Enterprise data storage grows daily, accumulating over the years. After decades of buildup, even with a curated data asset, unlocking its potential is a formidable task.

At Roche, a global pharmaceutical giant with a history of nearly 130 years, messy and incompatible systems hindered efforts to mine deeper insights from customer data. Modernizing an ecosystem spanning over 100 countries also helped pave the way for AI adoption.

"Transactional systems vary widely," Joao Antunes, chief engineer for platforms and data strategy at Roche, told CIO Dive. "Name any technology, and there's always some country using it."

Four years ago, when Antunes' team launched a large-scale cloud modernization effort, the company was feeding customer data into multiple CRM systems scattered across a global network.

The goal was to build a system that could provide sales teams with a unified view of each healthcare professional. By consolidating platforms, Roche paved the way for a machine learning-based assistant, which Antunes likened to the recommendation engine driving Netflix. These efforts also led to the deployment of a generative AI summarization tool with chatbot capabilities.

"In a heterogeneous environment, scaling any analytics solution is extremely difficult," Antunes said. "It's nearly impossible to get a holistic view of the customer."

When executives look at vast enterprise data assets, they always see the potential for unrealized value. Modernizing legacy data systems, breaking down data silos, and making IT assets pay off for investments in analytics and AI capabilities require time, resources, and strategic foresight.

According to Gartner, this is a process that enterprises across industries are tackling. In a February report, the firm said nearly two-thirds of organizations lack the data management capabilities needed to implement AI, or are unsure whether their processes measure up. Among chief data and analytics officers surveyed by Gartner last year, more than a third put data architecture reform on their agendas.

For Roche, the value of data modernization has become clear in both daily operations and the broader picture of system simplification.

"As of now, we have just one CRM system that provides a unified customer view—whether in Switzerland, France, Brazil, or Canada," Antunes said. "You see the customer profile, and you also know this customer is the one who writes scientific publications or is a trendsetter on Twitter. We didn't have that information before."

Streamlining systems to upgrade analytics has also brought broader returns.

"For a company of Roche's scale, adopting the same platform everywhere is both difficult and not necessarily beneficial," Antunes said. "But if you're ingesting the same customer data in multiple places through multiple vendors, you can save significant costs just by moving to a common platform."

Enterprise-wide buy-in

Major wins stem from small wins, especially in an organization as vast as Roche. Modernizing customer analytics is an incremental process that requires executive support and bottom-up buy-in.

Looking back from the perspective of 2025, things are looking quite good.

The company hired Chief Digital Technology Officer Wafaa Mamilli in January to lead global IT strategy. Mamilli, who held technology executive roles at Zoetis and Eli Lilly, succeeds Alan Hippe, who served as both CIO and CFO, in overseeing the company's informatics functions.

In the appointment announcement, Mamilli prioritized using data and AI to "prevent, stop, and treat disease."

"This appointment shows that we are treating data as one of the key pillars of our future strategy," Antunes said.

Roche's customer analytics improvements have been incremental, with Antunes' team targeting practical applications to build enterprise-wide momentum for multi-year projects.

"We didn't try to do everything at once—we learned as we scaled," Antunes said.

As the data team grew from five people to over 100 engineers, the team focused on rolling out a content recommendation system, which was piloted first in Brazil and Canada before broader deployment.

"We migrated their local solutions to our platform, decommissioned parts that were no longer needed, added valuable features, and then started gaining traction," Antunes said.

The first tool validated a larger concept: the value of a leaner, more agile data architecture.

"It was never just about the content recommendation system," Antunes said. "For us, it was more about laying the foundation for many other analytics products."

There was resistance along the way, as processes needed to align with the data architecture. The central platform, leveraging AWS Redshift cloud data warehouse and Salesforce CRM, had to meet the needs of stakeholders with varying regions, technical proficiency, and workflows.

"Implementation goes far beyond the technical side," Antunes said. "It's more about people and processes, gaining alignment and buy-in, and pushing forward decisively when necessary."

Narrowing the scope of data modernization was also key to the plan. Roche has decades of records. The migration was limited to data supporting use cases with business value.

"You need to be strategic about resources," Antunes said, pointing to the time required for data movement and processing.

"If you have 20 years of data, people will tell you they need all of it," he said. "What can you achieve with 20 years of data that you can't with four or five years? You have to have that conversation."

Modernization efforts can also get bogged down in an endless pursuit of perfect technology. While some tools are better than others, focusing on the specific needs of the project has its own value.

"We stay on top of the latest technology and try to get the best solutions," Antunes said. "At the same time, we're not bound by 'choosing the best tool for every component.' Instead, we prioritize building seamless integration between systems, which is often overlooked."