5 Pitfalls of Data Analytics Approaches

As more credit unions design and test their approaches to data analytics, a few common traps that slow success are emerging. During their talk at the 2018 NAFCU Annual Conference, our own Tim Peterson and Shazia Manus talked through five of these pitfalls and offered advice for side-stepping them.


  1. Innovating for Innovation’s Sake

    With so much hype surrounding analytics, especially when it comes to artificial intelligence, machine learning and deep learning, too many organizations are jumping in head first without fully understanding how the technology can make a difference for their end-users.

    To avoid this pitfall, sit down with your teams for a strategic level set. Identify one or two key opportunities for improvement, both internally and externally. That is your critical first step, and it comes before anything else – before hiring data experts, before finding a data partner, and definitely before deploying data technology.


  2. Reading Bad Data

    Bad data leads to bad conclusions. It can also empower self-serving biases, which are common in legacy industries like financial services. Leaders who are new to analytics can struggle to identify which of that information is signal and which is noise. It makes sense when you consider Pitfall No.1. If a team isn’t sure of the outcome it’s after, it can’t be sure which data is most meaningful.

    To avoid this pitfall, hire or designate a data translator. Data translators understand the human side of analytics strategy, which includes the temptation to chase shiny new objects. Their role is to keep the credit union focused on the problems it wants to solve, seeing through the hype to the real potential.


  3. Failing to Prep for Data Surge

    Not long after deploying an analytics strategy, leaders begin to see the appetite for data growing larger and larger. Managing the increase in data volume is one of the biggest challenges for incumbents working with legacy infrastructure and disparate systems. To lessen the complications of working with too much information, organizations often try to work with small sample sets of data from sources they know and are already comfortable with. This doesn’t work. Small data sets create misleading insights. When you’re working with small, isolated sets of data, variances are often ignored. Analysts have no way of knowing if the anomaly is worthy of attention. A larger data set, on the other hand, helps analysts see the outliers as part of a pattern with a definite signature.

    To avoid this pitfall, insist that your systems, processes and people are prepared for an exponential increase in data volume. Design a data strategy that is as encompassing as possible, pulling in data from as many places as you can and synthesizing that data around the core problem you are trying to solve.


  4. Neglecting Governance from Day One

    Cyber attacks, data breaches and privacy regulations in the European Union have raised consumer awareness around the vulnerability of personal data. They are paying more attention to the brands, firms, apps and others who have their data. And, they’re asking more questions about how it’s being collected, used and shared. Credit unions have earned consumer trust in a way not many brands today have. That’s why we can’t afford to lose sight of the protections on member data.

    To avoid this pitfall, bake strong data governance into your strategy from Day 1. Consider early how you will manage everything from availability, usability, integrity and security. Make a plan for being transparent with members so they can understand how your use of their data will improve their credit union experience.


  5. Failing to Consider Culture

    Not establishing and nurturing an analytics culture is the surest way to fail. Without the culture to support it, data analytics strategy will die on the vine. Analytics is rooted in exponential technology. To be successful with it, leaders must have a nimble strategy that will allow them to evolve alongside the technology's capabilities.

    To avoid this pitfall, take the time to build, and then set a plan for nurturing, an analytics culture. As you do, consider a couple factors: First, the fail fast concept is key to a culture that understands and embraces data analytics. Because you are working with an iterative mindset, it’s not only okay to fail, it’s expected. Second, inside an analytics culture, learning is dynamic. Learn to experiment and experiment to learn.


5 Best Practices for a Smooth Journey

  1. Outline the problems you want to solve – for the credit union and the member.
  2. Identify the data most important to helping you solve those problems.
  3. Prepare your systems, process and people for growth, ensuring you have the contextual intelligence to avoid misidentifying trends.
  4. Establish strong data governance from Day One.
  5. Build and nurture a nimble, iterative analytics culture.