Christina Sandema-Sombe, Co-Founder and Chief Data Analytics and AI Officer at Datum Cafe, speaks with Peter Geovanes, Founder and CEO of Juris Tech Advisors, in a video interview about her professional trajectory, the need for cross-functional relationships, fostering shared understanding, and why data fabric is not enough without accountability and collaboration across the organization.

From mailing Excel sheets to leading global data strategies, Sandema-Sombe’s journey through data has been exemplary. Reflecting on her early career, she shares how her first role at CARE International introduced her to data in challenging ways.

At CARE International, she was tasked with building a system for impact measurement and performance marketing globally. Back then, the internet was unstable in the countries the company operated in. To address this unreliability, she bridged the gap by writing technical manuals by hand and shipping those via DHL to field offices with Excel sheets that enabled offline data entry. Field teams could upload data once internet access became available.

Following her time at CARE International, Sandema-Sombe transitioned to Deloitte, where she spent nearly a decade wearing multiple hats and ended her stint there managing taxonomy and reference data. Her journey then took her to Nike, where she was the Chief Data Steward.

Speaking of the consistent thread in her role, Sandema-Sombe says, “I always worked in a global role in highly matrixed environments and always stood in the gap between technology and the business of the organization.”

Now, alongside former Nike colleague Amy Harrelson, Sandema-Sombe co-founded Datum Cafe, where she is also the Chief Data Analytics and AI Officer.

When asked about best practices to break data silos and fragmented data, she emphasizes that it boils down to understanding that “it’s a two-way relationship with your data team.”

It is convenient to receive requirements and focus on delivering value quickly to prove the relevance of the data team, says Sandema-Sombe. But, for a data team to truly add value, support must come from and be extended to other departments.

“The data team is only as effective as the partnerships it has with other organizations,” she says. “And that education has to happen at the top level of the organization.”

Using HR as an example, Sandema-Sombe states that the HR team may have requests to the data team about what they want to do with the data. Similarly, the data team should also be able to ask HR for support in changing certain things to bring the best outcome.

Further, Sandema-Sombe highlights the challenge of identifying data stewards, as the role is not defined across organizations. Similarly, she says, it is tough to get people to prioritize time to collaborate with the data team when the work is not recognized through performance management systems.

In such a scenario, the data teams have to work through several things with each department and serve as enablers or accelerators for delivering real value, says Sandema-Sombe.

Sharing her insights on having a 360* view of data, she states that having a comprehensive view is not effective without understanding the data context. What data quality means to the business side, which often owns the data, may be misinterpreted by the data team if it does not have the business insight.

Elaborating on this, she shares a recent conversation with a colleague who ran into problems with the engineering team. The engineers flagged financial data showing accruals in a future year, which seemed like an error.

In reality, the organization uses multiple calendars, and some of them include future dates. To the business, this was not an anomaly, but rather an expected behavior based on how their reporting calendars work.

“Without that context, it’s easy for a data or engineering team to assume the data is incorrect. But when there’s collaboration, someone can step in and explain it, and then you’re able to do more interesting, accurate things because now you understand how to use it,” says Sandema-Sombe.

Sharing another example, she emphasizes the complexities of working with location data because different parts of the organization had different interpretations of what location meant. One challenge was the presence of employees working remotely in countries where the company had no physical operations.

Another layer of complexity came from legal and compliance obligations tied to office closures. Sandema-Sombe notes that even if the offices had closed, the company had to legally retain legacy location data for seven years.

Addressing these factors, she states that it is critical to understand why the business is doing things a certain way. “The contract between the technology teams and the business teams is critical to getting the right outcomes from your data,” she affirms.

When it comes to modern data infrastructure, Sandema-Sombe acknowledges the value of a data fabric approach but also stresses that visibility alone isn’t the full solution.

“When we talk about the data fabric, what we’re aiming for is the ability to find and see all our data, no matter where it lives,” she says. “But that’s only one half of the equation.”

Even with advanced tools that provide a consolidated view, the real challenge lies in understanding and managing the data meaningfully, cautions Sandema-Sombe. Without context, business engagement, and people accountable for the data, the outcomes will not be favorable.

Expanding on this point, she states, “If you think about it visually, your data fabric is like having multiple rooms. One with procurement contracts, one with legal contracts, and another with HR contracts — all managed in different systems,” she says. “The fabric lets you look across all of them, sure — but when the time comes to act on that data, who’s responsible?”

Next, Sandema-Sombe emphasized the importance of establishing a decision-making body, a stewardship group responsible for this data. She advocates for a shared understanding of how each department uses data, whether everyone agrees on definitions and quality standards, and if cataloging is clear for others to use.

“Without that shared understanding, people will interpret the same data differently — and that leads to misuse.”

In conclusion, Sandema-Sombe points out a common misconception in highly automated environments: that technology can fully replace human oversight. Referring to ChatGPT-powered descriptions, she maintains that auto-generated insights can be misleading if they do not reflect what that data means in a specific context.

CDO Magazine appreciates Christina Sandema-Sombe for sharing her insights with our global community.