Executing Advanced Data Analytics - Do's and Don'ts

Credit unions are more data-centric now than ever before. They emphasize collecting, storing and harnessing data, as well as laying strong data-based foundations. To make decisions that are member-centric, and to be able to take actions where they are needed most, credit unions need to have a relevant analytics strategy and the capabilities to execute it in an efficient manner.


In this article, we discuss the different approaches to consider and use when executing an advanced data analytics strategy and the pros and cons of these approaches. The insights are drawn from a discussion of panelists at a webinar on Executing your Advanced Analytics Strategy hosted by CULytics. The panelists included Lee Brooks, SVP, Enterprise Data Analytics, Virginia Credit Union; Ken Kondo, VP of Innovation and Software Development, Redwood Credit Union; Scott Kaylie, Data and Analytics Leader, Solarity Credit Union; Shubhankar Jain, Advisor, CULytics; and Troy Del Valle, VP, Business Intelligence, Hudson Valley Credit Union.


Foundation to have before investing in artificial intelligence and machine learning.


Data is a tool and not the end goal. Here are some key actions you can take to build a strong foundation:


  • Focus on four main factors

  1. Data warehousing

  2. Business intelligence

  3. Data science

  4. Data governance

  • Data warehousing is very key for any predictive behavior. It helps to build a rich history that the credit unions should build upon. Propensity pay models and check fraud models are some examples of how advanced analytics strategy can be beneficial and drive value.

  • Bring in the right data from the right resources into one data warehouse. This will help you leverage power BI tools and take on various machine learning initiatives.

  • Evaluating use cases for more advanced analytics can provide value.

  • It is equally important to understand what particular data elements mean.

  • Engage with business functions and understand the business value.

Where do use cases come from and how are they prioritized?


How do credit unions put the use cases together? Here are a few examples:


  • Ideas from executives and reaching a consensus between the IT team and C-level leaders.

  • From the direction where the business is headed.

  • From servicing lines, for example the human resources department.

  • Prescriptive and reporting requests.

These use cases are prioritized based on the impact they have on business, and by being inclusive of all business functions. Analyze the member-facing impact, which gives a common language to the entire organization.


How long does it take to build a data foundation? Can it be built in one day or is it a continuous effort?


This is an ongoing effort. With business functions, there are always going to be new data sets, which means they will need to be consolidated as often as possible. One important factor in the time it takes to build the foundation is what you are working on, e.g., whether it is a machine-learning model or a reporting architecture. The process can also depend on the understanding you have the data.


What is the technology stack you should be using? What are the pros and cons?


There are two ways to look at your technology stack: Choose the stack that you believe your team can manage and provide value from or provide the tools to data-savvy employees. Some common tools for a technology stack include SQL, Informatica, SSRS for reporting with Tableau, R as predictive modeling platform, and Tableau for self-servicing dashboards. ETL and BI tools allow you to scale. Informatica is a good tool for loading. Augmented analytics and Data IQ are other common tools. A practice of setting R file templates for machine learning, and implementing power BI for data visualizations and reporting so as to build critical thinking is a unique practice adopted by credit unions.


What are some challenges which have been encountered? What is the way forward?


Some common lessons learned by credit unions are as follows:

  • Bad data will always be there. You have to come up with ways to counter that.

  • Unexpected work can derail progress towards goals.

  • Collecting the data and helping the organization develop an understanding of the process can be challenging.

  • Not being nimble with the process is a common mistake.

Credit unions are now moving towards using machine learning for a self-service approach, data warehousing, and business intelligence. Building self -service dashboards, improving on data governance policies, building predictive models and implementing them effectively are other objectives.


Conclusion


Having business units visualize what they can achieve with data and aligning with them can make a positive difference, and it increases transparency and data literacy. For data foundations, it is important to keep evolving as you move forward. Always weigh the return when you are going after an initiative. There is always room to grow.


Watch the full session shared as part of CULytics Webinar here.


This On-Demand Content and other similar transformation sessions are available as part of CULytics Membership.