Addressing the Productivity Challenge in Drug Discovery in Times of Increasingly Complex Modalities

It is well known that the pharmaceutical industry faces a productivity challenge. Despite unprecedented global R&D investment, the number of new medicines reaching patients remains stubbornly flat (Global Trends in R&D 2025, Alexander Schuhmacher et al., “The R&D Productivity Challenge). The productivity crisis has many root causes, not only stemming from the discovery phase, but also toxicity, efficacy in complex biology, trial setup, and increasing regulatory complexity. To solve this productivity challenge, we need to rethink the discovery process, specifically how and when critical risks are identified, and introduce approaches that surface those risks earlier in the discovery and investment timeline. 

For decision-makers, late-stage risk identification does more than delay timelines; it undermines the entire discovery cycle. When scientific uncertainty persists into later phases, organizations make large investments in programs whose underlying biology or chemistry could have been recognized months or years earlier.

Drug discovery is a fragmented process that hasn’t changed in decades, and that begins with target identification, proceeds through lead candidate optimization, moves into preclinical development, and eventually reaches clinical trials. Each stage uses distinct technologies, success criteria, and failure modes. A promising target may fail because binding isn’t specific enough. A potent compound may fail because it aggregates in formulation. A molecule that works in biochemical assays may fail in cells due to poor permeability.

These failures are costly, and the costs grow exponentially the later they occur in the development process. Generating more insightful datasets early in the discovery process and integrating them to identify patterns that better forecast late-stage success or failure is critical to ensuring that investment decisions can be made more efficiently and in a timely manner, before significant resources are committed.

Turning Early Measurements into Predictive Power

The more high-quality data is available to describe biophysical properties and later-stage behavior throughout the development pipeline, the easier it will be to predict liabilities early on. While this type of data is easier to obtain for established modalities, enhanced characterization effort is needed to properly assess the biophysical properties of the increasingly complex candidates and targets in the pipeline today, such as transcription factors, scaffolding proteins, nucleic-acid complexes, lipid nanoparticles, molecular glues, membrane proteins, or intrinsically disordered proteins. Because of this growing complexity, no single method can provide a full picture of how early molecular behavior impacts downstream success. And very likely, we do not even have sufficient methods available yet, for the characterizations we need for these highly promising novel approaches. Impactful correlations come from comprehensive datasets that demonstrate and predict important aspects of molecular behavior. This requires high-quality, integrated, and orthogonal data. Without them, risks are misjudged. Early clarity has the potential to improve attrition rates, technical success, and time-to-value in R&D. However, a potential downside of this early in-depth characterization is that it can lead to analysis paralysis or over-filtering. Emerging AI tools may help navigate this complexity, though validation is ongoing.

Traditional drug discovery has been shaped by biochemical enzymatic assays, which are powerful tools but are structurally misaligned with many modern targets. Orthogonal assays, especially biophysical ones, are essential because they illuminate different facets of molecular behavior not addressable with enzymatic approaches. This is also an advocacy for integration. It would not be helpful to solve the productivity challenge simply by setting up even more complex workflows. Integration converts those facets into a complete picture. Without integration, orthogonal assays are just additional steps. Ideally, integration means multiplex measurements carried out in the same piece of equipment or at least in highly automated systems, thus requiring minimal to no sample preparation.

The Growing Importance of Native-State Characterization for Complex Modalities

Many biophysical methods are surface-based, optimized for specific types of molecules. For simple, well-behaved proteins, these surface-based approaches do work well and have been essential tools for decades. For complex molecular assemblies, however, surface-based approaches can make it very difficult to observe the behaviors researchers need to observe. High-resolution, easy-to-obtain molecular insights into increasingly complex modalities are essential for the development of next-generation therapies. 

This is where in-solution or free-solution biophysical measurements become indispensable. They allow scientists to study molecules in conditions that preserve their native structure and function, generating data that is more relevant, more reliable, and more predictive of real-world behavior, indicative of downstream failure, and aligned with later-stage outcomes. As the field moves toward transcription factors, antibody-drug conjugates (ADCs), molecular glues, proteolysis targeting chimeras (PROTACs), and complex delivery systems, free-solution characterization is becoming central to discovery workflows. In our work with biopharma partners developing novel modalities, we do observe unexpected behaviors emerging in later development stages, including liabilities that weren’t visible with the standard characterization data available at the time of candidate selection.

Building the Foundation for a Smarter Future

Overcoming the productivity challenge in drug discovery is essential to maintaining a sustainable pharmaceutical sector that continues to bring novel, affordable medications to patients in need. While the crisis has many root causes, spanning discovery, clinical development, and regulation, improving how we characterize and select candidates early in the process represents a significant opportunity to work smarter, not just faster.

The drug landscape has expanded well beyond small molecules and simple antibodies to include biologics, gene and cell therapies, and other novel modalities. Each behaves differently, creating opportunities but also demanding new ways to measure and understand molecular behavior. Precision in depth and high-quality biophysical characterization, particularly in free solution, can help researchers characterize biophysical properties more meaningfully and identify high-risk candidates before significant resources are committed.

The key is integration. Collecting more discovery data without making workflows overly complex and connecting early biophysical data with downstream outcomes to build institutional knowledge are essential. This requires investment in comprehensive workflows, data infrastructure, and the discipline to fail fast on unpromising candidates while advancing only the most robust molecules.

The path to a world where every disease is treatable begins not with faster pipelines but with smarter ones, built on integrated insight, rigorous early characterization, and the courage to make difficult decisions earlier in the discovery process.