It was a sweltering summer day. I had big plans for riding my horse with friends which involved loading my young horse, Smoke, into our new horse trailer and then driving an hour north to the mountains. I knew that once we arrived at our cool and shady destination, my horse would be quite happy on our adventure. But Smoke was not convinced of the soundness of my plan.
At first, he was annoyed at being pulled away from his barn, then he became scared and then angry. No matter what I did, he just would not walk into the new trailer. He weighed 1200 pounds, so without one of us being injured, I realistically could not get him into that trailer unless he wanted to go in.
In my mind, this scenario is a lot like getting a group of data owners and users to go along with a new Data Governance initiative. As the project lead, you know that if they cooperate, and invest their time in meetings and training, then their data will be much better, and the company will run more efficiently in the long run. However, the truth is this: most teams won’t go forth willingly with a plan like that unless they WANT to. And if you try to force them — well, your career might be injured in the process.
So, what’s the trick to getting horses and people to do new things? Take small steps forward and make sure they have some input in the process.
Over the last decade, I have come to understand that Data Governance is a collaborative and continuous improvement process. It is not just about purchasing great software and training users. It is really about MANAGING CHANGE.
Data Governance changes the way the business manages and interacts with its data, and a successful implementation is about communicating, leading, and listening.
- Listen by conducting workshops and asking questions to understand how the business does things today. Ask them how dirty, redundant, or missing data and processes are impairing their work today.
During this discovery stage, listen for and acknowledge any fears they might have. Respectful listening and acknowledgment is the key to turning initial negative statements into baseline starting points for planned improvements.
- Lead the business by helping them understand the upstream and downstream impacts of bad data. People want to know “why” are you asking them to change.
Next, solicit their help in the development and execution of the new processes. This creates “ownership” in the plan. It’s less likely for a team to be resistant to an initiative they feel they had a hand in creating.
And be sure to speak respectfully of participants’ efforts. Ignoring or being dismissive of people often leads to stonewalling and project sabotage.
- Communicate clearly and frequently about the project’s initiatives, activities, tasks, and timelines. People need time to process new concepts – so don’t spring new things on a team overnight and expect instant obedience and cooperation the next day.
Unless you have great buy-in and a groundswell of support from the very start, it is best to start your governance initiative with one or two smaller projects with quickly achievable ROI’s.
Small but measurable successes followed by praise and publicity (“talking it up”) get others excited about joining your successful program. Avoid using the old-school model of big-bang IT-driven projects spurred by political pressure.
In the end, it took two hours to calm my horse down and convince him that it was OK to walk into the trailer. Had I made the effort to get him comfortable with the new process in short sessions ahead of time, I might have been able to rendezvous and ride with my friends that day.
So instead of forcing a team into a new uncomfortable place with a single large goal at risk, invite the business to participate in a series of smaller ROI project successes as a part of a larger program.
Once the business has experienced a taste of success, they will move forward as a willing partner, eager for new opportunities to shine and help the company become more efficient and profitable through improved data quality.
Daryl Crockett is the CEO of ValidDatum,
