The first step in deciding how to use data is to define an appropriate use case. When looking for use cases, many banks reflexively tend toward the customer interface. This choice is understandable because banks also want to position themselves to their customers as innovative institutions with modern solutions. However, ideal use cases are often found in other areas as well.
In the blog post “Data, the new gold” we commented on the risks and necessary precautions surrounding the handling of data. In particular, finding a balance between “seizing opportunities” and “avoiding risks” is a significant requirement for many companies. In this post, we will explain how to define a data use case properly.
A use case describes the way a user interacts with a system or product. Use cases are employed to set requirements, define the scope of a product or service, and limit risk. A use case can specify success or failure scenarios, as well as any critical variations or exceptions.
In its guidelines on “Handling data in day-to-day business”, the Swiss Bankers Association (SBA) lists various customer-oriented use cases. Therein, specific legal questions arise based on the customer data analyzed: artificial intelligence (AI) for compliance (Know Your Customer KYC & onboarding, transaction monitoring), credit checking, trend analysis and benchmarking, biometric authentication, personalized offers and guidance, and loyalty programs.
However, it is worth considering other processes with the potential for data analysis. Most banks have process maps, such as the one shown here. Alternatively, reference models such as BIAN are good orientation aids. It helps to keep an open perspective. That way, one can identify the most suitable applications for the entire company.
Effective instead of “defective” use cases
In our experience, if a company is at the very beginning of its journey towards becoming a “data-driven company,” use cases with customer data are less suitable. Use cases at the customer interface are often more complex than cases without direct customer interaction:
- The handling of customer data brings additional risks and, thus, requires mitigating measures, as described in our first article of this series.
- The issue of customer centricity is raised increasingly in product development. That’s why companies often test prototypes of products with end customers. However, many people are very sensitive when it comes to data processing. This raises the question of whether the still low maturity of the tested solutions reduces the desired marketing effect.
The Eraneos Use Case Identification Workshop
To clarify the question “How should I use data?”, it helps to proceed systematically. When introducing data analysis in companies, we like to start with a Potential Workshop. There, we identify and evaluate all possible use cases.
There are no wrong ideas in this workshop. We record all topics and evaluate them according to potential or effort. The workshop should also be composed as interdisciplinarily as possible to cover the entire range of activities of the institute. In Use Case Workshops, we consciously focus on all of a bank’s processes. Often, we find quick wins that achieve noticeably good results with little effort far away from the front (e.g., in the operating or support processes). An ideal first use case shows measurable results quickly and meets a specific need of various departments. An ideal first use case shows measurable results quickly and corresponds to a particular need of the respective specialist departments. Data processing and evaluation should only begin after the use case has been identified.
The example here shows how the area of “Compliance Analytics” was identified in the Use Case Workshop. Three fields of action were derived.
Six steps to a compelling use case
To flesh out the use case that has been identified, I recommend an iterative procedure with the following six steps:
1. Set story scope
A compelling use case requires a story. Setting boundaries is essential! In the Planning Meeting, the scope of the story should be determined quickly. Especially in the initial phase, the following applies: less is more. Start with a manageable scope; it can still be expanded later.
2. Ask questions
In the Question Breakdown Meeting, we deepen the story. Which questions should the analyses answer? It is easy to get caught up in the flurry of existing KPIs. It is helpful to try to tune them out and look at other dimensions. If, for example, the stability of a process in operations is analyzed, the focus is often on the end-to-end automation rate, the “straight through processing” (STP) rate. However, the analyses could also delve into other aspects, such as at which steps the continuous processing fails. This could be done, for example, to identify process flows for which a fundamentally different processing method would have to be examined.
3. Analyze data
The questions are clear. Now, it’s time to find answers in the data. One quickly gets to a point where a data source for automatic evaluation is missing. In the first iteration, “minimal viable product,” it is sufficient to integrate a new data source, for example, a static data source. A real-time connection can be set up in a later step.
At this point, at the latest, it becomes clear that the handling of personal data is many times more complex than use cases without sensitive data. If, for example, customer data is involved, it must also be protected in a prototype according to the principle of “need-to-know,” including extensive authorization mechanisms.
4. Documenting results
After we have found and linked the data, we visualize it. It is important here to always keep the original questions in mind because these need to be answered. In the Visualization Design Meeting, we combine the different opinions: technical and business specialist experts discuss the results with the users.
5. Representing added value
To anchor the added value in the company, it is also crucial to shed light on the application of data evaluation. In which processes do I use evaluations? What are the benefits from them? This is the only way for the analyses to find their way into the organization.
On the journey to becoming a “data-driven company”, companies must also pay attention to change management. A new dashboard will not be used if users do not trust the content. In corresponding storytelling sessions, it is vital to reduce skepticism about the data and evaluations and to show their added value.
It is easy to find use cases for the use of Data Analytics. To anchor these in the organization, we recommend following a prioritization process. Identify the use cases that promise the most appealing cost/income ratio. It is advisable to look at the entire company here; a process map helps for orientation. The selected use case can then be detailed and implemented using the six steps shown. With such a multilevel approach, you can ensure that you use data efficiently and create sustainable added value for your company. Especially for the first steps in dealing with data, it is advisated to start with simple applications – if possible without personal or sensitive data.