The shortage of data science skills continues to frustrate organizations in their quest to become more data driven.
CIO.com’s 2023 State of the CIO research found that data science/analytics is one of the top three tech-related skills CIOs are trying to hire – and 22% said it’s one of the three most difficult to fill.
With shortages likely to continue, IT leaders must optimize their data team investments. A key question many are asking is how to structure a team that best fits their organizations.
The answer should be guided by the higher-level needs of the organization, says Shayde Christian, Chief Data & Analytics Officer at Cloudera. A decentralized approach works best when the business requires high responsiveness with minimal dependency on cross-functional collaboration. On the other hand, a business that needs efficiency to scale may be better served by a central team that provides functions like data governance, platform engineering, architecture, and data engineering to all areas of the business. Heavily regulated industries tend to centralize.
A hub-and-spoke staffing model combines both approaches by centralizing governance while sowing best practices throughout the organization according to the needs and abilities of individual business functions.
“In staffing shortages, I strive to improve data literacy and promote the use of business glossaries and data catalogs, so technology-savvy business users (aka citizen developers) are well-equipped to serve their own needs,” Christian says.
In that context, organizational excellence in data science is more about creating a common understanding and framework for data interpretation than standardizing and consolidating the data assets themselves, Christian says. While those functions are important, they don’t address the bigger question of how to make the culture of the organization more data driven.
External accelerators
Regardless of the structure, the most important goal is to establish a culture that internalizes the value of data and its importance in making smart business decisions. “Give people the information they need about what data means rather than just telling them where to get it,” Christian advises.
Third-party partners can be valuable allies in addressing the need for both talent and tactics but be targeted in reaching out for help, Christian advises.
“I’ll partner when my business asks for data products outside my wheelhouse,” he says. “If I’m predominantly a data governance and engineering shop and I need advanced analytics, that’s a different mindset so I order out.”
Contracting for managed services with well-defined deliverables can be preferable to bringing in hourly labor because the third party is accountable for results. Only contract for skills you need. For example, the data prep work that comprises about 80% of data science projects may be done more quickly and cheaply by internal staff who know the business.
However you partner, ensure that the professional services experts have talent transfer in the contract so your people can learn from theirs. “A managed service contract should absolutely include in-house talent development,” Chrisitan says, “or else you’re running the air conditioning with the windows open.”
For more information on building a data team, visit www.cloudera.com
Cloudera
Shayde Christian, Chief Data & Analytics Officer at Cloudera
Business Intelligence