People data is inherently messy. Many people touch each record, and every hand that enters data is apt to make mistakes.
In all of our projects that require data migration, we include data cleanup as a project milestone. We always attach a caution – cleanup should apply only to mission-critical information. Spending money on data that doesn’t impact the business makes little sense.
One of the most important decisions in any implementation is whether to migrate data or leave it behind. Before making that decision, we recommend careful analysis of the costs and benefits.
Most times, it is an easy decision. In recruiting, there is no need to migrate candidate data if you close out requisitions in the old system and start over in the new one. Where you have extensive employee profile information or training history, it might be a harder decision.
What matters is how important the information is. For example, in an employee profile, how important is it that the names of institutions in an employee’s education history are exactly correct?
How much money, time, and energy do you want to spend on correcting bad data?
Thomas C. Redman, the “Data Doc,” says focusing on cleaning up past data isn’t the right way to go. We should think about today’s data more than yesterday’s. We agree with Redman: how much does cleaning up bad data matter if we will go on creating bad data? Will it really help if you leave bad information in an old system but don’t change your data practices?
The responsibility for good data lives with the people who create it, yet nobody knows about bad data until they use it.
So, data users spend 50% of their time correcting bad data or validating data they don’t trust [1]. Even if it is half of that estimate, it is a lot of wasted time and lost productivity.
What is worse is the impact on decision-making. When you don’t trust your data, you go back to the old ways of deciding on gut instinct. Unless you do something about bad data, your organization will never achieve the ability to make data-driven decisions.
Dr. Redman recommends getting data creators and data users together. When data creators realize the impact of bad data on the organization, they can take on the responsibility of identifying root causes and addressing them.
Sometimes the solution can be astoundingly simple. Working with an accounts payable group on a problem with returned mail to ex-employees, we found the problem to be a matter of the size of the window on envelopes. HR was using a 36-character limit in address fields. AP’s check writer system truncated addresses over 30 characters so they would fit the envelope window. Fixing the check writer system would have been expensive, but limiting address fields to 30 characters in the HRMS was cheap and easy.
Our recommendation is to get data users, data creators, and software platform managers together to create a company-wide data governance regimen. Change the conversation about data accuracy from cleaning up the past to getting the present right so you can build the future.
Reference:
[1]. Redman, Thomas C. "Data's Credibility Problem." Harvard Business Review. December 01, 2013.
PhenomeCloud is a comprehensive technology solutions provider committed to empowering businesses to overcome challenges, enhance their workforce capabilities, and achieve superior outcomes.