Keys to DataOps Transformation

Video produced by Steve Nathans-Kelly

In his Data Summit Connect 2021 presentation, DataKitchen founder and CEO Chris Bergh discussed the essential steps for measuring and ensuring successful DataOps transformation in your organization.

"Think about how you measure the processes that your teams work on," he said. "Can you measure the problems of your data providers, which ones are good and bad? Can you measure how often you're meeting your SLAs or not meeting your SLAs, how recent your data is, how good your data is, how often you're meeting your build times?"

These considerations are important, especially because in some organizations there are  hundreds of pipelines in production. "And then the second part is the opposite," Bergh said. "Your team is developing new pipelines. How are they getting into production? How well are they collaborating? How many tests they have, what test coverage. And so there's a set of metrics that you can develop."

It is critical to be able to look at this and drive change, said Bergh. "I've worked with data people for a while, and even the most cynical ones will start to believe that you can go fast and not break things when they can see the data to back it up," he noted. "And so gathering that data not only can help your team help the nonbelievers on your team but also help you as a leader look good because you're able to show these metrics and show the kind of progress that you're making as a team. And, you know, there are lots of different ways to measure team productivity and some of which come out of tools like JIRA, some of which come out of tools like our software, but it's a very rich area that I think we're going to see a lot of work in, in the next few years."

In terms of the key steps for doing enterprise DataOps transformation, Bergh said the first is avoiding the "big bang." It is important to start small, get a small team, show value, get customers talking about it and kind of get the social proof, he advised. "Don't ignore these soft things."

It is vital also to help people see this as an opportunity, and not as a way to adjust yet another way to blame them for "screwing up," said Bergh. "One of the ways to do that is just measure the benefits and show that your team is more productive, gets more story points, more feature points, more functions into the data; track your error rates to show that you have much less errors from quarter one to quarter two. And those things can be very concrete ways to show that you're actually making progress on these goals."