What are some of the actions your organization take when operationalizing data governance?

Robert S Seiner

What are some of the actions your organization take when operationalizing data governance?

What are some of the actions your organization take when operationalizing data governance?

My Dataversity webinar today just focused on this subject. I would love to hear from you about what actions you are taking to get your program "off the schneid".


Robert S. Seiner

KIK Consulting & Educational Services

The Data Administration Newsletter (TDAN.com)

[login to unmask email]  @RSeiner  @TDAN_com


Charles Harbour

RE: What are some of the actions your organization take when operationalizing data governance?
(in response to Robert S Seiner)

When starting from scratch, I think it's very helpful to get a business partner involved.  In the spirit of non-invasive DG, you don't have to scare people away/risk the wrath of a micromanaging boss by calling them a steward or a SME or an owner (even if those terms apply).  The idea is to start on that data quality feedback loop - again, you don't need to call it issue management, or spend a lot of time defining workflows in tools that may or may not work well with DQ issues.  But pick something, something of value to the business owners, and work the problem.


As you work the problem, you will undoubtedly uncover gaps in responsibility, ownership, accountability, more.  Depending on the impact to your little Skunkworks project, you maybe don't need to solve all of these problems in order to make progress on your DQ improvement task.  You probably don't want to focus on a problem that exists in the source system - for that, you need buy-in and budget from the system owner, and is not recommended for these early issues.  Maybe pick something that has value - say, a non-mandatory field has high value, but is filled in by the field reps/call center staff only half of the time.  You could create new training materials, review the value of this field with the staff (educate them on why it's important), maybe track who fills in the most in a given week and send them cookies (no, I'm not making that up).


As a DG practitioner, you should baseline the completeness of this field and then track it over time - are people complying?  Can we send a report to each manager that shows their employees' completion percentage, and ask to follow up with troublemakers?  Can we demonstrate the value of having this field filled in (say, if this is a loan contract, what is the time to deliver to underwriting?  Did that time go down?  Did it go up (and why)?).


Now that you have results (and a business champion), then is the time to report to all of the stakeholders and their managers about the value realized by performing this task.  And maybe it's not truly productionized (need to get enough of these to justify a governance council and your own budget), but you have demonstrated value, you have mid-level management buying into the concept, and you've fleshed out at least one area where you can make an impact (maybe it's just training, but could be something like an automated business rule/lookup that repairs simple data errors without human intervention) and solve real data quality problems.


A real life example was working with the department of transportation on the problem above - field reps weren't filling in a key field because it wasn't required in the application they were using.  We updated the existing training materials, retrained existing field reps, sent a weekly report that showed each reps completion rate, as well as a trending report that showed overall completion percentages - and we tracked the amount of time-to-complete for each mini-project.  For straightforward cases, the time-to-complete went down, but what we discovered was that the field reps weren't filling it in deliberately on complex cases (and in this case, the time-to-complete actually went up, since they were spending additional time researching). 

This was helpful information to the business process owners - they weren't even aware that this was an issue.  So even though our KPI had mixed signals, the additional insight then created our next task (and at that point, we did consider this productionized, in that we initiated the backlog and established a strategic review to prioritize that backlog).


And yes, we did send cookies.