Ownership is only part of the data quality equation, though. An organization also must have processes to control the data. It must devise processes to
monitor and uphold data quality, such as controls that prevent modification
of data or injections of false or incorrect data. In addition, it needs processes to
remediate issues as they arise, cleanse data, move data among different sources,
convert data into a single usable format, and document data lineage (the life
cycle of data, including where data is created and where it travels over time).
The employees were tapping different sources to reach their answers. The
core system had an “active status” field and would send a notice to opera-
tions when an account closed. Operations would update a spreadsheet of
parties no longer allowed to do business with the organization. Some of the
queried staff members based their answers on the core system, others used
the spreadsheet, and still others defined active accounts as those that had a
transaction within the last two years. Disparate processes led to disparate
answers and unreliable, untrustworthy data.
In many cases, data quality comes down to how well the people on the
front lines are trained in and adhere to the established processes. After all, no
technology can address underlying human issues. One bank had a business
process that allowed tellers who knew a customer conducting a currency
transaction to check a box on their computer system and bypass the usual
process implemented for current transaction reporting. But the tellers often
could click the “known customer” box regardless of whether they actually
knew the customers, and therefore failed to capture the required information
about them. Other institutions continue to struggle to collect data for the
new fields for Home Mortgage Disclosure Act (HMDA) reporting, causing
significant challenges across the industry. Without proper data, not only
will technical HMDA violations occur, but data analysis for fair lending
and Community Reinvestment Act purposes also will be affected. There is
nothing the most advanced AI tool can do to remedy such shortfalls.
Data governance also should include data validation. It is one thing to have
the proper people, processes, and tools to move data while maintaining reliability.
It is another to make sure they are working as intended and that the data is accurate and collected, handled, moved, and mapped properly. Financial services
organizations must regularly undergo independent reviews to confirm this.
The Big Picture
The value that AI and automation tools can bring to compliance obligations is clear. But, as some financial services organizations have already
discovered, having the will to get on board gets an organization only so
far. Without the available data and the people, processes, and tools needed
to maintain its quality, automation efforts are unlikely to ever make it off
the ground. ■
Similarly, governance includes getting everyone on the same page. At one
financial services organization, seven staff members were asked to provide
a list of the total active accounts—and each returned a different list. The
problem was not technical but was with the business processes.
ABOUT THE AUTHORS
JOE DURHAM, CRCM, CAMS is a senior manager with Crowe, LLP., and can
be reached at (616) 233-5624 or email@example.com.
ABA Compliance 2018_half page_FINAL.pdf 1 5/29/18 3:21 PM
CHRIS SIFTER, PMP, is a principal at Crowe, LLP., and can be reached at
(312) 857-7363 or firstname.lastname@example.org.
Schedule Your Free Demo Today
• Vendor Compliance
for a complete Risk Suite!
• Vendor Risk Management
• Contract Management
Reach this advertiser through links.aba.com