Why Analytics Might Not Be Working in Your Credit Union
It has become clear in business, and proven in science, that data-driven decisions can often yield higher efficiency and accuracy, especially in areas where we do not have established expertise. However, there are still many credit unions who are struggling to get their data working for them. In this article, I will discuss some of the reasons this may be true.
Let’s start with some key terms that need to be given some clear definitions:
Big Data – This term is used at increasing frequency, almost like “Big Oil” or “Big Pharma” are used to describe the Oil and Pharmaceutical industries. Big Data is not an industry, although there are some who are trying to make it that. Big Data refers to the velocity and variety of data, and in its purest since, is referring to data outside of your credit union, as well as data within. Think about it this way; data is being collected from electronic devices and sensors everywhere. In theory, all this data could be integrated and used in decision making. The truth is, as credit union, many of us do not have a real good handle on our internal data, much less external data. So, using a “Big Data” tool to make data-driven decisions at this point, is like using a flame-thrower to light a candle. Way too much learning required to achieve limited results. That’s not to say that Big Data couldn’t or shouldn’t be used, but that we need to tackle and master our own, small data first.
Data Scientist – As a trained data scientist, I cringe when this term is used incorrectly. Data Science is a combination of three important factors: technological skills; mathematical skills; and business subject matter expertise. Too often the term is applied to a data analyst who builds reports using existing metrics and tools. A Data Scientist, on the other hand, understands your business and what drives success and can test hypotheses with custom metrics and proxies, providing decision-makers statistically significant measurements for consideration. The Data Scientist needs to understand, not only how to build data visualizations, but how to tell a story with data. If you have a true data scientist on staff, you know it, because they were most likely difficult to find, and they cost you a bit more money than the average data geek.
Data Warehouse – A data warehouse is an organized database structure that integrates data from disparate data sources into a centralized source of facts – let’s not say truth at this point. The key here is that the data is organized and integrated. The data warehouse must also contain modeling that allows business users to access data without assistance from IT. These principles were defined in the Data Warehouse Toolkit, written by Ralph Kimball, and have become the road map for data warehouse architecture. If you have a database with random tables from disparate sources that is not normalized or connected in any way, this is not a data warehouse. Perhaps if we don’t want to call it a data dump, we can more tastefully call it a data attic or garage. Think about the organization in your attic or garage and the organization at an Amazon distribution center. A data warehouse is more like an Amazon distribution center than your attic or garage.
Once we have the Big Data buzzwords out of the way, we can focus on the real issue, which is the absence of a data-driven decision culture and/or strategy. I’ve seen many credit unions seek to implement analytics tools to solve a single problem, let’s say member marketing, loan portfolio analysis, or addressing regulations like Current Expected Credit Losses (CECL). As they begin to implement these tools, they soon hit what I have dubbed the blind curve of data science – data integrity and disparity. In other words, as we attempt to build decision models, we realize we don’t have all the necessary data and that the expense to organize and collect data for a single discipline outweighs the value one would get in the end. The data was either not collected, not accessible, flat out bad, or is in the hands of a third party. Some credit unions encounter the blind curve and sadly miss the turn altogether and fall into the abyss of an implementation cycle that lasts for months or years. Others push through the curve but, on the other side, realize they have lost some important cargo and some or many of their original objectives are not met. Finally, there are those who see the curve and turn back, ending their journey into data-driven decision making, for now.
None of the above scenarios need happen, if the credit union understands the principles above and starts with a strategy, not a destination. The first part of strategy making is to include all stakeholders in the beginning. Data projects run much smoother if one has buy-in from all the data gate-keepers – the people who collect the data – and stakeholders throughout the organization can see the value in data. Second, your team should define what the future state looks like when decisions are made using true and verifiable data. And third, you must give your team permission to use data to make decisions. There is nothing that will kill data projects faster than an executive who insists on making intuitive decisions in the presence of contradictory data.
If something in this article resonated with you, it may be time to sit down and take a close at your credit unions culture and strategy to see if it supports data-driven decision making. If not, you may want to contact us at CUBI.Pro for an initial free consultation. We could help you create a data strategy at your credit union that will support a data-driven decision making culture.
Comments
Post a Comment