Description and acquisition of the research material

The research material for this study is acquired in multiple steps, the first one being sam-pling the material from existing e-commerce sites. The e-commerce sites under investi-gation were determined on the following clauses. Firstly, the companies had to have a website for e-commerce purposes, and secondly, they had to be on the industrial auto-mation line of business. One exception was made for this in order to get an in-depth overview of customer intimacy in e-commerce by taking a closer look at how a known customer intimate online retail company works. Some of the e-commerce sites sell just the company’s products and some work as a platform for third-party products as well.

However, this is not relevant for the research and is therefore not addressed. The e-commerce sites and companies included in the research are listed in table 2. The empir-ical research material was collected from the e-commerce sites during May and June 2021.

1. Preliminary development task or problem

2. Familiarizing with the phenomenon in practice and theory. Specification of the development task.

3. Collection and analysis of empirical data

4. Development proposals

Table 2. Companies and their e-commerce sites included in the research.

Company E-commerce site URL

ABB ABB Ability Marketplace™ https://eu.marketplace.ability.abb Siemens Siemens MindSphere https://siemens.mindsphere.io/en Schneider Electric Schneider Electric Exchange https://exchange.se.com

Honeywell Honeywell Marketplace https://marketplace.honeywell.com

PTC PTC Marketplace https://www.ptc.com/en/marketplace

Amazon Amazon https://www.amazon.co.uk

The second part of the research material is generated in the research process by con-ducting an interview because when wanting to know how a customer really perceives customer intimacy and how it should be developed, it is simplest to ask directly. Inter-views are the most common data acquisition methods in qualitative research, and it can be seen as a conversation that has a beforehand defined goal (Juuti & Puusa, 2020, p.

99). Conducting an interview is a great way to collect information about previously un-known and little studied subjects (Hirsjärvi & Hurme, 2015, p. 35). There are several in-terview techniques that vary depending on their degree of structuring, which means how freely the interview is conducted (Juuti & Puusa, 2020, p. 105). This research uses a semi-structured interview which suits the needs and objectives of the research best. A semi-structured interview has an interview structure that is determined in advance, but it is more flexible than a structured interview, in which the answer options are defined as well (Juuti & Puusa, 2020, p. 105).

The interview was conducted in June 2021 via Microsoft Teams. The one-to-one inter-view lasted approximately 44 minutes. The interinter-view questions were provided to the interviewee a week before the actual interview so that he had an opportunity to get to know them and familiarize himself with the subject. Tuomi and Sarajärvi (2018, pp. 85-68) recommend this as it helps to obtain as much information as possible about the sub-ject. The interview was recorded, after which it was transcribed in order to be able to analyze it. The interviewee was selected for the interview because the thesis commis-sioning company had knowledge that the company where the interviewee works has

been eager to further develop e-commerce. Therefore, it could be presumed that the interviewee would have meaningful points of view on the subject. The interviewee works in the steel industry and his role is closely related to procurement. The viewee’s anonymity has been protected and he is therefore called by the term inter-viewee and not his name.

However, interviews are usually expensive and time-consuming, so when wanting to get answers from a larger group of people, surveys provide an easier way (Tuomi & Sarajärvi, 2018, p. 86; Ojasalo et al., 2014, p. 40). Therefore, the research material also includes a survey that was conducted in 2020 by the thesis commissioning company ABB Oy. The survey was conducted in 16 countries across the globe with a total number of 260 survey respondents. 35 % of the respondents were from Europe, 35 % from Asia-Pacific, and 30 % from the Americas. The respondents are divided into four groups: channel partners (131), original equipment manufacturers (37), end-users (24), and finally local ABB sales units (68), which will be excluded from this research because the focus is on customers.

4.3 Analysis

The data analysis provides clarity on the research material in order to produce new in-formation on the research topic (Eskola & Suoranta, 1998, p. 100). The aim is to summa-rize the material and bring out clear and meaningful observations from the seemingly fragmented material (Eskola & Suoranta, 1998, p. 100). This research uses content anal-ysis as the data analanal-ysis method, which means analanal-ysis of written, seen, or heard content (Tuomi & Sarajärvi, 2018, p. 78). It is one of the most common methods in qualitative research and can be applied to a wide range of studies (Juuti & Puusa, 2020, p. 144).

Qualitative content analysis is good at answering descriptive research questions (Schreier, 2012, p. 43). The analysis thus brings clarity to the data so that reliable and reproducible conclusions can be drawn from the phenomenon (Krippendorff, 2013, p.

24). Therefore, the analysis must be credible and truthful, and the solutions must be justified (Juuti & Puusa, 2020, p. 144). Qualitative content analysis can be applied to

various materials, such as websites, company brochures, newspaper articles, and inter-view transcripts (Schreier, 2012, pp. 2-3). This study uses both interinter-view transcripts, sur-vey answers, and websites, which means that there are both primary and secondary data involved.

Content analysis can be implemented in three ways: data-based, theory-based, or the-ory-driven (Tuomi & Sarajärvi, 2009, pp. 95-98). This study uses thethe-ory-driven analysis, in which previous knowledge and theories aid the analysis, but the analysis is not entirely based on theory or research data (Tuomi & Sarajärvi, 2009, pp. 97-98). This means that both theoretical background and the researcher’s own interpretations that are based on the observations are affecting the analysis. Whereas in data-based analysis, the theoret-ical concepts are entirely based on the research data and theory-based analysis is guided by previous knowledge and the categories are formed based on it (Tuomi & Sarajärvi, 2018, pp. 97-98). Content analysis is generally thought to consist of certain stages, which include the selection of the analysis unit, familiarization with the material, reduction of the material, classification, theming and typing of the material, and interpretation (Juuti

& Puusa, 2020, p. 144; Tuomi & Sarajärvi, 2018, pp. 104-105). Qualitative research usu-ally collects data in several stages and using parallel methods, which results in the data acquisition and analysis stages being partially overlapping and the analysis is carried out through the research process (Juuti & Puusa, 2020, p. 141). This is typical for a qualitative research (Juuti & Puusa, 2020, p. 10), and happened in this research as well. Some re-searchers consider content analysis to be a method that combines qualitative and quan-titative data and analysis (Juuti & Puusa, 2020, p. 144). This is also done in this study since the survey results are clarified by counting the numbers of responses and their percentages. This helps to understand the findings and their prevalence.