Playing vs. Leading with (Big) Data
If you were asked to identify a major corporation without a current strategy or, minimally, ambition to pursue enterprise solutions leveraging big data, you will be hard-pressed to identify one. In most cases, big data initiatives have been positioned to help organizations better address potential improvements in areas ranging from de-risking product development, monitoring critical systems, and optimizing global operations. To ensure definitional clarity, big data initiatives, for the purpose of this discussion, include any efforts to create infrastructure to interrogate data sets too big to be analyzed by traditional means. Unlike other instances where disruptive technology deployments have occurred in a comparatively homogenous fashion – mobile applications or even cloud storage come to mind – big data initiatives have emerged more amorphously, thereby preventing a broad group of public and private organizations from visibly aligning around the benefits that related technologies offer. Ask three executives about how big data initiatives have evolved in their respective organizations and you’ll likely get three different responses. Perhaps as a result of these dissimilar approaches and a relative lack of shared standards to be measured against, there are relatively few leaders among the many players in the big data space.
Differentiating factors will be discussed here, offering insights into why developing leadership in big data has been anything but a straightforward task while highlighting several common characteristics of successful organizations in the big data space that have allowed them to emerge as leaders.
The Data Challenge: The most sizable challenge related to demonstrating leadership in the big data space relates to the ‘nature’ of the data, not the ‘availability’ of relevant data. IT systems infrastructure has generally evolved adequately to support the amount of data to be stored, processed, and visualized. The greater challenge pertains to data collection as integrating data assets from heterogeneous sources is not a trivial task and complexity can increase disproportionately based upon the nature of the data sources. As an example from the biomedical space, and in the quest for improved overall health, precision medicine approaches involve a wide range of data sources from providers, insurers, diagnostic laboratories, electronic health records, and the biopharmaceutical industry. These data are inherently heterogeneous and frequently difficult to normalize for meaningful analysis. Ensuring data of the highest quality are available and accessible across the entire organization is done more effectively by leaders in big data.
Big data initiatives have been positioned to help organizations better address potential improvements in areas ranging from de-risking product development, monitoring critical systems, and optimizing global operations
The Analytics Challenge: Bold tactics operationalizing big data ambitions, specifically as it relates to the execution of big data strategies, require a focused and disciplined approach related to systems infrastructure. Many organizations continue to house critical data assets in siloed repositories where data cannot be easily retrieved nor shared. Normalizing heterogeneous data is key and pursuing a disciplined approach to data analysis is important. Unless data stewards are empowered to execute on critical systems infrastructure decisions, the likelihood of realizing fit for purpose infrastructure that is transparent, universally accessible, and is characterized by interoperability is low. Many available tools promise to be data agnostic as well as easy to use for end users, but more often than not that is not the case. Take the example of text and data-mining functionalities, which have shown great promise and relevance, yet advanced skills and decision-making capabilities in natural language processing, semantic enrichment, and cognitive computing are required to generate superior and relevant insights. Many companies underestimate the complexity and skills upgrades required to succeed here. Organizations that don’t demonstrate the ability to leverage evolving technologies systematically and deliberately tend to be players but not leaders in big data.
The Prioritization Challenge: Not every question that can be addressed by deep text analytics and or cognitive computing warrants being pursued in the context of an organization’s strategic priorities. In big data more so than in many other areas where innovative technologies have been deployed, the prioritization of goals pays great dividends. Wanting to solve any and all business challenges with the big data ‘hammer’ is tempting, but has been shown to yield diminishing returns in insights. Take another example from drug development. The promise of personalized medicine and the availability of a genomic profile for every patient enrolled in a clinical trial have the potential of rendering the current gold standard for clinical trial conduct, namely the double blinded, randomized clinical trial, unethical because biomarker analyses and genetic profiles can be expected to determine which patient will actually benefit from a given treatment as well as to predict any undesired effects said treatment might have. Among the more popular applications of big data analytics in the biopharmaceutical industry are target identification, pattern recognition, and safety signal detection -- all aimed at surfacing new treatment options or making existing therapies safer -- each introducing a level of complexity that is difficult to manage and highlighting the importance of effective prioritization. Resisting the urge to boil the proverbial data lake is key and sets big data leaders apart from the players.
The People Challenge: As is so often the case, those with the requisite informatics expertise who assemble the datasets will frequently lack the subject matter expertise to index these data effectively and determine their relevance. Unless time is spent here, entire big data initiatives can fail, and huge investments go to waste. In big data applications more so than in many other areas of development, top talent is scarce, and the difference between surfacing meaningful results when mining heterogeneous datasets and simply producing noise can make or break the project. Companies that have mandated scientists to be ‘bioinformatics proficient’ beyond their basic scientific expertise, thereby generating data with a compatibility goal in mind, have seen greater returns on their investments in informatics infrastructure.
Disruptive effects from players like Google and Amazon entering the health care field are widely expected, but not because these are organizations with legacy strength in therapeutics, drug manufacturing, or regulatory affairs. Their ability to leapfrog established players will be the direct result of a superior ability to consume a broad range of data assets in innovative ways and inform decisions only few established companies in the sector have been able to achieve to-date. In addition to a core engineering commitment, leaders in big data will be those companies with a relentless focus on initiatives that lead to the creation of fundamentally new business models. Focused investments in (competitive) insights generation, informatics infrastructure, and the talent required to improve operational performance across the value chain will allow them to unlock business opportunities they had previously limited reach into –and to ‘lead’ with data.