How UCB modernized its statistical computing environment

Q: What does GxP software validation require when deploying a statistical computing environment on a cloud platform?

Validation requires documented, reproducible evidence that the platform operates exactly as intended across development, acceptance, and production environments. UCB moved away from standard checkbox exercises, focusing instead on validating real-world, end-to-end user scenarios, such as moving raw data, creating analysis datasets, and exporting final outputs. Involving QA stakeholders in early workshops allowed them to understand cloud-hosted mechanics and map out written agreements on review cycles before code hits production.

Q: How do pharma organizations prevent scope drift during a multi-year SCE platform migration?

Scope drift happens both accidentally and intentionally as multiple business units request specialized features over time. UCB countered this by establishing a tight, co-sponsored governance team between IT and the business to ruthlessly filter priorities. The project team focused on establishing a core functional foundation before attempting to roll out flashy automations or advanced features, ensuring they delivered on time rather than getting stuck in endless development cycles.

Q: What training and documentation model is required to support hundreds of programmers moving to a new environment?

Generic onboarding documentation fails because practitioners need to see a day-in-the-life workflow to trust a new environment. UCB structured a persona-based training curriculum that moved from live, instructor-led walkthroughs to self-paced online modules. To ensure continuous support without pulling their core team away from their daily jobs, they published a compliant, searchable online user manual directly to the corporate intranet, giving statistical programmers immediate troubleshooting steps for common script and credential errors.

The hardest parts of a regulated statistical computing environment (SCE) modernization are not technical. UCB's multi-year DELTA program shows how process re-engineering, strict scope discipline, and early QA alignment determine success far more than the technology itself.

Tim Williams

Statistical Solutions Lead

UCB

What you'll take away from this session

Inventory your workflows before touching any system configuration

UCB found that no single person fully understood their end-to-end statistical computing workflow from data ingestion to submission. Completing this mapping discovery first prevents rework during final GxP validation.

Bring QA and regulatory teams in as architects rather than final gatekeepers

Involving compliance stakeholders in your planning workshops from day one gives them technical visibility and transforms them into active, collaborative participants. This removes the bottleneck of late-stage review cycles.

Defend project scope against accidental and intentional drift

Requirements naturally expand through gradual edge-case accumulation or direct stakeholder additions. Teams must formalize a recurring cadence to revisit and actively contain what they deliver.

Pick a low-stakes study for your first live production run

Every new SCE comes with an initial learning curve that causes a temporary efficiency dip. Onboard a standard, non-critical study first to allow users to build confidence.

Role-based training drives adoption at scale

Generic platform training fails to change daily programming habits. Adoption requires day-in-the-life workflows and should start with instructor-led sessions before transitioning to online modules.

Keeping the core execution group small prevents decision paralysis during implementation

A tight core team drives faster alignment on critical platform configurations, while an extended champion network handles the broader work of mentoring individual study teams.

In April 2026, global biopharmaceutical company UCB went live with a fully validated, GxP-compliant statistical computing environment built on Domino under their multi-year DELTA program. Tim Williams, Statistical Solutions Lead, maps out the blueprint of this transformation, arguing that technical implementation is secondary to people and process discipline.

Tim compares upgrading your statistical computing environment to moving to a new house. Leaving reveals what’s been accumulated over decades and the true complexity of your legacy footprint only becomes entirely visible when you begin packing up what needs to move. You can’t install modern technology directly into a decades-old workflow and expect it to run faster. You must rebuild the core business processes around how the new platform functions.

To manage this shift without impacting submission timelines, UCB structured three distinct environments that mirror the lifecycle of a trial: a development workspace for process testing, an acceptance environment that serves as a risk-free training sandbox, and a locked-down production layer for live clinical deliverables. They bypassed the typical checkbox validation exercise, choosing instead to run real-world study scenarios through the platform to surface workflow gaps early. By building a compliant, interactive user manual straight into their intranet and deploying local champions directly onto early study teams, UCB turned what could have been a disruptive migration into a self-sustaining foundation for advanced computing.

FAQ

What does GxP software validation require when deploying a statistical computing environment on a cloud platform?

How do pharma organizations prevent scope drift during a multi-year SCE platform migration?

What training and documentation model is required to support hundreds of programmers moving to a new environment?

Transform the work that matters most

See how Domino helps the world’s most regulated enterprises build, scale, and govern AI-powered applications.

Watch demo Talk to an expert