Thriving with innovation: Maximizing knowledge acquisition from customers
Isak Edlund Sara Larsson
Industrial and Management Engineering, master's level 2018
Luleå University of Technology
Department of Business Administration, Technology and Social Sciences
ABSTRACT
Purpose – The purpose of this study is to advance the understanding of how choices regarding the combination of type of customer and knowledge acquisition technique influences the characteristics of the knowledge that can be acquired, in the early phase of the innovation process.
Method – Data was collected through a single case study, varying the type of customer (lead users and non-lead users) and knowledge acquisition technique (survey, interviews and focus groups) in order to create six different combinations. The data was coded using content analysis, after which the codes were quantified by expert scoring on three different characteristics of knowledge (novelty, relevance and feasibility). Calculations were made, highlighting differences between the combinations, verified by t-Tests.
Findings – The findings show that data collected through interviews and focus groups among lead users score significantly higher on all characteristics of knowledge than all non-lead user interactions. Comparing surveys, knowledge from lead users only score significantly higher on relevance and feasibility. Within the two customer types for all knowledge characteristics, varying only the type of customer, interviews and focus groups score significantly higher than surveys.
Theoretical implications – This study contributes to the literature by deepening the understanding within the fields of knowledge acquisition and customer involvement.
Our findings challenge the unilateral view of customer involvement versus no customer involvement (Christensen and Bower, 1996; Ulwick 2002; Gemser and Perks, 2015) by providing a more nuanced picture, taking the process itself into consideration.
Practical implications – This study highlight important factors for effective knowledge acquisition from customers, stating that companies should (1) categorize the customers, and (2) conduct focus groups and interviews with customers categorized as lead users, independent whether incremental or radical innovation is desired.
Keywords – customer involvement, knowledge acquisition techniques, types of customers, knowledge characteristics, innovation.
ACKNOWLEDGMENTS
This master thesis is the final course in our master program Industrial Engineering and management with specialization within innovation and strategic business development at Luleå University of Technology (LTU). The master thesis has been carried out during the spring of 2018 in association with a startup company in the United Kingdom.
We would like to thank everyone who has supported us during the course of work.
Firstly, we would like to direct a special thanks to Sara Thorgren, our supervisor at the university, for support and valuable input throughout the whole period. Secondly, we would like to thank our supervisor at the case company, Niklas Fallsjö, who has done a great job guiding and supporting us throughout the work with this thesis. Finally, we would like to direct a big thank you to our wonderful colleagues, both students at LTU and employees at the case company, who have given their input and insight into our project, supporting the progress from beginning to end.
A special thank you also goes to family and friends who have supported us during all our academic years.
______________________________ ______________________________
Isak Edlund Sara Larsson Luleå, June 2018
TABLE OF CONTENT
1. INTRODUCTION ... 1
2. KNOWLEDGE ACQUISITION FROM CUSTOMERS ... 5
2.1 Characteristics of knowledge ... 5
2.1.1 Relevance of knowledge ... 5
2.1.2 Novelty of knowledge ... 6
2.1.3 Feasibility of knowledge ... 6
2.1.4 Benefits of diverse knowledge characteristics ... 7
2.2 Types of customers... 7
2.2.1 Lead users... 8
2.2.2 Non-lead users ... 9
2.3 Knowledge acquisition techniques ... 9
2.3.1 Surveys... 9
2.3.2 Semi-structured interviews ... 10
2.3.3 Focus groups ... 10
2.3.4 Comparisons among the techniques ... 11
2.4 Connection between theory and research questions ... 12
3. METHOD ... 13
3.1 Data collection ... 13
3.1.1 Knowledge collected through surveys ... 14
3.1.2 Knowledge collected through interviews ... 15
3.1.3 Knowledge collected through focus groups ... 15
3.1.4 Customer categorization ... 16
3.2 Data analysis ... 18
3.2.1 Knowledge coded through content analysis ... 18
3.2.2 Expert scoring of collected knowledge ... 19
3.2.3 Calculations for data comparison purposes ... 20
3.2.4 Time spent throughout the study ... 22
3.3 Quality improvement measures... 22
4. FINDINGS ... 25
4.1 The combinations’ influence on knowledge characteristics ... 25
4.1.1 Novelty score ... 27
4.1.2 Relevance score ... 28
4.1.3 Feasibility score ... 29
4.1.4 Total score ... 30
4.2 Cost approximations ... 31
4.2.1 Costs associated with the categorization of customers ... 31
4.2.2 Cost associated with interaction techniques ... 32
5. DISCUSSION AND CONCLUSIONS ... 34
5.1 How the combination of type of customer and knowledge acquisition technique affect characteristics of knowledge acquired ... 34
5.2 Taking cost and innovation outcomes into consideration ... 35
5.3 Theoretical contribution ... 37
5.4 Practical contribution ... 38
5.5 Limitations and future research ... 39
5.6 Conclusions ... 40
6. REFERENCES ... 41 Appendix A – Pilot survey ... I Appendix B – Knowledge survey ... IV Appendix C – Sampled knowledge survey informants ... VII Appendix D – Semi-structured interview survey ... IX Appendix E – Semi-structured interview guide ... XII Appendix F – Sampled semi-structured interview informants ... XIII Appendix G – Focus group survey ... XIV Appendix H – Focus group guide ... XVI Appendix I – Sampled focus group informants ... XVIII Appendix J – Expert interview informants ... XIX
1 1. INTRODUCTION
Knowledge acquisition from customers, referring to the gathering of information from customers through different types of media and techniques (Zhang, Zhao, Lyles and Guo, 2015; Yli-Renko, Autio, & Sapienza, 2001), is by many considered important for innovation (Soo, Devinney, & Midgley, 2007). Customers’ important role as providers of knowledge into companies’ innovation processes is since long well established (Rothwell, Freeman, Horsley, Jervis, Robertson & Townsend, 1974). Several studies have also found such contribution of knowledge to be positively correlated with innovation and market success (e.g. Li, Kankanhalli & Hyun Kim, 2016; Mahr, Lievens
& Blazevic, 2014; Laine, 2012; Poetz & Schreier, 2012; Fang, Palmatier & Evans, 2008;
von Hippel, 1998), often resulting in the most important innovations for companies (Franke, Schirg & Reinsberger, 2016). Studies have found this true both for radical innovation outcomes, referring to innovations new to both the company and the market;
as well as incremental innovation outcomes, referring to innovation only new to the company (Laursen, 2011). From the customers’ point of view, a feeling of accomplishment and access to better products are highlighted as the motivators for sharing knowledge (Gemser & Perks, 2015). Hence, customer involvement has the potential to create a win-win situation for both the firm and the customers (Gemser & Perks, 2015;
Mahr et al., 2014).
Previous research is, however, not in complete consensus regarding the value of customer involvement in the innovation process. The famous words uttered by Henry Ford: “If I had asked people what they wanted, they would have said faster horses” marks the battle line between those who believe customer input to be crucial for innovation and those who argue that true innovation is created by singularly gifted visionaries, without customer input (Vlaskovits, 2011). While researchers such as Gemser and Perks (2015) and Enkel, Perez-Freije and Gassmann (2005) conclude that customers are one of the main sources of input to both radical and incremental innovation, other researchers, such as Christensen and Bower (1996), and Ulwick (2002), argue that customers may actually lead companies on the wrong track. One part of the research community state that
2
companies need to take the customers into consideration whilst innovating to be competitive in today's business landscape; explaining potential problems arising as shortcomings that can be mitigated by developing a well-functioning process of gathering information from customers (Martelo-Landroguez, Albort-Morant, Leal-Rodríguez and Ribeiro-Soriano, 2018; Santos-Vijande, López-Sánchez, & Pascual-Fernández, 2018;
Gemser and Perks, 2015). On the other side of the battle line, researchers state that customers are not informed enough to contribute to the innovation process, making most of the customers useful only for confirming or indicative purposes (Carbonell, Rodríguez-Escudero, & Pujari, 2009; Ulwick, 2002). Some researchers have even gone so far as to conclude that customer involvement may negatively impact product outcomes (Knudsen, 2007); especially if the type of customer is not considered (Roy, 2018; Lüthje and Herstatt, 2004).
The conflicting findings suggest that there is great potential value in involving customers in the innovation process; but that it is hard to practically achieve and if done incorrectly the effect can even be negative. Further, the need to investigate the cost-benefit ratio of using different types of knowledge acquisition techniques has recently been emphasized (e.g. Guest, Namey, Taylor, Eley, & McKenna, 2017; Namey, Guest, McKenna, &
Chen, 2016). Thusly, there is a need to further investigate the effects regarding choices on how to acquire knowledge from customers, but also from which customers the knowledge is acquired.
Knowledge is often acquired through techniques such as surveys, interviews or focus groups, typically in the early phases of the innovation process, referring to the ideation and concept creation phases, as this is where companies recognize the biggest need to involve their customers (Bosch-Sijtsema & Bosch, 2014; Fabijan, Olsson & Bosch, 2015;
Stenmark, Tinnsten & Wiklund, 2011). Although many companies are expressing a desire to involve customers more in their innovation process (Stenmark et al., 2011), they are often struggling to extract value from doing so (Li, et al., 2016; Fabijan et al., 2015).
Scholars have expressed a need for further research on the usage and success of customer involvement in innovation processes and, in particular, how and when to use different
3
knowledge acquisition techniques in those processes (Janssen & Dankbaar, 2008). This, since studies have found that companies tend to use these techniques on an ad hoc basis, not considering how and when to use what technique (Bosch-Sijtsema & Bosch, 2014).
When customers are involved, studies have found that different types of customers (e.g.
lead users and non-lead users) contribute with different types of knowledge; separated by the different characteristics of the knowledge (e.g. novelty, relevance and feasibility)(
(Schuurman, Mahr, & De Marez, 2011; Mahr et al., 2014). The characteristics are in turn associated with different innovation outcomes (e.g. incremental and radical) (Mahr et al., 2014), thus making it highly relevant for companies to acquire appropriate knowledge depending on their desired use of the knowledge. Companies tend to involve a relatively small number of customers in the early phases of the innovation process; both due to resource restrictions (Bosch-Sijtsema & Bosch, 2014) and because involving more customers does not necessarily increase the quality of the knowledge acquired (Leitner, Warnke & Rhomberg, 2016). This highlights the importance of deliberately choosing which types customers to involve, depending on if radical or incremental innovation is desired.
By failing to involve the right kind of customers and using the right kind of technique, companies risk to prioritize the wrong projects and develop products that the customers do not want (Yaman, Sauvola, Riungu-Kalliosaari, Hokkanen, Kuvaja, Olivo &
Männistö, 2016; Fabijan et al., 2015). This implies a need to investigate the involvement of different types of customers through different types of knowledge acquisition techniques for innovation purposes. With the lack of research on how the combination of different knowledge acquisition techniques and how different types of customers affect the characteristics of knowledge that can be acquired, companies are left fumbling in the dark when it comes to practically achieving their desired innovation outcomes. By providing companies with more insight into how they can tailor their customer involvement, they would be able to more effectively and efficiently acquire appropriate knowledge.
4
Springing from the above, the purpose of this study is to advance the understanding of how choices regarding the combination of type of customer and knowledge acquisition technique influence the characteristics of the knowledge that can be acquired, in the early phase of the innovation process. Specifically, the present study addresses the following questions:
RQ1: How does the combination of knowledge acquisition technique and the type of customer influence the characteristics of the knowledge that can be acquired?
RQ2: For what innovation outcomes, radical or incremental, would each combination of knowledge acquisition technique and type of customer be most effective in terms of cost of acquiring the knowledge?
The research questions will be answered through a single case study conducted at a software company based in the United Kingdom (UK). The study will contribute to the literature by extending the understanding for knowledge acquisition techniques, different types of customers, and how different combinations of the two influence the characteristics of the knowledge that can be acquired in innovation processes. This contribution will also have practical implications for companies trying to involve customers in their innovation process, enabling them to tailor their customer involvement process based on if radical or incremental innovation outcomes are desired;
enhancing both the effectiveness and the efficiency of their innovation process.
5
2. KNOWLEDGE ACQUISITION FROM CUSTOMERS
In conducting this study, testing different knowledge acquisition techniques on different types of customers, we were informed by several literatures. First, literature on knowledge characteristics (Mahr et al., 2014; Poetz & Schreier, 2012), suggests that the understanding of knowledge acquired is enhanced if it is viewed through its relevance, novelty and feasibility. Second, given that knowledge acquisition may vary depending on the type of customers, the lead user literature (Schuurman, et al., 2011; von Hippel, 1976) informed us what distinguishes different types of customers from each other, and how that can be operationalized methodologically. Third, and which was critical to create a solid study design for testing different types of knowledge acquisition techniques against each other, we reviewed the literature on knowledge acquisitions techniques (Guest et al., 2017;
Flynn, Sakakibara, Schroeder, Bates, & Flynn, 1990). After reviewing each of these areas we explain how their integration provides a platform for how we designed and conducted our empirical study.
2.1 Characteristics of knowledge
This section focuses on three characteristics of knowledge, namely: (1) relevance of knowledge, closely linked to incremental innovation outcomes (Mahr et al., 2014); (2) novelty of knowledge, closely linked to radical innovation outcomes (Mahr et al., 2014);
and (3) feasibility (Poetz & Schreier, 2012), closely linked to the ability of utilizing the knowledge acquired. The characteristics are further explained in the following sections.
2.1.1 Relevance of knowledge
Mahr et al. (2014, p.602) defines relevance of knowledge as “knowledge that is appropriate for the particular project, provides details related to the tasks, and is easy to understand and implement”. As the definition indicates, relevant knowledge is particularly useful for exploitative activities, such as implementation and refinement (Mahr et al., 2014), and in doing so closely corresponds to incremental innovation outcomes. Further, highly relevant knowledge does not necessarily bring value in itself,
6
if the receiver was already aware of benefits that the knowledge brings (Im & Workman, 2004).
2.1.2 Novelty of knowledge
Not completely opposite, but very different and in many ways complementary to relevance of knowledge, is novelty of knowledge. Mahr et al. (2014, p.602) define novelty of knowledge as “knowledge that provides new insights, unique inspirations, and a broad range of opinions and ideas”. Novelty links closely to experimentation and discovery, which closely corresponds to radical innovation outcomes (Mahr et al., 2014).
Even though novel knowledge is generally harder to acquire than less original and ground-breaking knowledge, it is highly important for the success of a company (Poetz
& Schreier, 2012). Novel knowledge produces superior and distinctive product features that increases the likelihood of customer retention and adoption (Im & Workman, 2004).
However, important to note is that there is evidence that a too high level of novelty might cause customers to resist adoption due to difficulties with understanding the new functionality, resulting in customer dissatisfaction (Mahr et al., 2014). Combining a high level of both novelty and relevance of knowledge accumulates to knowledge that has high potential benefits (Im & Workman, 2004). However, the real benefits are still small if the knowledge cannot be realized.
2.1.3 Feasibility of knowledge
Springing from Poetz’s and Schreier’s (2012, p.250) definition of feasibility of an idea, the present study defines feasibility of knowledge as “how easily it [the knowledge] could be translated into a commercial product”. Poetz and Schreier (2012) highlight that both technical and economic costs should be considered. It is important to note that the cost referred to relates to realizing the knowledge, not acquiring it (Poetz & Schreier, 2012).
Thus, a high level of feasibility (for example through low costs of commercialization) is a benefit of the knowledge acquired, regardless of the resources spent to acquire the knowledge.
7
2.1.4 Benefits of diverse knowledge characteristics
As the three different characteristics described above are complementary, ideal knowledge would score high on relevance, novelty and feasibility. Studies have, however, found trade-offs between the characteristics, where relevance and novelty are negatively correlated with feasibility (Kristensson, Gustafsson & Archer 2004). In addition, no real value can be extracted from the knowledge if one of the characteristics is completely absent; e.g. knowledge with high novelty and feasibility but which lack relevance would by definition not be useful (Mahr et al., 2014). Thus, an adequate measurement of the overall benefit would be the product of the three characteristics (e.g. Poetz & Schreier, 2012), since this calculation would award the high scoring characteristics, but also account for very low values on one of the characteristics. Important to keep in mind, however, is that depending on the desired innovation outcome, the overall benefit score might be misleading due to the characteristics close correlation to innovation outcomes (Mahr et al., 2014). For example, if incremental innovation is desired, high relevance and feasibility would be valuable as long as the knowledge also holds some novelty to the company (Mahr et al., 2014).
2.2 Types of customers
The most known concept of customer involvement in innovation processes is the lead user involvement, a concept that origins from von Hippel (1976). The literature surrounding lead user innovation connects different types of customers to different aspects of knowledge that can be generated throughout the innovation process (von Hippel, 1976). While different types of customers may be useful for acquiring different types of knowledge, there is also a cost associated with segmenting and finding specific customer types (von Hippel, 1976). Therefore, it is important to carefully consider the choice of customers to involve and how to find them (von Hippel, 2005).
In the literature, various ways of categorizing customers into segments exist, such as lead users and non-lead users (e.g. Magnusson, 2009; Hoffman, Kopalle, & Novak, 2010). In the present study, we follow the categorization by Schuurman et al. (2011), as they
8
provide a solid foundation and explanation of customer types, when distinguishing between lead users and non-lead users.
2.2.1 Lead users
Lead users, defined as “users who experience a need for a given innovation earlier than the majority of the target market” (Morrison, Roberts, & Midgley, 2004, p.352), are highly important since they contribute to a proactive approach to innovation (Morrison et al., 2004; Schuurman et al., 2011). This makes them desirable to interact with during innovation processes since their knowledge, needs and wants, tend to pave the way for successful innovation (Morrison et al., 2004; von Hippel, 2005). Scholars have found that lead users provide great and novel product ideas, often better than other types of customers (Hoffman et al., 2010).
Lead users are often fewer than non-lead users, thus, making them harder to identify (von Hippel, 1998; Lüthje & Herstatt, 2004; Schuurman et al., 2011). In order to be able to identify lead users, it is important to know their specific characteristics (von Hippel, Franke & Prugl, 2006). Von Hippel (1988) found two characteristics distinguishing lead users. Firstly, their enhanced capability to innovate as they are ahead of the market and secondly, their strong motivation to innovate sprung from a high level of expected gain.
Lüthje and Herstatt (2004) provide some further guidelines for determining which users that are lead users. One should first determine which major trends one predicts for the market in the future, and then using that to identify users who take a lead in that direction (Lüthje and Herstatt, 2004). Furthermore, the dissatisfaction is highly important, as lead users are typically not satisfied with the current offerings (Lüthje and Herstatt, 2004).
According to Lüthje and Herstatt (2004, p.563); “The process of searching for Lead Users is a creative one that must be tailored to the specific conditions of the relevant search field”. One of the most used methods for identification of lead users is screening, where nearly all users are being evaluated in relation to lead user characteristics (von Hippel et al., 2006). This method is especially favorable when no lead users are known from the start (von Hippel et al., 2006). Screening is often carried out in a sequence of steps,
9
starting with quantitative data (e.g. surveys), followed by more qualitative assessments as the number of users become narrower (von Hippel et al., 2006).
2.2.2 Non-lead users
While non-lead users can be categorized further into different types, they share many of the basic characteristics (Schuurman et al., 2011). For example, these users are normally making up the majority of the market, following a relatively standardized pattern of usage and are rarely dissatisfied with the market (Magnusson, 2009; Kristensson & Magnusson, 2010; Schuurman et al., 2011). Some of them are characterized by advanced usage and some, due to their habits of usage, tend to put new innovations "to the limit" (Schuurman et al., 2011). Non-lead users are mainly associated with less novel knowledge, product evaluation and incremental innovation (Ozer, 2009; Rothwell, 1992; Voss, 1985).
However, some studies have also been able to connect knowledge from non-lead users to radical innovation (Magnusson, 2009).
2.3 Knowledge acquisition techniques
Surveys, semi-structured interviews and focus groups are the techniques most commonly used when trying to acquire knowledge from customers (Roach, 2007; Wackerbarth, Streams, & Smith, 2002; Seaman, 1999; De, Mellenbergh, & Hox, 1996; Flynn et al., 1990). These techniques are used in a variety of different compositions, both isolated and in combination (Wackerbarth et al., 2001; Morgan 1996; Flynn et al., 1990). While the research is extensive, researchers have not yet gained a common understanding regarding what techniques are the most fruitful, both in terms of benefits and limitations, (Guest et al., 2017; Wackerbarth et al., 2002). It is, however, possible to distinguish characteristics for each technique, presented below.
2.3.1 Surveys
Surveys, referring to a knowledge acquisition technique that gathers information from customers through standardized forms, most commonly distributed to a large population without any personal interaction (Flynn, Pagell, & Fugate, 2018; Kerlinger and Lee, 2000; De, Mellenbergh, & Hox, 1996; Flynn et al., 1990), is often referred to as the most
10
widely used technique for information gathering and opinion seeking (Flynn et al., 1990). Generalizability and theory testing are often the primary goals when conducting survey research (Krause, Luzzini, & Lawson, 2018). Roach (2007) concludes that surveys are a quick and cost-effective way of generating a lot of less in-depth knowledge and that they tend to reveal information and knowledge that informants would not feel comfortable sharing face-to-face. Researchers agree that surveys are the least time-consuming way of gathering information from customers (Flynn et al., 1990; Roach, 2007). Surveys are often complemented with more in-depth information from other knowledge acquisition techniques, when used for innovation purposes (Flynn et al., 1990). Studies have shown that surveys may deliver novel results but highlight that careful consideration in respect to the context in which they are used is required (Flynn et al., 2018).
2.3.2 Semi-structured interviews
Semi-structured interviews are referred to as a knowledge acquisition technique that builds on predefined questions but also allows the researcher to ask follow-up questions based on the informants’ answers (Wackerbarth et al. 2002; Flynn et al., 1990). The technique is popular when trying to gain an in-depth understanding of customers’ needs throughout the innovation process (Flynn et al., 1990). This since it gives room both to investigate details of a topic and to be flexible, ready to hear something unexpected (Seaman, 1999). Many researchers agree that semi-structured interviews can produce a great deal of detail and have the ability to extract more details into a respondent’s personal thoughts, feelings and views (Roach, 2007; Morgan, 1998; Knodel, 1993). The technique is, however, often perceived as time consuming and thus costly (Seaman, 1999).
2.3.3 Focus groups
Focus groups, referring to a knowledge acquisition technique that, through group interaction, enables collection of data on a topic defined by a researcher (Guest et al., 2017; Morgan, 1996), are used to gather data from a group of customers with the help and monitoring of a moderator (Morgan, 1996). It is a technique actively encouraging an interactive discussion between the participants (Morgan, 1996). The technique is
11
widely used and has increased its presence in innovation processes the last decade (Guest et al., 2017). There is, however, little research conducted on how fruitful the method is in terms of delivering knowledge valuable in an innovation context (Guest et al., 2017).
According to Wackerbarth et al. (2002), focus groups generate a variety of advantages, such as being less time consuming and enabling the opportunity to evaluate one person's thoughts against a larger group. Further, focus groups tend to generate a wider range of views and ideas than one-to-one methods (Krueger & Casey, 2015; Kidd & Parshall, 2000; Robinson, 1999; Kitzinger, 1994). In general, the larger part of the cost associated with conducting focus groups is the time allocated to the administration and planning prior to the focus group itself (Guest et al., 2017).
2.3.4 Comparisons among the techniques
Wackerbarth et al. (2002) compare focus groups to surveys and interviews, concluding that focus groups have to be managed properly. Failing to do so can limit the comprehensiveness of answers given and affect novelty and in-depth understanding of answers and opinions shared. Semi-structured interviews are, on the other hand, said to be much more time consuming than surveys and focus groups but has a tendency to generate both in-depth and unexpected knowledge (Seaman, 1999). In cohesion with above sections, Stokes and Bergin (2006) states that interviews provided greater depth and detail, while focus groups offered greater width and context. While many researchers struggle to distinguish the differences between the knowledge generated from interviews compared to focus groups, there is a consensus that surveys generally contribute with a large amount of less novel information (Guest et al., 2017).
The time spent to conduct the full process of each technique is highly important to measure since it is one of the main drivers of the costs associated with each technique (Namey et al., 2016). Thus, the costs must be set in relation to the value of the knowledge acquired in order to determine which technique is most suiting to use (Namey et al., 2016). Even though researchers agree that surveys are the least time consuming, opinions differ when it comes to focus groups and interviews. Many agree that conduction of focus groups is less time consuming than for interviews, but some argue that the far more time-
12
consuming administration required prior to focus groups balances out the less time- consuming conduction (Guest et al., 2017).
2.4 Connection between theory and research questions
As a foundation for our study, literature has provided the basis for used techniques, categorization, interaction with customers and different types of knowledge important in the context of innovation. Springing from the research purpose, different types of customers as well as different knowledge acquisition techniques are assumed to be associated with different advantages when it comes to collecting knowledge, but also different costs. We are informed that value of knowledge acquired is a function of the benefits from characteristics of the knowledge in relation to the cost to acquire that knowledge (Namey et al., 2016; Bailey & Pearson, 1983). As proven in the literature, the cost of innovation is often related to finding the right kind of customer (von Hippel et al., 2006) and, subsequently, using the right technique for knowledge acquisition (Flynn et al., 1990) in order to extract the desired knowledge from those customers. In order to assess perceived utility (i.e. knowledge value), referring to “the relative balance between costs and considered usefulness” (Bailey & Pearson, 1983, p.542), the benefits are compared to the costs of acquiring the knowledge. Originating from literature, this study will therefore investigate different combinations of the most commonly used techniques (surveys, interviews and focus groups) of knowledge acquisition (e.g. Roach, 2007; Wackerbarth et al., 2002), incorporating two different types of customers, lead users and non-lead users (e.g. Schuurman et al., 2011; von Hippel, 1976). This will lay the foundation for investigation of the cost-benefit ratio associated to the combination of different techniques and customers in relation to the characteristics of knowledge, novelty, relevance and feasibility (Mahr et al., 2014; Poetz & Schreier, 2012). Further, this will contribute to the overarching purpose of advancing the understanding of how choices regarding knowledge acquisition techniques and different types of customers in the early phase of the innovation process influence the characteristics of the knowledge that can be acquired.
13 3. METHOD
To answer the questions of (1) how the combination of knowledge acquisition technique and the type of customer influence the characteristics of the knowledge that can be acquired, and (2) for what innovation outcomes, radical or incremental, each combination of knowledge acquisition technique and type of customer would be most effective in terms of cost of acquiring the knowledge, a case study was conducted. The case company was chosen since it had been struggling with both effectiveness and efficiency in their customer involvement efforts, and therefore, was representing a good case regarding the identified problems.Below is a short description of the case company.
The case company is a fast growing, software-focused gambling company currently based in, and mainly operating in, the UK. The company has since its launch in 2016 been growing from around 10 to 50 employees, and is still growing quickly, both in terms of employees and customers. The industry is highly competitive and the company’s main focus is to bring new innovation into an industry that according to them “[...] has been snoozing for a while” (Personal communication, Head of Sportsbook). This makes their customers an essential source of information in order to ensure the effectiveness of their innovation efforts, making sure they are focusing on the right things; as well as their efficiency when involving customers, making sure they are doing it in the right way. The case company also has over 18,000 active customers, providing a stable base for the study.
3.1 Data collection
The study started in January 2018 and consisted of survey answers, semi-structured interviews and discussions during focus groups. Throughout the study, the time spent was documented in order to assess the costs associated with performed activities. Surveys were sent to all participants, of which one was used for collecting knowledge in order to evaluate the technique; while the remaining surveys were used for connecting with people who would be willing to participate in interviews or focus groups and categorizing them as lead users or non-lead users. In total, 40 survey answers were used for knowledge purposes, 18 semi-structured phone interviews were held and two focus groups were
14
conducted (six participants in each). All collected data was treated confidentially and the customers were therefore defined as codes rather than their identity during the processing and analyses of data. Since the product of the studied company is rapidly evolving, it was important that the participants were up-to-date with the current product. Hence, the population was defined as all active customers, referring to customers who had used the product in the last ten days. Customer data on activity and proximity to company office were extracted through the company’s Customer Relationship Management (CRM) tool. An overview of the sampling used for data collection is presented in Figure 1.
Figure 1 - Overview of sampling and data collection.
3.1.1 Knowledge collected through surveys
To collect knowledge through the survey technique, an internet-mediated self- administered survey, including both closed and open-ended questions (Saunders, Lewis
& Thornhill, 2009) was carried out. The survey questions were developed in iterations
15
with the company, and improvements were made based on a pilot study sent to 1,013 customers (Appendix A). The final survey was sent out to 4,407 respondents (Appendix B). A reminder email was sent to participants who had not opened the first email in 5 days. Because of the need of physical presence for focus group data collection, only customers not close to the company’s office were sampled for participation in this survey, using random sampling from that strata. Results for further analysis were selected using another round of stratified random sampling, with strata defined as lead users and non- lead users (based on customer categorization questions in the survey). The limiting stratum was lead users with a total of 27 customers, from which 20 were randomly sampled. The same number of customers identified as non-lead users were randomly sampled, resulting in a total of 40 survey answers selected for analysis.
3.1.2 Knowledge collected through interviews
To identify customers to be interviewed, a survey was sent to 4,544 respondents (Appendix D). A reminder email was sent to participants who did not open the first email in 5 days. The sample was selected in the same way as those targeted for the knowledge survey, with the exception that customers who were selected for the knowledge survey were excluded. All customers completing the survey with self-selection represented the new sampling frame, further divided into the two strata lead users and non-lead users, based on categorization questions in the survey. With 18 customers identified as lead users, the limiting factor for the number of customers to sample from each stratum was time spent conducting the interviews. Nine customers from each stratum were selected using random sampling, resulting in a total of 18 semi-structured telephone interviews conducted (Appendix E) (Saunders et al., 2009); each ranging between 20 to 30 minutes.
3.1.3 Knowledge collected through focus groups
To identify customers for participation in focus groups, a survey was sent to 675 respondents (Appendix G). A reminder email was sent to participants who did not open the first email in 5 days. The sample was selected in a similar way as for the one targeting survey and interview participants but selected from the close to company’s office stratum.
All customers completing the survey with self-selection represented the new sampling
16
frame, further divided into the two strata: lead users and non-lead users, based on categorization questions in the survey. A total of nine customers were categorized into the lead user stratum, of whom six were able to participate, making six the limiting factor for each group in the focus group. From the non-lead user stratum, random sampling was used to select another six customers. All the selected customers were grouped based on user type, one group for each type, to participate in their user-specific focus group interview (Appendix H) (Carson, Gilmore, Perry & Grønhaug, 2001). This resulting in the conduction of two six-person focus groups, each around 60 minutes long.
3.1.4 Customer categorization
As previously described, participants targeted for the knowledge survey, the interviews, and the focus groups all answered the exact same questions for categorization into one of two strata; lead users and non-lead users. The screening followed a two-step process as described below, consistent with what is described by von Hippel et al. (2006). Table 2 defines how the participants’ answers were translated into values, used for categorization calculations.
Table 1 - Mapping answers to categorization questions into values for calculations.
17 Step one:
- Using the values associated with questions one, three, four and five to calculate the initial lead user score, using the following formula:
!"[$%&'(] ∗ ( ( 1 − !/[$%&'(] ) + ( !2[$%&'(] − 1 ) ∗ ( −6 ) ) + ( !4[$%&'(] ) )
- Removing initial lead user scores lower than 9.
Step two:
- Qualitative assessment of Q2 answers by letting experts rank the answers on a scale from 1 (low) to 6 (high), based on how well the answers align with what the opinion of the expert.
- Removing customers with expert scores lower than 3.
- Adding the expert score to the initial lead user score, then removing scores lower than 15.
- Remaining customers are lead users, all other customers, removed in any step of the process, are non-lead users.
The first step sorted out any customers who said they did not have any ideas, thus non- lead users. It gave a higher score if the customer was unsatisfied with the market, had ideas that were new to the entire market or had ideas that they felt would have a big impact on customers’ gambling experience; all of which represent lead user characteristics. The choice to draw the line at 9 is sprung from 9 representing 75 % of the maximum score possible (12), ensuring that the customers show promising lead user potential but not removing too many in the first step. In the second step, customers who provided qualitative answers to Q2 that aligned with the opinion of industry experts were given a higher score, since this is a lead user characteristic. Removing customers with expert scores lower than 3 removed incomplete answers to Q2, but also put a large weight on the qualitative assessment, which has been found to be an important part of assessing lead users. The choice to remove scores lower than 15 after addition with the expert score was made due to its synergy with the first step. Users with a low initial lead user score could only be categorized as lead users by getting a high expert score, and vice versa. For example, a customer coming into step two with the lowest possible score of 9, had to receive the maximum expert score of 6 to be categorized as a lead user.
Through this screening method, 27 customers were categorized into the lead user stratum for the knowledge survey (of whom 20 were analyzed); 18 customers were categorized into the lead user stratum for the interviews (of whom 9 were interviewed); and nine
18
customers were categorized into the lead user stratum for the focus groups (of whom 6 participated in focus groups).
3.2 Data analysis
The data collection resulted in a total of 349 quotes which was coded using content analysis into a total of 94 codes under 11 themes. By using industry experts, the codes were scored with respect to the level of novelty, relevance and feasibility, after which calculations were made in order to compare the data, with differences verified by t-Tests.
3.2.1 Knowledge coded through content analysis
Content analysis with an approach of concretizing sentences was used to codify and interpret the knowledge collected from the survey, interviews and focus groups (Graneheim & Lundman, 2004). The goal was to bring raw data from longer sentences into smaller, more handable, codes and subsequent themes. The same method was used on all data gathered, independent of the technique of data collection. The meaning unit, referring to the knowledge shared by the customer in the survey, interview or focus group, was broken down into codes which were, in turn, put into themes, see example in Table 2. This method is useful when analyzing large amounts of qualitative data (Graneheim & Lundman, 2004). Doing this enabled grouping of the answers into different sections of knowledge contribution, while still preserving the core content of every meaning unit. Thus, minimizing the associated parts in the sentences that do not immediately contribute with valuable knowledge, making the work of expert rankings feasible. Furthermore, the process was done in iterations, going back and forth between meaning unit, codes and themes, continuously refining the codes and themes. This is a common way of working with this kind of data (Graneheim & Lundman, 2004).
Table 2 - Example of codification.
19
3.2.2 Expert scoring of collected knowledge
Since the definition of relevance of knowledge is rather subjective (Mahr et al., 2014), it was assumed that this could best be assessed on an industry-specific basis, preferably by experts within the specific industry. While the definition of novelty of knowledge is not as context specific as that of relevance (Mahr et al., 2014), an expert level of experience in the specific industry was still assumed to be needed in order to accurately evaluate the novelty of knowledge; this since the knowledge acquired had to be compared to existing knowledge within the company and on the market. The economic aspects of feasibility are assumed to be company or industry specific (e.g. company financial resources or market size); whilst for technological aspects, the expertise could be both general (i.e.
general technology advancements) or company specific (i.e. the company’s technological capabilities). Thus, an industry expert is assumed to be best suited to evaluate the feasibility as well.
In total, three sessions with experts were held to assess novelty, relevance and feasibility.
Experts were chosen from different fields (marketing, sportsbook and development, see Appendix J) to get a variety of perspectives included in the scoring. The experts were identified as suitable during the time spent at the company’s office, through informal conversations as well as observations of their daily tasks and responsibilities.
The scoring was done using a matrix as shown in Table 3, where the expert ranked the codified knowledge on a scale from 1 to 6, where 1 represented a low level and 6 represented a high level of the knowledge characteristic scored. The reason for conducting the sessions face-to-face was to be able to explain the depth of the codified knowledge to the experts and answer any questions that arose during the session.
Table 3 - Example of expert assessment table.
The scores from all three experts were added together and divided by 3, in order to get a single mean value for each characteristic of knowledge, ranging between 1-6. By doing
20
this, each code got a nuanced (scores from all experts equally important) single value for each characteristic of knowledge to be used for further calculations. To be able to assess the overall benefits of the knowledge, a new metric was introduced by multiplying the three means (relevance * novelty * feasibility), called the total score (T).
3.2.3 Calculations for data comparison purposes
The data was structured in a matrix with rows representing codes and columns for each combination of type of customer and knowledge acquisition technique as well as columns listing the mean scores for each code. Table 4 describes this structure where SU represents survey, IN represents interview, FG represents focus group, T represents total score mean, N represents novelty score mean, R represents relevance score mean and F represents feasibility score mean.
Table 4 - How the data was structured.
To be able to compare how different combinations of type of customer and interaction technique relates to the acquired knowledge characteristics, several metrics were calculated. Firstly, a calculation was done for the number of codes each type of customer (lead users and non-lead users) contributed to in combination with each interaction technique (survey, interview, focus group). This was achieved by adding together all the occurrences of whether the code was acquired through the combination or not, for each column in Table 4, as described below. The reason for using a binary variable rather than counting the number of customers relates to the qualitative focus of this study, with the reasoning that a combination either provided the knowledge or it did not. Thus, from a purely qualitative perspective, there is no added value in multiple customers providing the same knowledge once that knowledge has already been acquired once.
21 Parameters:
c = code number = [1, 2, …, C]
i = knowledge acquisition technique = [survey, interview, focus group]
x = type of customer = [lead users, non-lead users]
Variables:
ocix = 1 if code c could be acquired from technique i with customer type x, 0 else.
Calculations:
5 o789 for i = [survey, interview, focus group]
J
7KL
Secondly, the average score per code for each knowledge characteristics (novelty, relevance and feasibility) was calculated. This answers the question “for a certain combination, taking only the acquired codes into consideration, how high did the experts score the knowledge on average, for each knowledge characteristic?”. Besides the average (calculated as described below), the minimum value, maximum value and standard deviation were also calculated.
New parameters:
k = characteristics of knowledge = [relevance, novelty, feasibility, total]
New variables:
skc = mean score of knowledge characteristic k in code c Calculations:
1
∑J7KLo789∗ 5(o789∗ sN7)
for x = [lead users, non − lead users]
for i = [survey, interview, focus group]
for k = [novelty, relevance, feasibility, total]
J
7KL
Lastly, the average score per customer aggregated over all codes for each combination and for each characteristic of knowledge was calculated. By aggregating over all codes, the results of the calculations more accurately describe the specific combination’s contribution. The choice to calculate on a per customer basis was done in order to be able to compare between different techniques, as there was a different number of participants in each.
New variables:
nxi = number of customers type x participating in acquisition technique i
for i = [survey, interview, focus group]
for x = [lead users, non-lead users]
for i = [survey, interview, focus group]
for x = [lead users, non-lead users]
for k = [novelty, relevance, feasibility, total]
22 Calculations:
1
n98∗ 5(o789∗ sN7)
for x = [lead users, non − lead users]
for i = [survey, interview, focus group]
for k = [novelty, relevance, feasibility, total]
J
7KL
Since the average score per customer aggregated over all codes was the main metric used for comparison between different combinations and how it affects the knowledge acquired, the significance of these values was analyzed using t-Tests. The t-Tests used were two-tailed independent groups heteroscedastic t-Tests, meaning that there is no bias that one group or the other is higher, that the data of the two groups do not come from the same participants and that there is no assumption of equal variance between the groups (Saunders et al., 2009). The choice to go for the stricter option, not assuming equal variance between the two groups was made because the participant sample size was relatively small.
3.2.4 Time spent throughout the study
To be able to assess the resources spent on collecting, processing and analyzing the data throughout the study and for each different technique, time spent was measured and sorted into the three categories (1) preparing, (2) execution; and (3) data management.
The decision to sort the data in those three categories was founded on the differences in the allocation of time between the techniques on these three levels. The first phase, preparing, consisted of planning prior to the data collection. For example, pilot surveys and scheduling. The second phase, execution, consisted of the actual data collection, such as conduction of interviews. The third phase, data management, incorporated transcription of the material gathered as well as go-through of all the interviews and electronic materials from surveys.
3.3 Quality improvement measures
According to Lincoln and Guba (1985), there are four criteria of reliability that can be used to convince the reader that the study is reliable; credibility, generalizability, reliability and adaptability. Below is an explanation of how this has been achieved in this study.
for i = [survey, interview, focus group]
for x = [lead users, non-lead users]
for k = [novelty, relevance, feasibility, total]
23
According to Lincoln and Guba (1985), credibility is achieved through three things;
enough invested time in order to reach the purpose, persistent observation and triangulation. The contact with the case company was established long before the study begun in order to ensure enough time to conduct the study. Lincoln and Guba (1985) mean that persistent observation is about identifying the relevant concepts in relation to the research question of the study. Researchers must therefore in detail be able to describe how research questions were developed and analyzed. The method part of this study is a description of how the results of the study have been achieved. Furthermore, to ensure a common ground and in order to keep the study and corresponding activities relevant, the researchers had continuous discussions throughout the whole study. In addition, both researchers were involved in all steps of the process. Even though few interviews were conducted together, both researchers conducted interviews and reviewed all the data collected. During the analyses, it was kept in mind that Gioia, Corley and Hamilton (2013) strongly advise researchers to ignore previous findings in order to keep an open mind throughout the study. Therefore, coding was not done until after all data was collected and transcribed, in order to remove any bias that might evolve during the codification process. Furthermore, triangulation was achieved by letting multiple experts from different fields score the different characteristics of the knowledge gathered, in order to minimize the risk of a biased expert scoring. In addition, the categorization of customers was kept exactly the same for all customers, based on characteristics springing from established researchers with broad experience within the field.
There are two steps to ensure reliability and adaptability (Lincoln & Guba, 1985); detailed description of the process that lead to the results and reviews conducted by others outside of the study. To ensure this, the procedure of this study has been described in detail in the section of the method. The surveys that were sent out can be found in Appendix A, B, D and G. The interviews and focus groups were performed based on guides found in Appendix E and H respectively. This provided continuity in the interviews and allowed for a better comparison based on knowledge acquisition technique and type of customer interacted with. An overview of the informants that participated can be found in
24
Appendix C, F and I. Reviews conducted by others has been achieved through seminars and constant discussions with supervisors, both at the university and at the case company.
To achieve generalizability, measures should be taken such that the study can be repeated with the same results in other contexts (Lincoln & Guba, 1985). In this study, generalizability was achieved in two steps. First by ensuring that the scoring was conducted by experts from different fields, making the findings applicable to different environments and, secondly, by not analyzing the gathered data before all data collection was completed, reducing the bias of the researchers.
To ensure a rigid process, additional measures were also taken during the critical phase of transforming data from customers to electronic data to be analyzed. The transcriptions of the interviews and focus groups were done as soon as possible after they were conducted. This in order to leverage as much as possible on the impression taken during the interaction. In addition, each focus group had, besides the participants, one moderator and one person taking notes, as well as being audio recorded. The note-taking and audio recording made sure key points were captured (Saunders et al., 2009).
25 4. FINDINGS
This section presents the findings. First, data illustrating how the combination of type of customer (lead users and non-lead users) and knowledge acquisition technique (surveys, interviews and focus groups) influence the characteristics of knowledge acquired (novelty, relevance and feasibility) is presented (corresponding to RQ1). Secondly, the costs associated with acquiring the knowledge are elaborated on (corresponding to RQ2).
4.1 The combinations’ influence on knowledge characteristics
Table 5 presents descriptive statistics for the three variables used for calculating the final score (available in Table 6): (1) the number of customers participating in each combination of type of customer and knowledge acquisition technique, (2) the number of codes each combination contributed with, and (3) each combination’s average score per code acquired, for each knowledge characteristic.
Table 5 – Descriptive statistics for the three factors used for calculating the final score.
26
Figure 2 presents the average total score per code (quality), taking all three characteristics of knowledge into account, in relation to the number of codes contributed to (quantity), and the number of participating customers (represented by the size of the marker).
Figure 2 - Number of codes contributed to, average total score per code and number of customers.
In order to be able to accurately compare the different customer and technique combinations, Table 6 presents the averages for each characteristic of knowledge aggregated over all codes contributed on a per customer basis; thus including the number of participants as well as both the quality and quantity of knowledge acquired.
Table 6 – The knowledge characteristic specific score, per customer, for the combination of customer type and technique.
Graphical displays of the data for each characteristic of knowledge, as drawn from the numbers in Table 6, are displayed in Figures 3 to 6. Further, Tables 7 to 10 show the ratio between lead users and non-lead users in combination with different techniques; or when comparing different techniques within the same customer type, the ratio between the two techniques. Tables 7 to 10 also highlight the significance of the calculated ratio, as identified through t-Tests.
27
4.1.1 Novelty score
Figure 3 - Novelty score per customer and technique graph.
As presented in Figure 3 and Table 7, there is a great difference in the novelty score acquired through interviews and focus groups with lead users when compared to any other combination of customer and technique. Applying these two techniques on lead users produce around twice as high novelty score than when non-lead users are involved using the same techniques. Further, lead user interviews and focus groups score almost three times higher than lead user surveys, and more than four times higher when compared to non-lead user surveys. Although the score is twice as high for interviews and focus groups than for surveys, with non-lead users, the two do not generate a significantly higher score when compared to surveys with lead users.
Table 7 – Ratios and t-Test of Novelty knowledge characteristic.
28
4.1.2 Relevance score
Figure 4 - Relevance score per customer and technique graph.
As presented in Figure 4 and Table 8, lead users outperform non-lead users in regard to relevance score when comparing the same knowledge acquisition technique, varying the type of customer. For interviews and focus groups, lead users generate a relevance score close to twice as high when compared to the same techniques for non-lead users respectively; and for surveys, lead users score around 50 % higher than non-lead users.
Although the score is twice as high for interviews and focus groups than for surveys, with non-lead users, the two do not generate a significantly higher score when compared to surveys with lead users. For lead users, interviews and focus groups score around four times higher than for surveys.
Table 8 – Ratios and t-Test of Relevance knowledge characteristic.
29
4.1.3 Feasibility score
Figure 5 - Feasibility score per customer and technique graph.
As presented in Figure 5 and Table 9, lead users outperform non-lead users in regard to feasibility score when comparing the same knowledge acquisition technique, varying the type of customer. For interviews and focus groups, lead users generate a feasibility score close to twice as high when compared to the same techniques for non-lead users respectively; and for surveys, lead users score around 50 % higher than non-lead users.
Although the score is twice as high for interviews and focus groups than for surveys, with non-lead users, the two do not generate a significantly higher score when compared to surveys with lead users. For lead users, interviews and focus groups score around three and a half times higher than for surveys.
Table 9 – Ratios and t-Test of Feasibility knowledge characteristic.
30
4.1.4 Total score
Figure 6 - Total benefit score per customer and technique graph.
As presented in Figure 6 and Table 10, there is a great difference in the total score acquired through interviews and focus groups with lead users when compared to any other combination of customer and technique. Applying these two techniques on lead users produce around twice as high total score than when non-lead users are involved using the same techniques. Further, lead user interviews and focus groups score almost three times higher than lead user surveys, and more than four times higher when compared to non-lead user surveys. Although the score is twice as high for interviews and focus groups than for surveys, with non-lead users, the two do not generate a significantly higher score when compared to surveys with lead users.
Table 10 – Ratios and t-Test of Total knowledge characteristic.
31 4.2 Cost approximations
Scholars have concluded that the main part of the cost associated with conduction and planning of each knowledge acquisition technique could be derived to the time it took to conduct each technique (Guest et al., 2017; Roach, 2007; Seaman, 1999). In addition to the costs of conducting a technique there is also a cost associated with categorization of customers, as this was done through a survey technique.
4.2.1 Costs associated with the categorization of customers
The costs of categorizing the customers can be assumed to be equal to the costs of conducting a survey, further elaborated on in the following section. However, the findings show that lead users are proportionally more likely to be willing to participate in focus groups than interviews, and further, more willing to participate in interviews than in surveys (as shown in Table 11 and Figure 7). Thus, the effort and cost of getting the desired number of lead-user participants are assumed to reduce accordingly.
Table 11 - Total number of lead users identified for each technique.
Figure 7 - Visualization of the percentage of lead users identified through each technique.
32
4.2.2 Cost associated with interaction techniques
The type of customer does not affect the cost of executing the interaction techniques, assuming the customers are already categorized. The following paragraphs elaborate on the total cost associated with each interaction technique, as well as how the time was distributed between different activities.
According to both the literature and our observations, the resources, and therefore also costs, associated with each technique were directly correlated to the time spent with each technique. The time itself can be divided into three different phases; (1) preparing, (2) execution; and (3) data management. All phases were measured and analyzed with respect to each technique, see results in Table 12 and Figure 8.
Table 12 - Total number of hours spent on each technique in each phase
Figure 8 – The hours adding up throughout the process for each technique
Although the survey preparations required a pilot survey including analyzing and rework of the original set of questions together with detailed survey design, it was the least time- consuming technique. This mainly due to the facts that respondents could answer the survey whenever they pleased, and that the material was already in a digital format when collected. For interviews, the bulk of time was allocated to the actual conduction of the
33
interviews and subsequent processing of the material through transcription and analysis.
Less time was allocated to planning prior to the interviews. This because of the fact that all respondents were independent of each other and thus, only had to synchronize with the interviewer; as well as the nature of semi-structured interviews being a quite spontaneous interaction, requiring little planning prior to conduction. In contrary to interviews, the bulk of time spent on conducting the focus group was mainly allocated to the coordination and planning that took place before it was held. This since the schedule of each informant had to be synced with that of all other informants; as well as the need to structure and plan the actual conduction of the focus group. Further, the focus group also required preparations in terms of location and setting. Less time was allocated to the conduction of the focus groups, thus resulting in a lot less time consumed processing the material when compared to the interviews. Figure 9 shows an overview of the relation of the times spent on each technique,
Figure 9 - Overview of the relation between the time spent on each technique.
In conclusion, the ratio of total time spent on interviews and focus groups is approximated to 1:1. However, there is a difference as the interviews required more time spent on the execution and data management; while focus groups required more time related to preparing. Surveys were found least time consuming, approximated to a ratio of 2/3 of the time spent on interviews or focus groups.
34 5. DISCUSSION AND CONCLUSIONS
This section discusses the impact of the findings presented in the previous section. First, the findings related to RQ1 are discussed by addressing the impact of the relation between type of customer, knowledge acquisition technique and knowledge characteristics.
Secondly, the findings related to RQ2 are discussed and further elaborated on by presenting and comparing the cost of the different techniques in relation to novelty, relevance and feasibility in the acquired knowledge.
5.1 How the combination of type of customer and knowledge acquisition technique affect characteristics of knowledge acquired
The findings allow a number of different conclusions. Firstly, there is a significant difference between the interviews and focus groups with lead users compared to any acquisition technique with non-lead users. Interacting with lead users through interviews and focus groups gives a higher score on all three characteristics of knowledge than interaction with non-lead users through techniques. Therefore, it can be concluded that the combination of lead users and interviews or focus groups bring a significantly higher value than the combination of non-lead users and any of the techniques.
Secondly, none of the interaction techniques with non-lead users score significantly higher than surveys with lead users, for any characteristic of knowledge. This implies that interviews and focus groups held with non-lead users do not contribute with a higher score than conducting surveys with lead users. Lead user surveys score significantly higher than non-lead user surveys on relevance and feasibility of knowledge. However, these findings suggest that there is no need to categorize customers for a survey if the goal is to maximize the combined results or to gain novel knowledge. Yet, only taking the knowledge acquired into account, the findings show that the benefit of conducting interviews or focus groups with lead users corresponds to double the score of surveys with either of the customer types or interviews and focus groups with non-lead users, leaving these alternatives rather unattractive.
