After more than a decade of talk about potential, big data is beginning to have a real impact in the biopharmaceutical industry. The companies that have realized the most potential—and that will continue to do so in the future—focus on much more than choosing the right software and hardware. Our experience shows that they also think carefully about where to deploy big data, how to develop the necessary capabilities, and how to establish the right operating model to drive appropriate usage.

Big Data Is Made for Biopharma

Big data can be used wherever a data set can help inform a decision. Given many degrees of freedom in decisions along the value chain from research to the real world, and as one of the world’s most information-intensive industries, biopharma has much to gain from data analytics. (See Exhibit 1.) The industry has become even more data driven in recent years as new technologies have become more affordable and widely available. The cost of genome sequencing, for example, has dropped from more than $10 million per genome 15 years ago to less than $1,000 today. As a result, the amount of genomic data available for biopharma R&D has increased exponentially. At the same time, the volume of patient, disease, and treatment-related information has expanded dramatically. For example, the National Health Service (NHS) in the UK adds 19 million inpatient records, 90 million outpatient records, and 18 million accident and emergency records to its databases every year.

To their credit, many biopharma companies have invested millions of dollars in technologies, capabilities, and processes to exploit big-data possibilities. Until recently, however, much of this investment was still a bet on the future. Now we are seeing an increasing number of actual applications that are having significant impacts. The following examples in three different areas—clinical trials, sales and marketing, and treatment outcomes—point the way for biopharma companies to harness the power of big data in the real world today.

Delivering More Effective Clinical Trials

Clinical trials pose the industry’s biggest financial risk and consume enormous resources and time. Trials typically cost more than $2 billion for every approved drug (including the cost of failures), and the average new drug has approximately a 10% chance of making it through trials successfully. Many new drugs fail on their own merits because of lack of efficacy or safety, but many also fail because of inefficiencies inherent in the way clinical trials are conducted. Big data is now beginning to address this issue.

Historically, patients have been selected for clinical trials on the basis of fundamental phenotypic features, such as age, gender, and disease presentation. These factors, however, account for only a small fraction of the variations among patients, and they may miss important biological features that affect the likelihood of drug response. In some instances, such as for certain cancers, the underlying biological differences that determine differential response are well known. But in most situations, the underlying drivers of heterogeneity remain a mystery. To help add precision to the trial process, biopharma companies are beginning to use big-data analytics to assess a much broader variety of features relevant to patient stratification, including genetic data, serological and other biomarkers, clinical features in electronic medical records, and even patient-reported outcomes.

We recently worked with one biopharma company that used big data to design clinical trials for a novel cardiovascular drug. The initial population of patients was selected using traditional phenotypic features, and patients were grouped by age, sex, body mass index, and so forth. The company then accessed electronic health records from a provider partner and combined them with de novo information on genetic, metabolic, and lifestyle parameters using a combination of diagnostic tests, monitors, and patient-reported data to generate a rich data set across layers of information types. When this information was analyzed and evaluated against published scientific and clinical data, the results revealed certain patterns—a healthy microbiome was correlated with a high likelihood of responding to cardiovascular treatment, for example, while low levels of physical activity reduced the probability of response. Using these approaches, the company was able to further segment the patient population and to identify the individuals who were most likely to respond to the new drug, thereby reducing the size and cost of the trial and increasing its chance of success. (See Exhibit 2.)

Optimizing Sales and Marketing with Geoanalytics

Biopharma companies typically spend 20% to 30% of their revenues on selling, general, and administrative expenses, and companies are constantly looking for ways to maximize the impact from their commercial spending. The application of big data to sales and marketing decisions provides one path forward. An example is using detailed geographic information on disease prevalence to optimize sales force design. (See Exhibit 3.)

BCG has helped several biopharma companies develop strategies based on data analysis that model the geographic market size for disease types using epidemiological and census data, among other sources. For each geographic unit (a state or group of states in the US, for example; a group of states in Germany; or a set of NHS trusts in the UK), we can estimate the workload of a sales force by calculating the coverage index and thereby determine the potential sales lift from rebalancing sales teams among over- and underindexed areas. Typically, we see a boost in sales of 10% or more from such programs.

Real-World Evidence to Deliver Better Patient Outcomes

Perhaps the most significant ongoing change in health care is the shift toward value-based health care, which assesses the value of treatments (including drugs) on the basis of outcomes (benefits and risks) achieved per amount of cost invested.

While, historically, costs were relatively easy to measure, outcomes were not tracked as systematically as they are today, so it was not possible to make such comparisons comprehensively. Big-data analytics, however, provide the tools to assess these outcomes.

One of the greatest benefits of big data and advanced analytics is that they enable physicians to better match patients with treatments. This is particularly true in oncology, where there is often a wide range of treatment options and where the choice of treatment can have a major impact on health outcomes.

For example, clinicians have used advanced analytics to predict any given patient’s response to certain treatments for colorectal cancer far more accurately by combining information about the tumor and its immunological environment with observable symptoms¹ 1 Galon et al., “ Towards the Introduction of the ‘Immunoscore’ in the Classification of Malignant Tumours,” Journal of Pathology, 232(2) (January 2014): 199–209. Notes: 1 Galon et al., “ Towards the Introduction of the ‘Immunoscore’ in the Classification of Malignant Tumours,” Journal of Pathology, 232(2) (January 2014): 199–209. , which previously were all that were available to practitioners for making decisions. Although this is still a new technique, early studies suggest that relapse could be prevented in more than 10% of colorectal cancer patients² 2 Anitei M.G. et al., “ Prognostic and Predictive Values of the Immunoscore in Patients with Rectal Cancer,” Clinical Cancer Research, 20(7) (April 2014): 1891-9. Notes: 2 Anitei M.G. et al., “ Prognostic and Predictive Values of the Immunoscore in Patients with Rectal Cancer,” Clinical Cancer Research, 20(7) (April 2014): 1891-9. . (See Exhibit 4.)

Reaping the Rewards: Big Data, Bright Future

For biopharma, the potential of big data is very real. Reaping the rewards, however, is still a matter of clear goal setting, strategy, and execution. In our experience, the companies that are achieving significant results do three things well.

Focus on value. Smart companies avoid the morass of big-data jargon and hype. Instead they prioritize where they use big data on the basis of impact and feasibility; they move from generalities to well-defined applications; and they spell out their approach and their goals very specifically. They don’t get distracted by gaps in their data, dwell on how they get it, or worry about whether they’re accessing all the data that’s available. Some companies take a portfolio approach, exploring options with different technologies, testing and iterating, learning from the outcomes, and adjusting their approaches accordingly.

Think strategically about capabilities. While many biopharma manufacturers have both the right tools and growing access to data, relatively few thus far have developed the capabilities to leverage big data fully. Proper application of big data to business problems requires employing data scientists who have a cross-disciplinary ability to translate domain-specific needs into analytical solutions. Thinking carefully about how to cultivate such data scientists internally, or acquire them externally, is a key differentiator of successful companies. While access to a wider variety of data sets, more sophisticated data architecture, and new data platforms all remain important, bringing the necessary talent to bear is increasingly the pinch point that companies encounter.

Develop operating models that work. To succeed with big data, teams need to work across organizational divisions—drug discovery, clinical development, manufacturing, sales and marketing, and so forth. Establishing a productive organizational model for big data is a new type of challenge for most biopharma companies. They must answer questions such as, Who owns the data? How are activities funded? How are big-data analytics governed? While the big-data programs that drive value share many success factors, there is no one “right” way to organize a big-data effort. Programs vary as much in their goals, structures, and approaches as do the companies that create them.

Successful companies believe that traditional organizational structures are unlikely to work and that new models need to be built, adjusted, and perfected over time. They prepare for a journey that is likely to take years. They understand that an important part of the trip is identifying, and acting on, areas for improvement. Despite their successes, many think that they can do a better job of readying their organizations to act on big-data insights and execute big data–based plans.

It’s still early days for big data in the biopharma industry, but the results so far point to significant gains in multiple areas for companies that get it right. Those that do will start with a clear vision of where they want to go and apply the lessons they learn with both rigor and flexibility as they move forward. They will also combine strategic foresight, adaptive experimentation, and excellent execution to make the most of the opportunity.