CROs In Early Drug Discovery, Part 1: Choosing The Right Partner

By Simon Cocklin, Ph.D., director of therapeutic discovery, Chan Soon-Shiong Institute for Molecular Medicine at Windber

This article begins a three-part series on working effectively with CROs in early drug discovery. Part 1 emphasizes the essentials — selecting the right CRO and establishing clear scientific expectations along with a well-defined statement of work (SOW). Part 2 discusses what occurs once experiments start and operational pressures come into play. Part 3 examines the practical aspects that influence whether outsourced work remains on schedule as programs expand or timelines become tighter.

Outsourcing has become a key part of early drug discovery, but most discussions about CRO involvement treat it as simple. In reality, the quality of outsourced work varies greatly. Over the years, I have seen both extremes: CROs whose discipline, communication, and scientific rigor accelerated my work and others whose lack of replicates, resistance to feedback, or poor assay execution produced data that could not be used. The difference between these outcomes rarely depends on the target biology. It is more about alignment, clarity, and the structure of the collaboration.

My perspective has been shaped by conducting discovery work in academia, working within a CRO, and now developing a therapeutic discovery effort at a private research institute. Even as an academic, I relied on CROs to keep projects moving and to prevent wasting months of student time on work that could be done more quickly and reliably externally, such as reagent generation. Doing this kept projects progressing meaningfully, and it allowed students to avoid spending months on necessary but routine tasks in favor of experiments that truly develop scientific reasoning.

Across all these settings, one pattern is clear: CROs excel as technical operators, but they are not responsible for your biological reasoning or strategic decision-making. They are not deeply involved in your hypotheses, the details of your system, or the effects that truly matter for your program. If you do not specify the scientific context, the necessary controls, and the decision thresholds, a CRO will make reasonable assumptions — but often the wrong ones. That responsibility lies entirely with the sponsor. When expectations, boundaries, and success criteria are clear, CROs provide high-quality, reproducible data at a pace that internal teams rarely match. When they are left to “fill in the gaps,” they may deliver exactly what was asked for, but not what was actually needed. Clear scientific guidance is what turns a CRO from a vendor into a dependable extension of your discovery efforts.

This article explains the framework I use to choose CROs, write effective SOWs, keep momentum, and ensure the data produced is reproducible and truly supports decision-making. The goal is simple: not to outsource less but to outsource in a way that reliably delivers the quality and clarity early discovery needs.

Choosing The Right CRO For The Job

A successful CRO partnership begins with selecting a team whose skills genuinely match the workload. Despite their marketing claims, CROs are not interchangeable. Some provide high throughput but lack scientific depth; others are excellent at solving complex, high-context problems but cannot handle thousands of samples each week. The key is aligning the CRO with the specific task, not the other way around.

Technical Fit

Most mismatches occur at the technical skill level. High-quality biophysical data — especially surface plasmon resonance (SPR) and biolayer interferometry (BLI) — can be surprisingly hard to gather consistently. Many CROs advertise these platforms, but only some provide clean baselines, reliable referencing, and properly controlled binding data. In contrast, routine construct creation (gene synthesis, codon optimization, subcloning) is usually performed much more efficiently and far faster externally than internally. A CRO might excel in one area but struggle in another: a high-volume screening lab might have all the instruments but lack the expertise to troubleshoot a temperamental protein or low-signal assay. Smaller, science-focused partners often perform better when a project needs careful assay development, mechanistic insights, or iterative scientific interpretation. No level of automation can replace a team that cannot engage with the biology. Ensuring technical fit up front directly impacts whether the data will be robust enough to support real decision-making.

Scientific Depth Vs. Throughput

A key early decision is whether the project requires a thinking CRO or a throughput CRO — in other words, whether scientific problem-solving or efficient execution is more important. For new targets, unstable proteins, or assays that need true scientific judgment, you need a team with Ph.D.-level scientists who can analyze unexpected results, suggest alternatives, and make adjustments in real time. When a CRO has strong scientific capabilities, both data quality and the speed of resolving issues improve significantly.

Throughput-focused CROs are ideal when the assay is already validated and you only need high volume. They operate efficiently, reliably, and at a large scale. However, giving an unoptimized or exploratory assay to a throughput-oriented organization almost guarantees repeated failures because these groups are not designed to develop assays — they are built to run them.

Practical Considerations

Logistics and communication can determine the success or failure of a collaboration. Not all CROs have dedicated project managers (PMs), and the difference is immediately apparent. A PM who is also a scientist — someone who genuinely understands the data and biological context — helps prevent miscommunication and identifies misunderstandings early. A practical sign is how confidently a PM can interpret basic preliminary data without needing to "check with the team" for fundamental points; hesitation in this area usually indicates shallow scientific knowledge. Without a strong PM layer, the sponsor must take complete responsibility for overseeing every operational detail.

Cultural alignment is also important. A major red flag is a CRO that does not listen. If early conversations seem dismissive or transactional, that pattern rarely changes. Conversely, a green flag is a CRO that works collaboratively: open to technical discussions, comfortable asking questions, willing to support grant work or publications, and interested in building a long-term scientific relationship instead of a one-off project.

Writing an Effective Statement of Work

Most CRO problems come from a poorly written SOWs. A good SOW is not just administrative paperwork — it is the operational backbone of the project. When the scope and expectations are clear, both sides stay aligned. When they are vague, you get scope creep, unusable data, and long, costly arguments about what "should have" been done.

Define Exact Experimental Objectives

An effective SOW clearly defines the purpose of each experiment. A vague description like "test inhibition" is useless. A well-crafted SOW details the biological context, readout, and conditions, such as, for example, “determine whether compound X inhibits enzyme Y at physiological pH and 37°C using a fluorometric assay, with appropriate positive and negative controls and triplicate measurements.”

I always define required controls and minimum replicates. Reproducibility remains one of the weakest points in outsourced work, so if an experiment is important enough to run, it is important enough to run at least in duplicate — ideally in triplicate.

When I am familiar with a method, I specify controls and readouts directly. When I am not, I outline what I believe is necessary and ask the CRO to suggest improvements. For example: “Here are the controls and quality control (QC) steps I believe are necessary — if there is a better or more reliable way to achieve the same goal, propose it.”

This uses the CRO's technical skills without sacrificing scientific purpose.

For expression constructs, however, I am never flexible: construct design is always sponsor defined. CROs cannot infer how a protein is processed or which domains matter.

Scope Experiments Into Milestones

My early-discovery SOWs follow a consistent staged structure:

• Protein expression → purification → QC
• Activity or binding confirmation (absolutely essential; a protein that does not bind its known partner is not correctly folded and should always be confirmed before any assay qualification.)
• Assay qualification (signal quality, reproducibility, Z′, control performance)
• Pilot testing
• Full compound execution

Each stage concludes with a stop/go review. I do not proceed until I review the raw data and confirm that the QC metrics are acceptable. This avoids significant rework later.

The most common failures arise from insufficient early QC, which only becomes evident weeks later when a protein is inactive or an assay fails to stabilize.

When appropriate, I also ask the CRO to suggest their own milestone structure and acceptance criteria. Their internal workflow often uncovers hidden constraints or bottlenecks early.

Specify Deliverables Clearly

Every SOW must clearly list deliverables. "Final report" is meaningless unless broken down into parts. I always ask for:

• all raw data (plate reader files, chromatograms, spectra, SPR/BLI sensorgrams)
• QC data for proteins and key reagents (purity, mass, concentration verification, and activity/binding confirmation)
• detailed methods (buffers, pH, temperatures, incubation times, plate maps)
• batch information and chain of custody
• a written summary with interpretation.

If I have the software to analyze proprietary formats, like SPR instrument files, I request those. If not, I still want exported traces — not screenshots.

My rule is straightforward: if the experiment is important, then complete data and full methodology must be provided.

Timelines And Contingencies

Clear timelines help keep expectations aligned. I prefer a straightforward Gantt chart showing expected completion dates for each milestone, with space for repeats or improvements.

I also set rerun triggers using the same criteria I use internally:

• Control failure
• QC metrics below threshold
• Unreproducible results
• Unexplained variability
• The protein loses activity
• Assay Z′ or coefficient of variation (CV) outside agreed limits

This clarifies whether a data set "counts."

I am also open to "limited try" clauses — for example, two expression attempts before escalation — which encourage early discussions rather than a month of quiet repeated failures.

Finally, SOWs should clearly specify fallback options:

• If assay A cannot be validated, attempt assay B.
• If protein activity is low, pause and troubleshoot before compound testing.

These contingencies stop CROs from improvising work that does not support the scientific objectives.

Conclusion

The challenges of working with CROs rarely stem from the target biology — they come from unclear expectations, mismatched capabilities, and assumptions no one realized were being made. Part 1 outlined the key foundations that prevent these issues: selecting a CRO whose strengths truly align with the scientific task, establishing expectations early, and writing an SOW that clearly communicates the experimental intent.

These steps do not solve everything, but they shape the overall direction. Projects that start with clear structure and shared understanding remain stable; those that begin with ambiguity tend to drift, which only becomes apparent after time, money, and momentum are already lost.

Part 2 builds directly on this foundation by focusing on what happens after the work starts — how to sustain momentum, avoid miscommunication, and ensure data quality during daily execution.

Part 3 then completes the framework by addressing the operational realities that quietly determine the success or failure of outsourced discovery: budgets, timelines, and the common pitfalls that can undermine even technically sound programs if not managed carefully.

Taken together, the three parts offer a practical end-to-end guide: from choosing the right CRO to maintaining aligned collaboration and ensuring the data you get is reproducible, decision-grade, and capable of advancing the science.

Read part 2 of this series.

About The Author

Simon Cocklin, Ph.D., is the founding director of therapeutic discovery at the Chan Soon-Shiong Institute of Molecular Medicine at Windber (CSSIMMW), where he leads translational drug discovery projects in immuno-oncology and fibrosis. At CSSIMMW, he is establishing a new Therapeutics Discovery Department and building scientific and infrastructural capabilities to support early-stage drug development aligned with the institute's goals.

Cocklin is the chief scientific advisor for Regenova Pharmaceuticals, an early-stage biotech that utilizes multi-omics and AI to develop antibodies against infectious diseases and cancer targets. He is also the cofounder and co-CEO of Bespoke Patient Solutions, LLC.

He previously held faculty and leadership roles at Drexel University College of Medicine, where he led NIH-funded research programs focused on drug discovery for HIV-1, infectious diseases, and oncology.