System architecture and data

The BI and analytics theories and frameworks presented in the previous chapters suggest that advanced analytics could give companies important competitive ad-vantage when big data (unstructured data) is used. The case company, however, is not utilizing any unstructured data or even external data in their analyses for management reporting at the moment.

The below figure demonstrates the current structure and connections of the different information systems of the Company X. The most important elements from the thesis’ point of view are the information the BI tool extracts from the Service Fee Engine (SFE), Opsware, which is the company’s customer relation-ship management tool (CRM), and ERP.

Figure 5. Current data flow for financial reporting in the case company

In the current setup, the BI tool uses ETLs to extract the data from SFE (rev-enue from active and paying accounts) and CRM (customer behavior, pricing plan and other customer account related data) but does not have the capability to extract the data in ERP. What this means in practice, is that every month, the controller pulls the needed reports from the BI tool, extracts manually the data from ERP and reporting tool, and then manually put the reports together.

According to the controller, it would be ideal if this process could be stream-lined so that putting the reports together would not require manual labor, but instead the BI tool could use ETLs to extract data from ERP and the reporting tool automatically as well (illustrated in the below figure).

“The sales data flows automatically because the fee engine feeds that data automat-ically to the BI system and I can get the data almost in real-time. What is missing is the expense data and that is not loaded to the BI tool so what we do is we compile the reports so that we take one part of the data from the BI tool and another part from the ERP. We still have not managed to get these systems to talk with each other.” (Business controller, 2020.)

Figure 6. Desired data flow for financial reporting

In theory, the Brain team has the capabilities to establish the connection be-tween the BI tool and ERP, but due to the sensitive nature of the ERP data, there are major concerns about sensitive data being displayed in the BI tool. The reason why this is a concern, is that in the case company, the BI tool is in organization-wide use. It was a management decision to have full transparency in the data and give full access to all members of the organization, meaning that everyone has the same access rights. The reasoning behind the decision to have all the data in the BI tool available for everyone is that it, for example, reduces the time to make and implement decisions as the assumption is that all employees have access to same data, which eliminates the risk of a team or an individual to argue that they were not aware of certain decisions being planned as everyone is equally in-formed.

Furthermore, the fact that the company wants to promote full transparency may pose limitations to full utilization of the BI tool’s capabilities in terms of re-porting and analysis scope. This is acknowledged in the company, but the man-agement’s view is that information symmetry across the employees outweighs the utilization of BI in its full potential, and therefore the sensitive information is handled elsewhere and not in the BI tool.

According to finance, the data quality has been good, meaning that there is rarely a need to amend the raw data. The Head of Brain agrees that in general the data quality is good, but depending on the task, the effort required to put the data together for an ad hoc report can be significant. This is because the data repository and infrastructure were not initially designed to enable flexible report-ing, therefore, the question may not be whether the data quality is good or bad, but rather whether it is fit for varying needs or not. Moreover, the system

developers emphasize tool performance over analytic capabilities, which is re-flected on how the data is stored in the repository.

“We rarely need to manually correct the data afterwards. In that sense, the data quality has been good. In addition, if we made manual corrections and someone saw the original report, we would have to use the time [in result review meetings] to explain why we had to make the corrections. It’s all very good data quality, we don’t need to do manual tweaking on the data itself.” (Business controller, 2020.)

“If we notice a clear error in the data, we do not make changes by ourselves, but add a side-note in the report.” (CFO, 2020.)

“We will sometimes have quite a big challenge to give the data that the business controller and the CFO want because the way that things were started. What I want to emphasize is that the amount of work needed to put in to get this [the requested data]

presentable. The quality of data is good, but it’s not made for data analysis, so we break our heads a little bit more to make that presentable to different stakeholders.” (The Head of Brain, 2020.)

Finance does, however, recognize that there are some improvements needed to ensure data consistency and confidence. According to the business controller, in the month-end reporting process there are clear cut-off days for month-end closing. However, it is not always clear when the data showed in re-ports have been updated. Therefore, from time to time there are rere-ports or dash-boards, which are partially populated with the latest data, while in other reports the data might be old, but it may not be distinct to the user.

“For example, I give you a figure for Finnish customers that are in pricing plan X, and you use that later for your own purposes. Later, you will check the figure with me, and I ask the Brain team to confirm the figure, but surprisingly they give a totally differ-ent number, which means that I have to manually check or ask the Brain team to manually check the data.” (Business controller, 2020.)

“One table might get refreshed during the night and the second may be updated at the same time as the first table, but then the third might be updated with latest data in the middle of the day and the fourth table can be updated only when the third has been updated. Due to these possible dependencies we have to know when the data loads are scheduled and whether the data has been updated or not.” (Business controller, 2020.)

“I don’t really see bottlenecks in our reporting process, instead, for example, if we have multiple tables or charts in a dashboard and they are not updated at the same time, it can confuse us an awful lot that the data is not consistent.” (CFO, 2020.)

From the above can be concluded that there could be a need for better plan-ning process. As the Head of Brain states that currently the team does not have the time to think and plan ahead meaning that they have to work reactively

instead of being proactive. This can have a significant impact on the Brain team’s workload as if the current data design does not fit the requirement, the team needs to build workarounds to reclassify the data.