Cooperative data analytics through a semantic layer

On a recent trip to Malaysia, I was able to play basketball with my brother-in-law (my wife is Malaysian). As we began the game, I realized that we were all relying on a single source of truth for the rules of basketball.  Even though we are from very different parts of the globe, we were operating under the same definitions of the rules of basketball. Imagine if we all began playing according to sources of the truth that dictated different ways to play basketball. Maybe my source of truth told me that I don’t need to dribble to play the game. The other team’s source of the truth dictated that they can tackle the other team. This game would end horribly and would probably escalate into a conflict quickly. The same is true for data analytics within the credit union movement today.

Different Sources of Truth

Within most credit unions, there are many different sources of truth. Marketing departments have their sources, Accounting has theirs, and Lending has as many sources as types of loans (i.e. credit cards, mortgages, student loans, etc.). Over the course of time, every department begins establishing their own language based on their sources of truth, which are usually centered around a specific source system. For example, the marketing team has an MCIF system, which has an abundance of data regarding households and members’ profiles. The lending department relies on its loan origination system, which displays information found within a member’s loan application. All the while, the contact center relies on their CRM, which houses data collected during calls with members. When there is a need to work together to accomplish a goal, these various departments come to a meeting speaking different languages and using a separate understanding of the rules of the credit union. Like the game of basketball without a common source of truth, the project or initiative often ends horribly.

Data Availability

In a previous blog (The Purpose of Analytics), I elaborated on the double-edged sword of data availability and access. As data access continues to get easier, credit unions’ employees are beginning to face “information overload” and are having trouble getting a clear signal in all the noise. This is resulting in many credit unions believing that they are “doing data analytics” when they are simply building disparate data siloes that are full of information but not integrated into one single source of truth.

Rogue Analytics

No, it is not a new movie coming out of Hollywood (although it might be soon), “Rogue Analytics” is the current state of most credit unions today. There are thousands of “analysts” throughout the credit union movement running around and developing reports that display data in various fashions. Unfortunately, these reports contain very specific rules that are heavily influenced by a source system a department relies on.

Establishing Common Rules: The Semantic Layer

For any sports team or credit union, to work together effectively, a common set of rules must be established. This set of rules should not come from the software systems used by credit union employees or from within a department. A common set of business rules must be established in the mold of the credit union movement. This common set of rules does not change when the credit union converts their core processing system, swaps out their mortgage origination system, or purchases the latest online banking platform. The layer that sits on top of the data analytics platform that the credit union uses to integrate all their different source systems is called a semantic layer.

A semantic layer produces data in clear business terms that are not specific to a database or department. It is a common set of definitions around data that all departments and teams work from when developing reporting and analytics. Especially when working in teams, the semantic layer provides a common set of rules for everyone to work from.

Credit Union Islands

Unfortunately, even if one credit union creates a common language for their analytics internally, they won’t be able to play with other credit unions because they don’t speak the same language or use the same set of rules. To establish a single source of truth that fosters collaboration throughout the credit union movement, we must come together to establish a common language that can be used throughout the entire movement. This will be essential for credit unions to compete with the megabanks and fintech competitors.

Cooperative Analytics

Once credit unions can establish a single source of the truth for the entire credit union movement, they can truly collaborate effectively. Through a common industry standard data model, an analytics platform can be established that will empower credit unions with a common language to use as they work together to further the credit union movement.