With over 1,000 branches, Fifth Third Bank is one of the largest regional retail banks in the U.S. Known for its strong community presence across the Midwest and Southeast, the institution is equally committed to modernizing its data and analytics capabilities to deliver trusted, personalized financial services.

In this first part of a three-part series, Shanna Anderson, Sr. Director, Customer Data and Insights, Fifth Third Bank, sits with Or Zabludowski, CEO of Flexor, as she reflects on her priorities, the bank’s evolving ModelOps journey, and how it is laying a foundation for trusted, explainable AI.

Anderson was recently recognized on CDO Magazine’s 2025 Global Data Power Women list. She leads customer data mastery, ModelOps, and insights delivery at Fifth Third, a portfolio that touches revenue, profitability, and customer experience.

From business to data: A perspective shift

Anderson did not grow up in technology — her roots are in the bank’s consumer business. This background, she says, gives her a unique vantage point. “I spent about 20 years in the consumer line of business working on the front lines with our branches, as well as on our consumer strategy side,” she explains. That experience allows her to partner with executives on outcomes, translating data capabilities into strategic business value.

For 2025, her team is focused on three priorities:

• Modern data strategy, giving business users transparency into data sources and pipelines.
• ModelOps transformation to streamline the journey from model development to deployment.
• Household growth, using insights to expand market share and improve customer experience.

Together, these priorities illustrate the pendulum swing: moving toward agility and innovation while holding firmly to control and trust.

“With great power comes great responsibility”

Anderson emphasizes that handling customer data in banking comes with extraordinary responsibility. “There’s a lot of trust that comes with that. I tell my team all the time, great power comes with great responsibility,” she explains.

For Anderson, this principle underscores the importance of quality, transparency, and keeping the customer’s voice central to every decision.

This philosophy guided Fifth Third’s transformation of its ModelOps supply chain. The bank treated its internal data scientists as “customers,” mapping out every step of the model-building process to identify duplication, misplaced steps, and inefficiencies.

“We spent a lot of time working with our data science teams as well as our governance teams, understanding each step in that full model process,” Anderson recalls. Using Lean Six Sigma tools, her team pinpointed gaps and opportunities to shorten timeframes. “If we can deliver models faster, then we also get to business value faster,” she adds. 

Reuse over rebuild: The power of feature stores

One breakthrough came with the creation of a “feature store,” a central repository of model features.

“In the previous state, if our data science teams built a feature, they would build it anew for every single model each time,” Anderson recalls. Governance teams often repeated the process, wasting both time and effort.

The feature store ensures reuse, embeds statistical calculations, and provides lineage and ownership clarity. It eliminates duplication while strengthening governance.

Cloud as a catalyst

Alongside the feature store, the bank has advanced its platform itself, shifting from on-premise to a cloud-based model development. The cloud environment expands access to big data sources, enables real-time decisioning, and opens the door to richer customer recommendations.

Together, these changes culminate in a leaner, faster, and more transparent ModelOps process — one that makes it easier for teams to innovate while upholding the trust customers place in the bank.

Governance in a black-box world

Looking ahead, Anderson acknowledges the opportunities and risks of generative AI and large language models. For a highly regulated institution like Fifth Third, governance and explainability remain non-negotiable.

“We have to be able to explain any type of model decisions, and biases and things like that are something that have to be monitored,” she emphasizes.

To that end, Anderson mentions that the bank has established governance frameworks, steering committees, and compliance checkpoints to ensure any use of AI or LLMs meets regulatory standards while protecting customers from unintended bias.

CDO Magazine appreciates Shanna Anderson for sharing her insights with our global community.