How AI Is Future-Proofing Pharma Demand Forecasting

Authored by: Rachel Howard, Roy Rogers and Bilal Babar

Uncertainty is the one constant in today’s pharmaceutical landscape. Clinical trials may miss endpoints, timelines may shift, and competitor activity can evolve unexpectedly. Add to that the broader pressures of political instability and economic volatility, and the path to market becomes even more complex. At the same time, patient journeys are changing, especially as large language models (LLMs) are empowering people with more information about their health than ever before. All these variables mean that traditional demand assessments – often built on static assumptions – are increasingly insufficient.

At Research Partnership, we’ve evolved our approach to demand assessment to better account for this shifting reality. By combining advanced Monte Carlo simulations with using LLMs to support deep secondary research, engaging avatars to present stimulus, and conversational probing to get deeper into what is driving stated demand, we help our pharmaceutical clients’ commercial teams make confident decisions in a world that’s anything but fixed.

From Static Forecasts to Simulated Futures 

At the heart of our approach is a simple but critical shift in how we deal with uncertainty: moving from manually-defined future scenarios to a probabilistically driven spectrum of possible outcomes.

Where traditional demand studies often involve pre-set scenarios, Monte Carlo simulations go further. They model uncertainty directly by running thousands of “what if” scenarios, considering unknowns around clinical trial outcomes, launch timelines, patient uptake, and more, and allowing us to adjust the inputs within a pre-defined range of possibilities for each parameter.

This method isn’t new to pharma: Monte Carlo simulations have long been a gold standard in Health Economics and Outcomes Research for developing Cost Effectiveness Models in the face of all this uncertainty. But until now its application in commercial demand assessment has been limited due to it being time consuming and often cost-prohibitive to run. AI computing power changes that, making it practical to scale simulations and allowing us to bring that same level of analytical robustness into demand forecasting, learning from the best when it comes to handling variability and risk.

For instance, instead of saying “Product X will capture 5% market share”, we can run simulations that quantify the likelihood of hitting exactly 5%, as well as showing that there’s a 90% probability your share will land between 4% and 6% – and a 0% chance of falling outside 3% to 7%.This probability distribution gives clients a much more realistic and resilient foundation for strategic decisions.

Importantly, we believe Monte Carlo enables us to take a more forward-looking and decision-oriented approach than simply validating a demand model by tracing alignment between real world outcomes and historical predictions. In our view, that approach is limited in assuming static conditions and a level of market stability that does not reflect reality. Monte Carlo, on the other hand, lets a 2030 launch forecast evolve, ensuring a demand assessment run today is not frozen in a 2025 view. As new data arrives, we can update inputs and rerun simulations, so the distribution reflects a 2028 lens, then 2029, narrowing uncertainty as launch nears.

Leveraging AI to Enrich Forecast Inputs

What makes these simulations even more powerful is our ability to use AI to enhance the complexity and sophistication of the scenario inputs, making simulations more dynamic and grounded in evidence. We can:

• Scan real-time clinical pipelines
• Map likely competitor launches
• Use the latest secondary data to evaluate trial success probabilities
• Assess broader market dynamics

AI also enhances patient sizing, aggregating incidence and prevalence data while factoring in demographic complexity – all without the model becoming unmanageable.

In a recent study, our AI-enabled deep-dive into secondary data allowed us to anticipate shifts in the competitive landscape and refine the clinical assumptions feeding into the design of the conjoint attributes and levels we subsequently tested with HCPs in an online survey. Rather than relying on generic comparator profiles, we were able to build scenarios that reflected likely future realities.

Engaging Stakeholders with Immersive Avatars

Each demand assessment is only as good as the quality of the primary market research data that feeds it. To generate more authentic responses from healthcare professionals when surveying them about anticipated future uptake of novel products, we use avatars to deliver consistent, engaging scenarios that help them visualize how evolving treatment pathways may shape their prescribing behavior.

With multiple academic studies supporting the improved quality of recall from combined auditory and visual information vs. either on its own, the use of multimodal avatars enhances survey respondents’ evaluation of new products, leading to richer, more reliable data.

Conversational Probing for Deeper Demand Insight

Traditional surveys often stop at the “what”. AI-powered conversational probing allows us to get deeper into the “why”. Trained on relevant therapeutic areas and embedded directly into quant surveys to engage respondents in a conversational dialogue, these probing questions dynamically prompt respondents to elaborate on the trade-offs they make.

This adaptive method yields:

• Nearly 3x more qualitative depth in open-ended responses
• Greater nuance in understanding trade-offs
• Improved interpretation of demand forecasts

Smarter Decisions Through AI-Driven Forecasting

In times where the only certainty is change, future-proofing demand assessment is no longer a “nice to have”. By simulating futures, grounding estimates in probability, and understanding stakeholders’ rationale behind those estimates more deeply, our AI-infused approach enables pharma teams to gain a clearer view of the road ahead. Critically, this enables them to make data-driven decisions on everything from manufacturing capacity to portfolio prioritization, as well as informing launch planning across marketing spend and sales force allocation. And as the landscape continues to evolve – whether from changes in the target product profile, competitor activity, or broader market dynamics, AI can help us continuously refine the scenarios being modelled, ensuring forecasts stay relevant and actionable.

Ready to Future-Proof Your Forecasts?

Talk to our experts to see how AI-powered forecasting can add certainty to our launch strategy and market planning. Get in touch.