Building an Integrated Cloud Forecasting Platform for a Scaling ADC Portfolio

A global pharmaceutical company with a leading oncology franchise was scaling its antibody-drug conjugate (ADC) portfolio faster than its forecasting infrastructure could support.

In roughly two years, the portfolio grew from a single on-market asset with about five indications to multiple launch-ready assets spanning fifteen-plus indications, with overlapping patient populations across breast, lung, and other tumor types.

The forecasting process spanned 17 disconnected Excel models, each owned by a different forecaster. Assumption management, version control, scenario comparison, and portfolio roll-ups were all manual, slowing forecasting cycles and limiting the team’s ability to answer leadership questions in real time.

Challenge

Rapid Portfolio Expansion

• The portfolio moved from one on-market product across five indications to multiple launch-ready assets across fifteen-plus indications, with overlapping tumor populations across breast, lung, and other types
• Forecasting complexity, stakeholder count, scenario management, and alliance coordination all grew in parallel

Fragmented Forecast Infrastructure

• Seventeen standalone Excel models across three ADCs with no shared data layer, no version management, and no single source of truth for locked forecast submissions
• Assumptions were siloed in individual models, with no centralized catalog to visualize inputs across assets and/or indications

Growing Analytical Demand

• Leadership’s growing needs on class-level dynamics, tumor overlap, what-if simulations, portfolio driver analysis, and scenario comparisons that the existing setup could not support at speed
• Increasing cross-functional stakeholder complexities, with leadership mandating harmonized, auditable outputs across the portfolio

Slow, Manual Reporting and QC

• Quality control was time-consuming and carried elevated risk of manual error during peak cycle windows
• Discussion material preparation for brand discussion, business unit meetings, and alliance reviews was repetitive and consumed considerable day-to-day bandwidth for forecasters

Approach

Excel-Native, Cloud-Powered Architecture

• Forecasters continued to work in their familiar Excel models through a lightweight Office 365 add-in, so day-to-day workflows required no retraining
• Every lock, save, and load now reads from / writes to a central AWS data lake that sits between the models and executive stakeholders

Pilot-Then-Scale Delivery

• A single asset served as a pilot to validate the end-to-end architecture before scaling the full portfolio
• The remaining models across the portfolio were then integrated, with scalable dashboards and outputs

Live Dashboards and Automated Reporting

• Power BI portfolio, indication, and asset storyboards gave stakeholders dynamic access to trends and business drivers, with what-if scenario controls
• One-click generation replaced manual deck assembly for recurring brand, business unit, and alliance reviews

Governance and Cross-Asset Learning

• A web-based Forecasting Portal managed and governed forecast processes – planning cycles, submissions tracking, through an access-based user control
• The Assumptions Catalog and Forecast Library enabled reuse of historical inputs, uptake curves, and testing rates across new pipeline builds

Self-Serve Scale Utilities

• A Model Builder, based on the client’s own validated model blueprints, let the team standardize new pipeline and go-to-market models without a custom build
• A Knowledge Hub with an audit trail for key assumptions, data sources, and scenario notes for all forecasts

Outcomes

End Deliverables

• Integrated cloud data connecting all seventeen Excel forecast models and aggregators across three ADCs 
• Office 365 Excel add-in gave forecasters access to local scenarios from the cloud, or save and lock directly to the cloud 
• Live Power BI – portfolio, indication, and asset dashboards with what-if scenario controls 
• Forecast Library and Assumptions Catalog for cross-asset use of historical inputs, with an audit trail that builds the institutional knowledge base 
• A web-based Forecasting Portal for cycle and access governance 
• Automated generation of standardized PowerPoint reports for leadership presentations 
• Self-serve Model Builder, for single-click generation of forecast models for new indications and assets 

Key Highlights

• A single source of truth lets leadership, forecasters, and alliance teams reference the same locked, versioned scenarios 
• What-if scenarios run in real time during executive meetings rather than over multi-day analysis exercises 
• Recurring brand, business unit, and alliance review decks generate in one click, reducing slide preparation from days to minutes 
• Standardized Model Builder templates accelerate new pipeline and go-to-market builds, reducing effort spent per new model development

Impact

• Forecasters saved time spent on consolidation, slide prep, and version control, allowing them to spend more time on analysis, insights, and business partnering 
• Auditable, reproducible outputs raised the credibility of forecasts in executive settings 
• Cross-asset learning from the Forecast Library strengthened assumptions quality for new pipeline builds 
• With the Model Builder, PowerPoint automation, and Forecasting Portal, the team can scale to new indications and cycles with minimal external support 
• Scalable architecture built to extend beyond the U.S. ADC portfolio to non-ADC assets and additional geographies