When Usama Fayyad stepped out of the chief data officer role at Yahoo! he said, "Never again." Never again would he take a position where the work was so demanding and thankless and the challenges were so difficult.
But six years later he returned to the role, this time for Barclays Group. The pay was one reason for his change of heart, but also, the shifting landscape of the big data world made his return smoother than he would have thought. At the RapidMiner World user conference in Boston, he shared his perspective on how the position of the chief data officer, and more generally, the business, has changed.
The main changes he has seen are that business groups today are less siloed and more willing to work together around common data sets. Top executives are also realizing that data has to have a voice in the C-suite.
Usama FayadChief Data Officer, Barclay Group
"There's a fundamental realization that data has value," he said. "To [analysts], it's obvious, but to many businesses it hasn't been."
But that realization alone isn't enough to improve business processes. There are certain processes organizations must put in place to make effective use of their data and analytics, Fayyad said. IT departments likely won't be able to handle all of these tasks, which is why it often takes a chief data officer. His suggestions:
- Data gains value when coalesced and loses value when siloed. Bring together all your departments' data into one database and give everyone access.
- Bring data together at its source. Cleaning up messy data in a hundred different spreadmarts and applications is a costly waste of time. Getting the source in line prevents this.
- Standardize everything. The only way to unlock cross-department value is for everyone to speak the same language.
- View data as a core competency, not a sideshow. If the data tells you something, listen. Don't use it as just another part of the decision-making process.
The last point in particular is beyond the skills of IT, Fayyad said. It requires a broad view of an organization's problems and goals to insert data-driven decision making into the heart of specific operations. But IT typically has a better view of what's happening in the trenches, dealing with individual problems as they arise. The chief data officer role is better suited to handling the strategic aspects of analytics projects.
Strategy is also important when it comes to buying technology. Fayyad said there are many vendors that can handle the various tasks associated with data management and analysis, but not every business needs every tool. Sifting through vendors and tools to find the right ones is daunting.
"It's my job to understand these [tools] and I don't even understand them all," he said. "Even the experts don't understand what's happening here."
Sometimes it may take some trial and error. Businesses need to find what works best for them, Fayyad said. During this process organizations can't get distracted by the latest cool thing. Just because something is getting a lot of attention in the media doesn't mean it will deliver business value to every enterprise. Value should be the ultimate deciding factor.
Dell's chief data officer talks data-driven culture
What's the division of labor between CIO and CDO?
How the CDO and CIO can get along