The data analysis process was started with the customer interviews. First, the interviews were listened to and the transcriptions of the interviews were read through twice in order to gain a general feeling about the interviews. Becoming acquainted with the data helped to understand the interviews as a whole and create an initial understanding of the customer’s relation to the research phenomenon. Using coding28, categorizing and thematizing the data was investigated for similarities and interesting aspects related to the research phenomenon. No predetermined coding scheme was used; the focus was on letting the data speak in broad terms. Throughout the coding phase the data was asked guiding questions such as how do customers describe the information affecting their value creation?; what is the role of the information in the customer’s everyday life?; what causes dissatisfaction toward the information and why?; etc. Altogether, the overarching goal was to achieve a broad and holistic understanding of the customer perspective to the research phenomenon at hand.
The coding scheme in itself was developed from the empirical data. The initial coding phase yielded altogether 42 distinct and initial categories, including
28 Coding can be understood as a method of connecting data and issues involving interpretations, data
categories such as ‘Not for me’ (information was not regarded beneficial), ‘Guiding’ (helping customers to achieve a more healthful life), and ‘Education’ (information as educator), ‘Easiness & Convenience’ (service as convenient to use), to name but a few. At this point, these categories were not organized into any hierarchical structure or under broader themes or entities. As the understanding of the empirical data increased, the categories were developed further including hierarchical elements. This was achieved using NVivo’s tree nodes as a way to organize data. In addition to the hierarchical structure of the data, some categories that were developed during the first round of coding were readjusted, merged into broader entities or divided into sub categories. Altogether, the reorganization of data was the result of perceiving the data more as a whole, identifying emerging patterns and similarities and understanding the interrelations of the empirically-grounded categories.
Customer feedback data was different from the interview data for it was already initially in textual form and was not characterized by the interactivity of the interviews. The majority of the feedbacks were rather narrow and limited in nature and thus far from the richness of the interview data in terms of the ability to describe the research phenomenon in depth. However, customer feedback data provided good insight into what customers regarded as good or bad in the information resulting from reverse use of customer data and what should be developed further in the service. Customer feedback succeeded in uncovering aspects that were considered important in the customer’s value creation. Although being one-sided and limited in size, they offered condensed and outlined perspectives to the research phenomenon. In addition, the total amount of customer feedbacks (456 pieces) can be regarded as rather large, which also contributed to the wide variety of different and diverse customer perspectives to the research phenomenon.
The analysis of customer feedback data started with reading all the feedbacks through twice. After that, the feedback data was coded and categorized using the hierarchical structure developed when coding the customer interviews. The existing categorization was used due to two reasons. First, the majority of the aspects found in the customer feedback data had already emerged from the customer interview data and hence, fitted well into the developed structure. Second, although data triangulation was used in generating data, it was utilized to generate data from the same phenomenon. It was not necessary to start with a new coding scheme for the
feedback data, but to let the additional data generation method enrich and shape the existing categorizations.
A critically important aspect in the data analysis process was to understand the nature of the data. In terms of data validity, it was vital to notice when the informant was talking about him- or herself, i.e. about actual experience, and when the focus was on general observations about the case context. Furthermore, it was important to distinguish whether informants were talking in conditional terms; i.e. not based on personal experiences, or through imaginary experiences of others. The focus was placed on teasing out what can be called experiential knowledge from what is only opinion and preference (Stake, 2004). This was in line with the chosen theoretical perspective, i.e. phenomenological hermeneutics, and the focus on interpreting customers’ subjective experiences. Thus, the process of data analysis focused systematically on the subjective experience of the informants in order to gain a trustworthy, valid and well-grounded description of the research phenomenon.
After the analysis the transcriptions of both the customer interviews and the feedback data were read through once more. This was done to evaluate how well the output of the analysis eventually reflected the data. Only minor adjustments were made.