Credit unions are often plagued by their data being locked up in the core processor and other standalone systems. Increasingly, they are adopting data warehouse solutions to free and integrate this data so cutting edge analytical tools can be employed to solve tough business problems.
However, a new problem then becomes apparent. Once business users realize long-imprisoned data is now liberated, they clamor for information. This puts pressure on the IT department to meet the growing demand for reporting and analytics.
Beleaguered IT executives seeking relief often look to the holy grail of self-service analytics. In this approach, data access and analytical tools are put in the hands of business users so they can do their own reporting and analysis. This seems like an attractive solution to both IT and information consumers. In fact, self-service analytics often is hailed as the height of “data democratization” or further freeing data by releasing it into the hands of the masses.
While this sounds very virtuous, poorly deployed self-service analytics can cause as many problems as it solves.
Data warehouses can serve as the place for all organizational data to be centralized. The goal is to provide a single data source for all analytics purposes. However, most enterprise data warehouses are very complex, interrelated arrangements of files. If semi-trained business users try to query these files without sufficient knowledge, very different results can appear. This is frustrating to the business users and cast doubt on the accuracy of the data warehouse.
There are important steps credit unions can take to a get the most out of self-service analytics while minimizing the risks.
Business-Friendly Data Layer (Semantic Layer)
Rather than expose the complicated array of files at the foundation of the data warehouse, a data layer needs to be constructed on top of these files that is easy-to-understand and easy-to-use. It features data elements named according to familiar business terminology (e.g. Loan Type, Risk Score Group, Branch Name, etc.). This layer also provides frequently used summary data while providing access to the transaction level data. For example, loan balances are rolled up by branch or zip code or any other meaningful aggregation, but the end users can also drill down to the detail.
The Right Tools and the Right Training
With the Semantic Layer, many easy-to-use tools are available to provide powerful analytical capabilities. However, even easy-to-use tools require some level of training to prevent business users from becoming dissatisfied.
Another important aspect of training concerns the data itself. Even though the Semantic Layer seems simple to understand, proper use of the tools requires that users be well-versed in the breadth and depth of the data. For the breadth of data, training is necessary to teach the definitions of the individual data elements. Depth refers to educating users about the different levels of aggregation available.
Know When to Call in the Experts
Even well-trained users with access to a well-designed Semantic Layer cannot answer every question. Credit unions need to manage their expectations about how much users can effectively do. There is a point where an expert third party should be called in to handle analytics and reporting tasks that are beyond the current skills of internal personnel. These experts may have provided similar services to other credit unions so delivery is more efficient and less costly than if the credit union team tried to grind it out themselves.
. . .
Self-Service Analytics has the potential to deliver the promise of data democratization by empowering business users to do their own information exploration. In having the proper data infrastructure in place, IT departments are freed to concentrate on other projects needed within the credit union while empowered end users can carry out the analytics necessary to drive growth and member satisfaction. This promise can be realized more fully if data, tools, and training are effectively deployed to support the effort.