Recently, the J+D Forecasting team welcomed a diverse group of professionals to Boston for our two-day Fundamentals of Forecasting multi-client workshop, run in collaboration with Evaluate. From emerging clinical-stage biotechs to global Top 10 pharma companies, delegates gathered to deepen their knowledge, share experiences, and engage in peer-to-peer learning in a relaxed, collaborative environment.
The atmosphere was dynamic. With US and Global remit roles spanning from New Product Planning Directors and Commercial Operations Senior Directors to leaders of Forecasting Centres of Excellence, the collective insight in the room was palpable. Everyone brought their own challenges, questions, and perspectives—making for rich discussion, and practical, grounded takeaways.
Breaking Down Forecasting Fundamentals – with CATS
The training was led by Andrew Ward, Head of Implementation, and Kris Barker, Senior Implementation Director—two of J+D’s most seasoned experts. They introduced delegates to the essential elements of building a high-quality forecast model using the memorable and light-hearted acronym CATS:
- Consistency
- Applicability
- Transparency
- Simplicity
While the content was thorough, the delivery was approachable and engaging. Delegates were fully immersed in each session, and the feedback was overwhelmingly positive. One attendee remarked, “Very well done—I wish I had taken this training years ago!” Another shared, “Facilitators were excellent, very knowledgeable.”
Day 1 Recap: Forecasting Foundations and Practical Methodologies
Day 1 focused on building a strong foundation in forecasting best practices. The morning began with sessions on defining and trending the market, followed by a discussion on what makes a “good” forecast. We explored the key stages of forecasting, including understanding the forecasting cycles, navigating internal workflows, and balancing stakeholder needs—a critical skill in ensuring the forecast is both trusted and actionable.
Delegates were introduced to the core forecasting methodologies—sales-based, epidemiology, opportunity-based, and patient flow—along with guidance on when to apply each one. These frameworks provided the basis for deeper discussions and hands-on modeling work that continued into Day 2.
Forecast Horizons: 3 Years? 10 Years? 20?
A hot topic early on was forecast duration. How far out into the future should models look? Unsurprisingly, the answer varied based on model purpose and forecast cycles—whether LRP, LBE, or AOP. Some Top 10 pharma companies revealed they forecast up to 10 years, sometimes even 20. Others were more conservative, especially for established brands where the outlook beyond 2 years becomes speculative.
Building one model to cover a long time horizon? That came with its own challenges. While a 3–5 year forecast can be managed in a single file, going further often leads to performance issues and complexity that undermines the model’s usefulness.
AI: A Support Tool, Not a Silver Bullet
Artificial intelligence inevitably came up, particularly regarding its role in forecasting. A common theme across big and small pharma: AI isn’t doing the forecasting—but it is helping streamline the process. From running multiple scenarios to rapidly contextualizing large datasets, AI tools are speeding up epidemiology research and even helping structure competitor analyses, such as SWOT frameworks.
However, adoption is still in its infancy. No companies reported full AI integration into their forecasting processes. As one big pharma delegate admitted, AI is promising, but trusted data and internal capabilities remain key limiting factors. There’s interest—but also caution.
Opportunity-Based Forecasting and Market Dynamics
One of the most insightful discussions came around opportunity-based forecasting, particularly modeling naïve vs. switch patients. A commercial biotech emphasized the need for consistent definitions, especially as scrutiny increases closer to product launch. A pre-commercial delegate introduced the use of persistency curves to inform when and how to model these dynamics.
The discussion deepened with mention of ‘warehousing’—a phenomenon where prescribers delay treatment switches in anticipation of a better therapy. This kind of nuance, particularly common in rare and orphan diseases, illustrates just how vital it is to have a robust patient flow model. (We explore this topic further in this article on rare/orphan diseases.
Day 2: Eventing, Real-World Data, and Analogs
Day 2 opened with the topic of eventing—adding future events to forecasts. While some already included competitor launches, others admitted this was new territory. The training emphasized how to integrate these elements while maintaining two of the core CATS principles: simplicity and transparency. Simplicity, we emphasized, does not mean sacrificing robustness—it means making models fit for purpose.
Following the workshop, several planned to incorporate market and product events, such as pricing shifts or guideline updates, recognizing their impact on therapy uptake.
Real-world data (RWD) was introduced into the discussion. A major pharma mentioned the challenge of sourcing international data, while a biotech expressed interest in better ways to include RWD in new product planning. There was a shared appetite for analog libraries and scoring models, both covered in detail on day two.
We also covered how to validate epidemiology-based models using sales data. By comparing treated patient estimates with actual sales—adjusted for pricing and dosing—teams can strengthen confidence in their forecasts and ensure greater alignment between data sources. This really resonated with those who were actively putting it into practice, and they openly shared the challenges they were facing with the group.
Delegates appreciated the hands-on approach: a full case study walked them step-by-step through market definition, event integration, trend modeling, and conversion. We also delved into statistical methodologies—how to choose the best-fit time series approach and when to override with manual trend selections. (For those keen to explore this further, check out our new TSM online training course.)
Communicating the Forecast: Story First, Numbers Second
One of the final, and most impactful, takeaways was around how to communicate forecast results. Instead of leading with the final number—which can spark premature emotive reactions—our facilitators recommended leading with the story. Walking stakeholders through the key assumptions first builds understanding and buy-in, making it easier to have constructive conversations about the output.
Future Events
As the workshop wrapped up, delegates left not only with exclusive training resources and access to J+D’s forecasting software, but also a new network of peers to support their forecasting journeys.
Thank you to all who joined us in Boston—we loved hosting you! If you missed this session, don’t worry. Our calendar of upcoming workshops across the US and Europe will be announced soon.
Stay tuned—and we look forward to seeing you at a future event.
