Learn more about Donna Burbank.
One of the things I love about data modeling, and data management in general, is that it gives me the opportunity to work with a variety of organizations around the world and understand what makes them run. In a sense, a good data modeler is a management consultant for information – understanding how a business works, how the data underpins the organization, and what can be made more efficient.
In many ways the mind of a good businessperson is similar to that of a good data modeler—continually asking questions and looking for areas of improvement. A recent example at a client of mine brought this to light. In building a conceptual data model for a manufacturing company, I was working with a senior engineer to understand the underlying data model for several functional business areas. As he had some previous experience with data modeling, in addition to the logic of engineering, I found this session particularly productive.
The data modeling process asks a series of questions that are almost childlike in their simplicity, but when done in a methodical way, can highlight important business rules that might not have come to light. Can customer have more than one account? Must a customer have an account to be considered an active customer? Can an account be associated with more than one customer at the same time? A data modeler hearing these questions might immediately start thinking of cardinality and relationships, but a businessperson hearing the same questions should be thinking about business scenarios. For example, “Ah, you’re right, if a customer doesn’t have an active account, we should mark them as expired—we hadn’t been doing that.” Often additional ideas come to light, spawning more questions. “If we’re not tracking expired customers, we’re not sending them renewal notices and are potentially losing follow-on business.”
Data modeling is a process, as much as a result. The very process of asking the questions necessary to build a model drives the value of the model. For the manufacturing client, using the typical types of questions used in data modeling, we were able to not only document how information supported key business functions, but we were able to highlight areas of inefficiencies or ones that needed clarity. Some of the types of questions that we asked included:
- Can there be different market arrangements for each product? Do these vary by region?
- Is customer support the same for wholesale and retail clients?
- Should all support tickets be tracked for each customer, or only ones that resulted in a valid issue?
Each of these questions prompted further discussions on how the business worked and what might be made more efficient or more in line with industry regulations.
Having dinner with a colleague later that evening, he mentioned that he had overheard much of our session. He asked, “Weren’t you frustrated that the client didn’t know more about his business?” I was shocked by the question, since I felt we had spent an entire day delving into business processes and highlighting areas for improvement. “Why do you think he didn’t understand the business”, I asked. “Well, he kept saying things like ‘I hadn’t thought of that’ or ‘We’ll have to look into this further’. And you had to keep asking him questions.”
But that was very the success of the session! Through asking these serious of questions, we were able to discover things that were unknown or unclear. The questions are part of the discovery process. Too often, we are afraid to ask questions for fear of looking unintelligent or wrong but if we don’t question, we don’t learn, and businesses (and minds) can stagnate without continuous learning and change.
A good businessperson, like a good data modeler, will always look for areas of improvement and to do that, you need to ask contrarian questions. Today’s economy is driven largely by data and technology, and through this powerful combination entirely new business models are emerging, with traditional business models being usurped. Think Uber vs. taxis, or online retail vs. brick-and-mortar stores. A business may have been running a certain way for decades. But what if it didn’t run that way? Would it be more efficient? What if? How? How many? That’s the language of change, and the language of data modeling.