected industries. Firstly, the moderator introduced all participants, presented the studies’ results and explained the purpose of this focus-group interview, namely: 1. Validate the studies’ results. 2. Interpret the results in terms of participants’ own experiences. 3. Unmask and classify related issues. Statistical analyses were also conducted and explained per industry9 to satisfy the expectations of the
participants (Appendix H)10. The summarised findings of the conducted focus-group interview
concerning the individual stages and activities related to Hypothesis 1 and 2 can be found in Appendix I. Overall, participants largely agreed and validated the results. For instance, executing the activities in the testing stage was only found significant for exploitative innovations (H1d) which is confirmed by the
participants since more radical innovations experience a merged final stage. As hypothesised, activities from a firms’ testing stage are simultaneously pursued as activities from firms’ launching stage in case of explorative innovations. This finding suits the significant interaction effect EXPLOR (P<0.1) of those two stages in Table 7. Another example included the technical development stage, since it was only found significant for explorative innovation (H1c). The participants largely recognise the importance of
this step in today’s ‘online world’ and concluded this stage is equally important for both type of
8 Addition analyses has been conducted to find other interaction effects among all independent and control variables.
Concerning EXPLOR three interaction effects were found: TD*Size (P<0.1), T*Size (P<0.1), and BAMS*Size (P<0.1). Concerning EXPLOIT only one interaction effect was found: IG*SL (P<0.05)
9 Industry specific analyses were discussed in the focus-group interview, but excluded from the result section of this study,
since it does not fit the scope of this studies’ hypotheses.
10 In these analyses, all variables – prior PCA - were used since the three subsamples were too small to conduct a factor analysis.
innovations. Due to the differences in terms of age, size and formality it remained unclear why the control variables showed no significant relationships in both exploitative and explorative innovation. Overall, managers in this dynamic business environment recognise the importance of following formal NSD stages in order to be successful. Despite small differences in the execution of the specific NSD activities and the different proposed NSD frameworks, the participants follow more or less the same stages which validates the results of Hypothesis 3.
5.5 Summary chapter
The results obtained by statistical analyses are summarised in Table 12. The focus-group interview added the importance of the technical development stage (H1c) in terms of exploitative innovations since
this is a stage were most changes occur and time-issues often create workload pressures. Moreover, the technical part of innovation is sometimes outsourced and overall the participants agreed on following formal stages to manage this stage in both type of innovations.
Table 11
Hypothesis confirmation and rejection
Hypothesis Confirmed Rejected
EXPLOR EXPLOIT EXPLOR EXPLOIT
1 H1a X X H1b X X H1c X F* X H1d X X H1e X X 2 H2 X 3 H3a X X H3v X X *F = Focus-group
6. Discussion
The purpose of this study was to verify if the NSD process in Dutch financial service institutions should be followed in a structured manner or if deviations in the process were allowed. Despite the scarce literature field of NSD, linear innovation process was accustomed which guided as a baseline for the study at hand. This NSD process was carefully reviewed and reconsidered from two theoretical perspectives in relationship with this studies’ results.
(2006) who concluded that this stage dictates all further stages of an innovation process and is therefore very crucial. Moreover, participants from the focus-group acknowledged the importance of this stage and found that systematically searching for and screening ideas is not the hardest part compared to the other stages. Furthermore, two independent variables are positively and significantly related to firms’ explorative new service innovation performance, namely business analysis and marketing strategy and technical development. These findings confirm the notion of Papastathopoulou et al. (2001) who conclude that the attractiveness of radically new ideas have to be examined in the market and within the capabilities of the firm. Furthermore, it suits the characteristics of explorative innovations like continuous interaction with the environment (Lievens, 2000). From a theoretical point of view, it seems that structured stages are needed to execute a full-scale introduction of the service to the market and the evaluation of its performance (Avlonitis et al., 2001). In line with the reasoning the participants of the focus-group interview, the importance of the technical development in explorative innovations has to be recognised as well. According to Melton and Hartline (2010) it is all about the combination of developing and operationally testing the core service which seems important in both type of innovations. The activities from the testing stage are positively related to exploitative new service innovation performance. This argues against the reasoning of Oke (2007) who concluded that incremental innovations do not have to be managed according to a structured NSD process. However, the results showed that following the activities from the testing stage in terms of more radical innovations are not supported which is in line with the arguments defined in the focus-group interview. Furthermore, it suits the line of reasoning behind Hypothesis 2 and confirms the notion of Aboulnasr et al. (2008) who states that any information shared prior launching can be a signal for competitors. Hence, negatively influences firms’ new service innovation performance. The final stage, service launch, is not significant and this is an unexpected finding, since this stage actually consists of critical tasks and are very challenging (Xie, et al, 2008). This can be explained by multiple reasons. Firstly, in line with the reasoning of Hypothesis 2, it could be argued that service launch activities are merged with testing activities concerning explorative innovation. Furthermore, in terms of more radical innovations the notion of Xie et al. (2008) makes sense because the positioning of a new offering experiences a highly experiential nature. Third, this stage involves, next to the market introduction, an evaluation of services’ performance in the market (Avlonitis et al., 2001). The performance scales used in this study, cannot be influenced by the evaluations after launching. This might be an interesting item starting-point for future research (e.g. different performance measurements). The Process Theory is successfully applied to the NSD stages and these results significantly adds to the scarce NSD literature concerning ‘process and stages’ as mentioned by Papastathopoulou & Hultink (2012). It improves the understanding of this theory used in practice.
deviations made from structured NSD processes, the worse firms’ new service innovation performance. Acknowledging (and confirmed by the focus-group interview) that small deviations from the formal process can be made due to firms’ specific characteristics, following the main steps of the NSD stages are beneficial. Respectively, too much deviations will harm the performance. This study confirms the notion of De Brentani (1989), de Brentani and Ragot (1996) and Froehle et al. (2000) who showed that following a formal process of NSD is equally important to the role that the formal development process plays in NPD. Regarding the hypothesis whether deviations from the ‘expert-profile’ were negatively associated with firms’ new service innovation performance, partial support could be found for explorative new service innovation performance. This could be explained by the method chosen concerning the ‘expert’ profile concerning the measurement chosen (percentage of new sales). Furthermore the ten ‘experts’ chosen were, after the profile was set, deleted from the database. This resulted in a sample of 104 instead of 114 respondents and could also be a reason of the non-significant relationships. Nevertheless, this study significantly add to the practical use of the Configuration Theory and in combination with the scarce NSD literature.
Overall, no significant relationships between the dependent- and independent variables with the control variables could be found. This was discussed extensively by the focus-group. The participants recognised the importance of firms’ specific characteristics (e.g. size and formality) in order to structured follow NSD stages. For instance, size and formality were recognised by the participants to correlate (supported by the correlations of Table 6). Despite the correlations, no significant relationship were found with the dependent- and independent variables and no conclusions could be drawn.
6.1 Summary chapter
Following the specific activities of the NSD process can positively influence the innovation performance depending on the type of innovation that a firm wants to pursue. Deviations from such a process can occur due to the uniqueness of each firm. However, too much deviation from the original process will harm the firms’ performance. This supports the Process Theory in which constant necessary conditions must be reached and it extends the relevance of the role the Configuration Theory in formal NSD.
