How to Clean Out the Data Junk Drawer

Disorganized office supplies and knick-knacks in a messy junk drawer

I was looking for take-out menus recently in a drawer that has held these and other assorted items collected over the years. Turns out, everyone in my family had started using that drawer to keep all sorts of other stuff since I first commandeered the prime kitchen drawer space. What was once a neat and organized spot to store useful things had become too jam-packed with everyone’s disparate items to be of value to anyone.

This reminded me of some of the data challenges asset managers face. Databases across the industry are riddled with junk. Firms are loading expensive data that people aren’t using into structures ill-suited for evolving needs, incurring costly processing time and maintenance, and making it very confusing for end users to efficiently find what they need when they need it. Confidence is falling and data consumers are often left to seek alternative paths of lesser resistance to get what they need outside of established and/or implied standards.

I recall a conversation with a chief investment officer who was frustrated by some internally generated reports, each telling very different stories about portfolio duration and offering conflicting recommendations for him to decide upon. Turns out, there were four unique duration data elements in their reporting data mart and the analysts didn’t have details available to differentiate between the fields. The relatively small differences in the individual security level duration values caused varied outcomes in their aggregated findings. Each of the four duration fields may have had full coverage and been completely accurate, but without knowing the right fit, the exercise failed, despite the efforts of well-intentioned professionals. The result was an angry CIO who was left to determine the best course of action to take on the portfolios during a time-constrained rebalancing period and confused analysts who thought they had done the right thing—not a good spot for anyone involved.

Some solutions:

  • Organize the key data assets to make them useable. This includes producing proper supporting documentation such as data taxonomies, definitions, and lineage so users can easily identify the right data for the right situation. If the analysts had a document or a clearly identified person (“owner”) to guide them to the right fitting data for their research, the resulting findings would have been consistent and pointed to a clear path forward.
  • Discard what is not essential. Stop paying for unnecessary data and investing effort into capturing and maintaining data with negative ROI.
  • Take an honest assessment of your data governance model. Ensure you have the right people at the wheel and processes in place to not only prevent the utility drawer from becoming a junk drawer (hint: metrics are essential to monitoring data health and identifying issues requiring remediation), but to also educate and prepare the organization for the future

Don’t kick the proverbial can and wait until spring for cleaning needed today! If you do, you’ll only delay the inevitable and make the job that much bigger and complicated down the road. (FYI—after discarding the non-essentials, we now have wall bins for papers and the drawer is back to its pristine state holding menus, rubber bands, pens, and bottle openers!)