Much has been written about the limitations of data silos. Less frequently discussed are the limitations that attend siloed data skills or knowledge. Some data scientists, for example, have the expected deep understanding of data, but lack the knowledge of business applications that would enable the data to have the optimal impact.
Health Care Service Corporation (HCSC) approaches data talent acquisition in a cross-disciplinary way – the company hires data scientists from both within and outside the industry, across seven different business lines. In the last three years, HCSC has hired more than 70 data scientists, including a combination of Ph.D.s, statisticians, and actuaries with experience within and outside of the healthcare industry.
In addition, the company trains data scientists using a “trilingual” model, enabling them to better understand the data, analytical techniques, and business context.
Vijay Murugappan, VP of enterprise analytics and process improvement at HCSC, spoke with Data Informed about the company’s approach to identifying, hiring, and training trilingual data scientists.
Data Informed: What does it mean to be trilingual in data?
Vijay Murugappan: Analytics can be effective only if it enables decisions or actions. In order to do that, we believe three key skills are needed – deep understanding of the business domain, training in analytical methods and techniques, and the technical ability to access and stage the data. Historically, the focus has been on the latter two areas.
First, a deep knowledge of various business areas is critical to effectively perform analytics – this kind of knowledge comes from first-hand experience with operational roles in key business areas. Secondly, experience in statistics, modeling, economics, and related analytical disciplines is important to ensure that data points to correct business observations. Finally, the ability to acquire, aggregate, and present data for analytical purposes is crucial to business insights. These trilingual skills are rare and are typically developed over time. It’s absolutely crucial not only to have domain expertise and analytical thinking, but also to be able to tell the story using the data as well.
How does the philosophy of being trilingual in data inform HCSC’s approach to hiring and training data talent?
Vijay Murugappan: Our talent strategy focuses on expanding our sourcing to attract stronger and more diverse analytics professionals, recognizing that the skills necessary to interpret, communicate, and execute the business insights are as critical as the technical skills necessary to gather and analyze the data. More directly, we are building a competency model based on these three “languages” to hire the best talent and develop and measure our current staff.
We believe that deriving value from the use of analytics requires focus on both the producers and consumers of analytics. As a result, we have developed a deliberate approach to enhancing the skills of those consuming analytics. We are accomplishing this through a combination of executive training, increased self-service analytics, and aligned incentives.
How does this model differ from how the organization used to approach hiring data talent, and what prompted the change?
Vijay Murugappan: It differs in a number of ways. First, we have implemented a cross-divisional talent strategy that allows for different divisional analytics areas to share a common talent pool, have visibility into where the analytic talent needs are across the organization, and potentially assist peers in the interview and selection process. The hiring and staff-development processes were fairly siloed in the past.
Second, we have a shared talent-acquisition resource who recruits cross-divisionally and regularly reports on the status of open positions and candidates in the pipeline for analytics and data-related roles. We are excited about building a cross-divisional rotation program so new talent can experience analytics roles in multiple areas of the business before landing somewhere more permanent. We also are enhancing our university partnerships by hosting events onsite with schools to recruit students and creating development programs with local universities for our employees to build upon their knowledge and skills.
The simplest difference in the new model is the greater and more consistent collaboration across divisions. This has been a part of our strategy all along, but it’s now top of mind for all of us looking for the best talent.
What kinds of results is the company seeing from this approach, in terms of hiring, retention, and the quality of the employees’ output?
Vijay Murugappan: We’ve just introduced the comprehensive talent strategy – our focus for a tri-lingual talent base shifted about a year ago, and we have been implementing some of these tactics incrementally over the last 12 months. Within months of transitioning to this model, we saw an increase in throughput, improvement in the quality and relevance to line of business leaders, and a flattening in costs associated with analytics resources.
Has this approach widened the pool of candidates? How has this impacted the company’s ability to identify and hire qualified candidates at a time when it’s widely acknowledged that there is a serious lack of qualified data scientist candidates to meet industry demand?
Vijay Murugappan: We agree that these trilingual skills are extremely rare and typically developed over long periods through exposure to all of the areas above. Retail and consumer products companies have developed these capabilities over the last two decades. Catching up to that level of maturity will require a structured talent acquisition and development plan. In the short term, three approaches can help fill the gap: A vertically integrated team of individuals who possess each of these skill sets, a rotational plan across functional and analytical domains, and a co-sourced model with analytics partners to fill short-term skill gaps and create an on-demand model for long term skill acquisition.
Does this approach to hiring weigh nontraditional factors in a candidate’s background more heavily as opposed to trying to find someone with the exact background typically sought for this role?
Vijay Murugappan: We have expanded the factors for consideration to include more tacit business skills that are not typically required for traditional analytics roles. We are hopeful that the inclusion of these non-traditional factors will help to diversify our talent pool and, ultimately, help us achieve our objectives. We will be evaluating any impacts of an industry learning curve as well as individual performance at regular intervals.
Has HCSC noticed any differences between the hires from within the industry and from other industries?
Vijay Murugappan: Inherent in our trilingual model is a bias toward domain knowledge. As a result, we have noticed that new hires from within the industry have a shorter ramp-up period than those from other industries. That said, our talent acquisition and development approach takes a balanced view on supplementing out-of-the-box thinking from other industries with the deep knowledge of the health plan industry.
What are some other companies/industries that you have seen that you think are taking the right approach to hiring analytics talent?
Vijay Murugappan: In developing this approach, we researched and/or directly talked to Fortune 100 companies in a wide range of industries – retail, agriculture, consumer products, high-tech, social media, and entertainment – as well as healthcare peers and competitors. We picked companies that we believed had an extensive culture of analytics. In some of these companies, analytics ceased to be its own thing and was an embedded competency among business leaders. As we parsed out the common elements that made these companies successful, we identified that they possessed varying degrees of trilingual elements.
Subscribe to Data Informed for the latest information and news on big data and analytics for the enterprise.