As a career, Data Science clearly is a winner. According to a recently released study by Burtch Works, Salaries of Data Scientists, median salaries can range from $91,000 for those with one to three years of experience, up to $250,000 for managers leading teams of ten or more. Median bonuses for top level managers are topping $56,000, and Data Scientists can realize 16 percent increases in their median base salary when changing jobs, notes the study from the executive recruiting firm that specializes in the placement of quantitative business professionals.
This is all happening as large companies are starting to hone in on the talent. Major corporations want in on the business performance benefits seen by the tech startups that were early to embrace Big Data analytics, and Data Scientists like the career stability those businesses offer in comparison to startups, Burtch Works reports. So much so that the proportion of Data Scientists employed by startups declined from 29 percent in 2014 to 14 percent in 2015, the report says.
As wonderful as this news is for Data Scientists, the question for the enterprises that are eager to access their talents revolves around how they actually can realize the greatest value from these individuals and the abilities they have of viewing data through a quantitative lens. A word of caution comes from John Akred, CTO of Silicon Valley Data Science, a consulting firm of Data Science and engineering teams specializing in data-driven product development and business transformation.
“A Data Scientist is very important if you have a very digital business,” he said during the CDOvision 2015 Conference at Enterprise Data World (EDW). “But I would submit that if you simply view it as hiring a Data Scientist, you probably are headed towards disappointment.” As to why, he commented that they must function as part of cross-functional teams that works iteratively across disciplines, including database management, quantitative, and engineering functions – not in silos. That way, the experimentation that can lead to successful insights can be enabled without bringing the company down.
The idea that a Data Scientist cannot succeed working in isolation has been raised before. A few years ago, for example, McKinsey Global Institute reported that for every Data Scientist, organizations will need ten data savvy managers with the skills and understanding to make decisions based on data analysis. This was noted in a report released last year by data analytics SAS which concluded that it is difficult for any single Data Scientist to have a complete mix of skills – technical, mathematical, creative, communicative – needed to answer all the needs of the organization. That means that companies should make data savvy managers a part of their mix and otherwise ensure they have complementary skills and experience represented in teams charged with delivering insights Big Data insights and follow-up actions.
Others at the recent CDOvision conference pointed to their own experience merging talents to ensure that the business sees results from its investments in making Data Science a critical part of the organization. For example, Robert Abate, Global Director, Enterprise Data Management and Analytics at Kimberly-Clark Corporation, discussed efforts he has undertaken in the past to mix internal resources and consultants with different skills sets – for instance, those with mathematics and statistical backgrounds and those with strong skills in understanding and using data findings – to create Data Science expertise in a market where there a single Data Scientist has a pick of dozens if not hundreds of job opportunities.
Data Science Working at Work
Despite the increasingly welcome embrace of Data Science in big companies and the positive results that can come of that, “Data Science still has to percolate over many large organizations,” according to Idriss Mekrez, Chief Technology Officer, Public Sector, at enterprise NoSQL platform vendor MarkLogic. “Currently the challenge in many organizations is that a lot of people are sitting on extremely valuable data, but they can’t change their business model fast enough to take advantage of that. They don’t have the skills and processes in place.”
Among the growing number of companies starting to see returns from building up their Data Science muscle is TD Ameritrade, where Chief Data Officer Derek Strauss hired a Chief Data Scientist to help the company take things to the next level when it came to driving deep insights and finding opportunity areas for the business.
Previous to this, the organization had many silos of analytics teams, often frustrated by the complexities of data and inefficiency of existing tools, and probably replicating at least 20 percent of workloads. “There was good raw material in those teams,” Strauss said at the CDOvision event. Putting a Chief Data Scientist in place to lead them as a federation functioning as elements of a center of excellence was the path to educating everyone involved about what Data Science really was, how it could be used, and why it was different from how analytics had been done in the past – as well as to figuring out a way to do things once that the whole enterprise could share.
With a real Data Science track in place, the organization finds itself able to get out of the mode it was in – that is, it was stuck continuing to ask tried-and-true questions of its data. “Exploration and discovery was lacking,” Strauss said. Things have changed since then. For instance, it’s now mining its own Big Data trove of text documents, such as reports, for information that could be interesting to investors if presented as data in a format they could use. And, he added, a world of interest can be found in grouping outliers among client segments. “Very often the way we deal with our clients is based on segments dreamed up over the years by marketing people,” he said. But the “real gold lies in outliers, those who don’t easily fall into segments and the clusters that start appearing.”
Now, TD Ameritrade is beginning relationships with at least one university to launch a crowd-sourcing program. It’s loaded data onto an Amazon Cloud instance and opened it up to students, hoping to nurture and access the best Data Science talent. “We told them you play with it, tell us what you think we should look at, what kind of studies you think you can deliver back to us, and we’ll take the top 5 to 10 ideas, give you some seed money and knock yourselves out,” he said. “Then come back and present to us and we will have our business partners directly plugged in.”
Leveraging the expertise of Data Scientists – and those working with them in related positions in their orbit – to grow business insights will continue to expand as a key strategic differentiator for many companies. Indeed, predicted Abate, those who don’t advance the maturity level of their organizations by bringing one (or more) Data Scientists onboard risk being out of business a short few years from now, in his opinion.
That’s not an argument for rushing things beyond what an organization is ready to embrace, though. Akred reminded attendees at the event of the importance of taking a software engineering discipline such as Agile for building reliable and predictable systems and applying that methodology to Data Science and data-driven projects. “There are so many possibilities of what you can do with data,” he said. But “the ability to do great harm with data is as profound as the ability to do great good with it.”