*4.3. Cross-Case Analysis*

In the cross-case analysis, the results of the case studies were analyzed in comparison to the relative maturity of the data science capability as reported by the interviewees, the perceived success of the data science outcomes from the perspective of the project team, and whether or not the outcomes were accepted and adopted within the primary business processes. Table 2 below compares the two case studies based on data governance maturity, data science outcomes, and the adoption status of the data governance outcomes.


### **Table 2.** Comparison of the case studies.

Table 2 above shows that Project A has an established data governance capability and that the outcomes of the project were accepted by the business. In Project B, the organization does not have an established data governance capability, and the data science outcomes were not adopted by the business.

4.3.1. The Role of Data Governance with Regards to Data Science as a Product of Human Agency

In Table 3 below, the role of data governance with regards to the successful implementation of data science as an organizational capability is compared between the cases.



In Table 3 above, we notice that although Project A required more data sets than Project B, data access was not considered an issue, and the project was able to be completed with a minimum of extra effort. In contrast, project B team members were required to set up the data infrastructure, find the data, and manage access and data quality themselves.

4.3.2. The Role of Data Governance with Regards to Data Science as a Medium of Human Agency

In Table 4 below, the role of data governance with regards to the acceptance, coordination, and control of data science outcomes is compared between the cases.

**Table 4.** Comparison of the cases with regards to the role of data governance in data science as a medium of human agency.


From Table 4 it becomes clear that in project A, data owners were involved from the start of the project until delivery. In addition, data owners were accorded ownership of the outcomes. As a result, the outcomes were accepted by the data owners. This is in contrast to Project B, in which data owners were not available, and business owners did not accept the data science results.

4.3.3. The Role of Data Governance with Regards to Organizational Conditions of Data Science

In Table 5 below, the role of data governance with regards to the coordination and control of organizational conditions of data science is compared between the cases.

**Table 5.** Comparison of the cases with regards to the role of data governance in coordinating and controlling the organizational conditions of data science.


From Table 5 we can conclude that Project A has a strict regime of coordination and control following a yearly review as well as well-defined roles and responsibilities. In contrast, team members of Project B were given little direction and no ownership was displayed by business leaders.

4.3.4. The Role of Data Governance with Regards to Organizational Consequences of Data Science

In Table 6 below, the role of data governance with regards to the coordination and control of organizational consequences of data science is compared between the cases.

**Table 6.** Comparison of the cases with regards to the role of data governance in coordinating and controlling organizational consequences of data science.


From Table 6 it can be derived that in Project A, the implementation of the data science outcomes was managed by a dedicated project manager in conjunction with the data owners. This was in contrast to Project B, in which no data owners were involved and a rival COTS application which had previously been acquired by business leaders created an insurmountable conflict for the project team.
