Is a Data Warehouse Really Your Best Next Step?


My wife is a talented crafter.  She can take bits and pieces of items we’ve used around the house and transform them into decorative gold.  Every time I turn around, she is repainting an old table or wrangling some wire mesh from the garage into an old window frame to hang earrings on.  It’s amazing to watch, but recently she may have met her match when we purchased a Cricut® (pronounced like cricket).  This little gadget, that looks a lot like a digital printer, can do several amazing things, including cut vinyl letters and create “thank-you” cards, just to name a couple.  While this little unit promises to raise her crafting to the next level, the learning curve is quite steep.  Lately, my wife has spent more time learning new software than she has completing a craft project.  I’m afraid this unit will soon become a pricey paper weight if she is not able to integrate it more quickly into her creative process.

This article isn’t intended to be a review of the Cricut® machine, but to question whether an Enterprise Data Warehouse is the best next step in your credit union’s data journey.  Whoa!  How did we get from crafting to data warehousing?  Well, like my wife’s purchase of a machine to support her crafting, many credit unions are purchasing or building data warehouses to support their data journey.  But, like my wife’s experience with her new tool, some credit unions are finding the data warehouse project to be a distraction from their primary goal, which is data visualization.  So then, how do we know the best time to incorporate more sophisticated tools into our creative process?
Allow me to take the analogy one step further.  Just the other day, my beloved had spent several hours at her computer when she said she was going to be crafting.  Puzzled, I asked her if she was having trouble and, sure enough, I found myself involved in a crafting project.  What she was attempting to do was use the Cricut® to cut container labels from a 12 x 12 sheet of paper with a colorful print AND write a content description on each label.  However, she was having trouble learning how to use the design software and it had consumed much of her afternoon.  Of course, with my design software experience, I was able to quickly solve her problem and coax the machine to produce what she wanted, but there was a point at which I wondered aloud whether it would have been simpler to cut the labels from the sheet using scissors and write on the labels by hand.  I ended the day with a net score of 0, plus one for helping, minus one for my ignorant comment.

A data warehouse for enterprise data management and business intelligence, like the Cricut® for arts and crafts, is an excellent tool.  But, should it be the first tool?  I’ve spent the last several months with credit unions that have stated that they either have a data warehouse, or they are building a data warehouse.  I’m usually called in to conduct some sort of analysis or for building some visualization of existing data.  But, when I ask the credit union to provide me with the requisite data to perform the analysis, I encounter a roadblock.  Either the client can’t provide me with access to the data, the data warehouse does not have the requisite data, or it is not clear how to get to the requisite data out of the data warehouse.  One should understand that each of these roadblocks above suggests that the credit union does not, in fact, have a data warehouse by its most fundamental definition.  Data warehouses should provide users with easy access to normalized data for real-time analysis.

So, what is really happening?  What is really happening is that credit unions are being sold a database architecture or one is being deployed by internal staff, but a database architecture does not constitute a data warehouse and a database architecture without a plan or strategy is useless.  This is discussed in another article, found here.  But ultimately, the issue is that the credit union has purchased or invested in a sophisticated tool without really knowing how the credit union would use the tool and who in the credit union would be responsible for making it work.  The real issue is that the credit union has not developed an Enterprise Data Strategy.  The key elements for creating an Enterprise Data Strategy are listed below but allow me to quickly return to my previous analogy.  If my wife had strategized her labeling plan, she might have concluded that using scissors and markers might have produced the same results as the Cricut®, in less time.  Never let the technology drive the strategy.  The strategy should drive the technology.

Stakeholder Interviews

Stakeholders are those in the organization that will require data.  This may be someone on the executive team that wants to see a dashboard of daily statistics, or this may be someone on the front-line staff who wants to see how they are doing on their goals.  It may be a Marketing Analyst who is trying to discover a target segmentation for a campaign, or a lender trying to determine decision rules for an automated decision engine.  Stakeholders are throughout the organization and you want to interview them for two reasons; One, you want to know what they need, and two, you want to know what they have to contribute to the overall data set.  Every department has tools that are collecting data and their knowledge is valuable when creating an inventory that can be integrated into the enterprise dataset.

Inventory Data Sources and Reporting

There are obvious data sources in the credit union, the primary one being the core data processing system.  But, there are other data sources, like a Loan Origination System, external data services like a credit bureau and offline data sources such as spreadsheets used to track member contacts.  All these data sources need to be inventoried and mapped in the enterprise data model.
Current reporting is an excellent way to identify requirements.  It is interesting that the same measures and dimensions seem to be used throughout the organization for different disciplines.  For example, Lending may use loan delinquency by credit tier to inform underwriting guidelines while Finance may use the same information to determine pricing and profitability.

Identification of Key Performance Indicators (KPI’s)

KPI’s are the diagnostic measures we use in any business to determine success of failure.  Profitability is one essential KPI, but other KPI’s help us to diagnose why another KPI may be out of line or predict what might happen next.  For example, if our delinquency KPI is increasing, we should predict a rise in credit losses.

Data Dictionary

A Data Dictionary may take many forms, but a Data Dictionary is required to create a common data language for the organization.  For example, one system might have a field called “Contract Date” while another system has a field called “Origination Date”, both representing the date a loan was originated.  A Data Dictionary would define one term to describe this data and that term and definition should be enforced throughout the organization.  In the end, you shouldn’t have different reports using the two different terms above interchangeably.  This creates confusion and increased lack of trust.

Short and Long-Term Goals

A common mistake by organizations is trying to do too much at the outset of their data journey.  This may explain why an organization has a data warehouse without any real idea how it might use it.  Realistic goal setting is critical for long-term success.  Instead of goal one being creating a data warehouse, perhaps it might be creating and deploying a simple Executive Dashboard.  As the organization begins to experiment with some of these projects in the short-term, it gains a better understanding of where they want to be in the long-term.

Once your organization has an Enterprise Data Strategy, you are now equipped to engage outside vendors if necessary, because you have a plan and priorities.  You now know what your priorities are, and you know what needs to happen first.  This will enable you to differentiate the strengths and weaknesses of different technology offerings.  If you think you might want some help with your Enterprise Data Strategy, feel free to reach out to us at ds@cubi.pro or visit our website at www.cubi.pro.  We can help your credit union develop an Enterprise Data Strategy.

Comments

Popular posts from this blog

Small Market Credit Unions May Have Greater Impact on Field of Membership

How to Estimate the Impact of CECL on ALLL

Odds Are, You Have Never Seen One of These