The Art and Science of Justifying DataOps
For chief data officers and data scientists, the business case for DataOps can be obvious. But for business executives, not so much.
DataOps, correctly done, can streamline data workflows, reduce errors, and offers transparency to the entire data operations. It improves efficiency, increases data trust, and gives more time to do analysis. No more time-consuming rollbacks, report delays, and questions about the validity of the insights.
For business executives, such benefits are not immediately apparent. So, getting the budget to build your DataOps can run into snags — right up until a business problem challenges your company’s core value proposition.
That’s what happened for Screenrights.
Data flood alert
Emma Madison, head of new business and technology at Screenrights, faced a challenge that all companies face: rising data volumes.
For Screenrights, a non-profit organization in Australia, the challenge the data deluge posed was it slowed down the ability to link royalties to the programs.
“It is a core function. We need to do it efficiently, and the high volumes of data were making it challenging to link royalties to programs,” says Madison.
It created a backlog that never seemed to go away. It also impacted business efficiency and overall business performance.
Madison and her team had a rules-based algorithm. But they soon found out that accuracy suffered in huge volumes.
“It was a prediction problem. So, we started to look at machine learning and asked similar organizations for potential solutions,” Madison recalls.
It is around this time the company came across Tamr.
The unintended DataOps adventures
DataOps and data mastering was not the first thing on Madison’s mind when she began the machine learning route.
“I did not think of the data mastering or DataOps challenge at all. We started building the business case for one problem based on the costs associated with it,” Madison says.
The challenge with data science is solving the immediate problem is only half the solution. You also need to do it cost-efficiently, not just cheaply. It means Madison had to widen her business lens to see other associated problems.
“With advice from Tamr, we quickly found out that we had other related problems to solve. So we focused on the three big ones,” Madison shared.
Ultimately, what Screenrights needed was a solid foundation and system to master their data so that they could unify and accurately understand it.
For Madison, it is where the human-guided machine learning DataOps approach from Tamr made business sense. “We can train the system so that it can learn how to master the data at scale,” she said
For Screenrights, an efficient DataOps now underpins their business model. In addition, it reduced data discrepancies that created problems with content owners.
“So, for an organization that relies on accuracy, having the agile machine learning approach from Tamr fit well,” says Madison.
With the benefits identified, Madison did an extensive cost-benefit analysis. “We focused on the problem and talked very little about the technology. And when it came to the senior leadership, we also highlighted what our teams would experience using Tamr and showed real results after limited rounds of training using Tamr,” Madison describes.
It’s what convinced the Screenrights senior management to come on board.
Everyone matters
Madison is still exploring the various benefits of data mastering. But it is a learning experience that her entire team is going through.
A significant benefit that Madison is following is rising data trust.
“The team needs to trust the suggestions. The human-guided machine learning approach allowed them to review these suggestions and measure the accuracy. This also allows us to quantify the risk,” says Madison.
Whatever the motivation to solve a data problem, Madison advises not to start the business case as one.
“Instead, articulate what the risks are for the organization if the data problem persists. When there is a better understanding of the problem, and it is articulated that ‘Hey, here is a solution,’ it enables a healthier conversation,” she explains.
It also gets everyone on the DataOps journey. “So everyone is learning how to work with the new system together. Over time, non-technical people can learn the system and understand how it works, which is critical to tackling future data problems.”
Winston Thomas is the editor-in-chief of CDOTrends, DigitalWorkforceTrends, and DataOpsTrends. He is always curious about all things digital, including new digital business models, the widening impact of AI/ML, unproven singularity theories, proven data science success stories, lurking cybersecurity dangers, and reimagining the digital experience. You can reach him at [email protected].
Image credit: iStockphoto/Rawpixel
