As it turned out earlier, it is vital for organizations who believe in the value of data and the competitive advantage that its utilization brings, it is important to understand where data modeling should start. We interviewed Ari Hovi (www.arihovi.com) and together with their leading data modeling expert Hannu Järvi, we went through the biggest pitfalls and best practices in data modeling.
First, it is important to recognize the facts. If you have dozens or hundreds of information systems, data warehouses, and businesses, it’s clear that it will take years to create an entire data model that describes the entire company and all of its functions, and it will become infinitely complex and difficult to leverage. Of course, this is possible to do but very rarely appropriate. It’s important to start with things that are the most important to the company, we call these focus factors. Let's imagine an example of a venue where the customer who comes to the event is at the center. Of course, it is possible that your focus may even be on the actual costs of an individual event and the services offered to visitors, but from this you can already see how important it is to model the structure in terms of what is important to your business.
Thus, the idea of a "customer-driven data model" for the industry of the transaction industry is created. When you put the customer at the center, you can start to form entities around it that affect customer satisfaction, the euros the customer spends and the customer's different situations. What is the customer's path in the event, what situations are involved in the path, what services are sold along the path, how are the services consumed, how different services affect the customer's satisfaction and so on. This story can be continued indefinitely. The real benefit of modeling is that we start to see how things relate to each other and the visualized overall picture helps to better understand what are the key factors for leadership. A good simple example of this can be a rock concert where customers arrive at arranged transportation. If the bus or train runs late and half of the entire gig is missed because of this, it is clear that customer satisfaction is not good. While this is an easy thing to think about without a data model but much more wider and complex environments, you really need such a task. Think about this through your own business so that the real reason behind the visible consequence is the sum of three or four (or even more) variables and if the interconnections between them are not described, it is impossible to get an overall picture or manage it effectively.
The next most important factor is to choose a business competitive factor. What is it about companies in terms of values, strategy, or defined distinctiveness? Analyzing this with a focus factor often yields a view where a very in-depth view of the analysis of a firm’s competitive factor and the factors that affect it is obtained. Remember that you cannot outsource this. Where all the most critical things in your business need to be in your possession. A consultant or subcontractor may not understand all of your functions, the relationships between them, or strategic directions such as members of the top management team.
As a final factor, I highlight the importance of visualization and terminology. Often when doing something new, it is clear that things are understood in different ways and little different terms are used. But a well-done and visualized structural data model explains more than a one view of written document in more than 10 pages. If model is not understood in the same way as the basics of the data model and the concepts used, the time spent on the data model may seem wasted. However, the situation is different, data modeling is never wasted and almost every time data modeling evokes in the management team "AHA" experiences of the connections between things and the fundamental logics in the company’s business.
Hannu Järvi states well that "people are a little afraid to describe complex and multilevel things because they are linear in nature. Data modeling is systemic and therefore can contain more than two dimensions and "sidepaths"", making it impossible to perceive cause-and-effect relationships without orthodox modeling. Because business environments are often complex, they cannot be derived by simplifying or ignoring indirect factors or all factors. That is, despite the fears, the work is always worth it.