AI Could Save Your Data Governance Program, but It’s Unlikely

Jirsak / Shutterstock

In the 1980s, there was a flurry of movies about robots coming to imprison or terrorize humanity. Forty years later, almost every business and technology publication seems to have reimagined the army of robots and artificial intelligence as trading their quest for world domination for the exciting world of business processing. It’s unlikely that most organizations have their business processes and workflows documented well enough to train the army of robots. In fact, the lack of documentation for how things work will stump the robots as much or more than it challenges the humans. For new data governance practitioners, there can be a tendency to rely too heavily on the promise of technology, AI, or machine learning to solve problems caused by sparse or missing workflows.

For those who insist on hiring robots to save your data governance program, make sure your workflows are well-documented and can answer these three questions.

1. Who is the decider and what do they decide?

It’s crucial to know where the deciders are in the workflows. Whether considering sharing data externally or internally, classifying data as sensitive, or purchasing a tool, it’s crucial to know who the deciders are and what they can decide. Launching and maintaining a data governance program is about convincing people with a limited amount of time why investing in data governance will save them money, make them money, or save them time. It wastes time and money routing decisions to people at the wrong level. So, document the workflows that require deciders.

2. Where are the decision points?

Which parts of the workflow are standardized and require no decision? These are the places where robots and AI come in handy for automation. Every node in the workflow cannot require a consensus or convening of the Data Governance Council — this would take too much time. Most parts of a streamlined workflow should not require a decision or consultation. However, the decision points should be clear.

3. What signifies the end of a workflow?

There must be a beginning and an end to every workflow. (What about the data lifecycle?) Yes, everything is a lifecycle, but even the lifecycle has phases because things should end. Workflows can end with the passage of time. For example, the decider may have 10 days to decide something. The workflow cannot linger indefinitely because the decider didn’t decide. Good workflows let people know what happens if the time expires. Perhaps, the workflow resets, or someone is sent a notification to take another action. However, there must be a beginning and an end.

For all the humans in the loop, I hope the prospect of turning over many of the monotonous administrative tasks of well-documented data governance workflows will not scare you from continuing to document. Data governance programs rely on good documentation and good people, much to the chagrin of the 1980s robots of my childhood. AI is poised to be a powerful tool to support people doing the work of implementing data governance in organizations big and small. However, to get the most out of AI tools, most organizations will need to prepare by documenting processes and procedures and/or developing training sets. Buying AI in hopes that the act of purchase will save a data governance program in lieu of investing in the people power is likely to have adverse effects.

Share this post

Carmen Robinson

Carmen Robinson

Carmen Robinson is a senior principal consultant at ABS Global Government Solutions. She has over 15 years of experience in data wrangling, data governance, and data risk. Follow her on LinkedIn.

scroll to top