Sobi: Thoughts on the State of Data in Commercial Biopharma

Dec 19, 2025 | Todd Bellemare

Research surveying 116 biopharma leaders shows 86% of leaders expect measurable AI uplift, yet 89% fail to scale more than half of their initiative.

I recently spoke with Ivan Cadiente, head of field and data operations at Sobi North America, about how foundational data issues slow progress across the industry. He shared recommendations for success based on his experience at a multi-brand organization.

AI needs context-specific training

Companies want to use AI to drive meaningful business value. But leaders should balance ambition with what’s possible given 96% say their data isn’t ready to scale AI.

Cadiente emphasizes that AI needs contextually specific data and a strong semantic layer. That semantic layer should define key entities (HCPs, HCOs, affiliations, etc.) and apply consistent meanings, definitions, and business rules across systems and data sources. Without that foundation, outputs are unreliable.

Sobi focuses on fit-for-purpose use cases where AI can add value immediately, such as pre-call planning and HCP summaries. Cadiente cautions against trying to do too much too quickly. Start small, prove value, and scale deliberately. That’s how leading organizations approach AI today.

Inaccurate affiliations undermine trust

Years of healthcare consolidation and a lack of investment in reference data have eroded the quality of information linking HCPs, HCOs, and broader organizational structures. When affiliations are inaccurate or outdated, the impact is immediate:

• Reps visit HCPs at the wrong locations
• Territories have too few or too many HCPs
• Accounts can’t be prioritized accurately
• AI produces unreliable suggestions

Cadiente says the one improvement that would make the biggest difference is fixing affiliations. Establishing an accurate, cross-brand view of where HCPs practice and how those affiliations roll up to parent organizations strengthens everything from field execution to analytics and AI outputs.

Data inconsistencies erode quality

Data inconsistency is a top challenge for 69% of leaders. Inconsistencies often emerge as companies add new data sources to support product launches and expansion. “Different sources describe the same organization differently,” says Cadiente. “It muddies the waters and slows us down.” Another inconsistency is how specialties are labeled. One dataset might use “Oncology” while another may use “Cancer.”

These inconsistencies lead to conflicting reports, duplicated analytics work, and hesitation in decision-making. For example, field teams question their targets, or analysts repeatedly clean the same issues.

If left unresolved, data inconsistencies erode data quality — and that erosion undermines trust in both the data and the tools built on top of it. “You can’t have data quality if you have fragmentation and inconsistencies,” says Cadiente.

Manual data stitching slows time-to-insight

Companies spend up to 100 days per year matching HCP data across systems, and even more managing third-party agreements (TPAs).

Cadiente says often, in multi-brand environments, teams must stitch together numerous sources to build a complete view. “The manual lift is immense — it delays insights the field needs.”

Field teams miss opportunities, brand teams react slowly, and operational decisions become less timely. Improving speed requires fixing foundational structures first.

Leaders agree: global harmonization unlocks AI’s value

Seventy-six percent of leaders said global harmonization is the top enabler of scalable AI. The right data partner delivers accurate, standardized, connected data that gives organizations a stronger foundation for AI and field execution — freeing internal teams to focus on strategy rather than stitching data together.

Cadiente agrees that companies need data partners who help them move faster and reduce fragmentation — not add to it. “Our partners need a roadmap and vision that keeps up with how the market is evolving,” says Cadiente.

Data challenges are tangible, but they’re solvable. Companies that adopt an enterprise-wide data model and prioritize clean, connected data will be the ones that unlock AI at scale and enable their teams to move with greater speed and confidence.