6 Ways To Get More From Data Science, IT
Data scientists and IT push the limits of what's possible -- whether that's operating more efficiently, taking advantage of new opportunities, or innovating. Here are 6 ways businesses can boost their effectiveness.
Businesses limit their potential when their IT departments are relegated to support roles. To be effective, IT and business leaders need to align technology strategy and business strategy and work toward common goals. The same is true for data science teams. They also need to work collaboratively with business leaders to make a strategic impact. However, what the data science team does, like what the IT team does, is not always well understood.
Data scientists, like CIOs, are expected to be business-savvy. Similarly, members of their respective teams should have a grasp of the business case so technology and data can be used for the strategic benefit of the business. Savvy CIOs understand the relationship between business objectives and technology initiatives. A good data scientist also strives to make a positive impact on the business.
However, to be effective, CIOs, data scientists, and their teams require the support of the organization -- including executive champions for important initiatives, adequate funding and tools, open lines of communication, and the latitude to make recommendations and suggestions. After all, there are always other technologies, methodologies, models, algorithms, and potential outcomes to consider.
Business leaders need to understand what technology and data can do. At the same time, they must also be prepared to sometimes hear what they'd rather not hear, including what cannot be accomplished with IT systems and data. Similarly, the organization may be required to make some adjustments for its own good, which may include changes to the organizational structure and reporting lines, as well as new approaches to problem-solving.
Whether a company is able to realize the maximum benefits from its IT and data science teams depends on a number of factors. We present a few of those considerations here.
Technology for the sake of technology and analysis for the sake of analysis have little or no practical value. The possibilities typically exceed what is practical or advisable, although not everyone might agree on the best course of action. Because different people tend to have different vantage points, there may not be common agreement about what the scope of an initiative should be, what it will cost, the time it will take, what its likely impact will be, and what the priorities should be.
"Too often data scientists are driven to build the optimal analytic model with less concern about how the organization might leverage that model to make or save money," said Ken Elliott, global director of analytics at HP Enterprise Services in an interview. "While the good data scientists can talk through the business use-case, there is often a very shallow understanding of the actual business processes required to leverage the analytics, and there's rarely a clearly articulated path to ROI. The business sponsors and data scientists need to be more realistic and business-minded about an analytic effort."
Highly competitive companies realize that IT can't be an afterthought. It has to be an integral part of the business. The same is true of data. However, some organizations try to bolt a data strategy onto an existing business model, when it may be more advantageous to consider how the business model should evolve to include data. In fact, data should be applied in the first place to determine the business model.
"Being data-science driven means embracing the field throughout the entire decision-making process," said Tom Fountain, CTO of distributed data company Pneuron, in an interview. "Doing so means everyone must be open to the results. Some companies may not embrace dissenting views or when the data results aren't what were expected. When that happens, it's seen as a potential threat."
IT teams have to integrate legacy systems and software, and IT and data science teams have to integrate legacy data with new data sources. While integration is necessary to enable greater value and insight, leveraging legacy systems and data can sometimes be a complicated and time-consuming task. It's a situation not everyone understands.
"Data scientists have to spend a significant amount of time building the data science platform itself, often weaving together narrow-purpose solutions with legacy technology, which means they don't get to spend enough time on the actual science part," said Nikita Shamgunov, CTO and cofounder of in-memory database company MemSQL in an interview.
Not all businesses require Hadoop developers or data scientists. But what's hot may sell, even though the decision may not be in the best interests of the company or the individual who has been recruited. It is not uncommon for organizations to put less thought than they should into why they need a data scientist in the first place or the questions they should ask a candidate during an interview.
"Data scientists are all the hype right now," said Michael Schmidt, data scientist and CEO of predictive analytics company Nutonian, in an interview. "Companies are hiring them for the sake of hiring them without knowing how to best utilize their skills. Businesses need to know why, and if they need a data scientist [they need to] know what they're looking to get out of the investment before bringing one on board."
Data scientists, like CIOs, don't work in isolation. There may be business analysts, data architects, DBAs, data visualization engineers, systems architects, statisticians, developers, business sponsors, and others involved, depending on the nature of the project. Even more important than titles are skill sets, especially since titles -- and the qualifications that justify them -- vary from company to company.
When full-time positions can't be justified (and even sometimes even when they can), companies may try to fill skill gaps with internal resources, which is hard to do well when specialized knowledge is necessary and scarce. In addition, organizations run the risk of losing good employees when the expectations of their capabilities, and their ability to handle an expanded scope of responsibilities, are not realistic.
"There usually aren't enough people with enough time to look at all the data or to try to do anything useful [with] it," said Tim Kirchner, director of data science operations at global services company Appirio. "Many companies simply lack the specialized skill sets in their human capital to do all of the work."
Data scientists, like CIOs, don't work in isolation. There may be business analysts, data architects, DBAs, data visualization engineers, systems architects, statisticians, developers, business sponsors, and others involved, depending on the nature of the project. Even more important than titles are skill sets, especially since titles -- and the qualifications that justify them -- vary from company to company.
When full-time positions can't be justified (and even sometimes even when they can), companies may try to fill skill gaps with internal resources, which is hard to do well when specialized knowledge is necessary and scarce. In addition, organizations run the risk of losing good employees when the expectations of their capabilities, and their ability to handle an expanded scope of responsibilities, are not realistic.
"There usually aren't enough people with enough time to look at all the data or to try to do anything useful [with] it," said Tim Kirchner, director of data science operations at global services company Appirio. "Many companies simply lack the specialized skill sets in their human capital to do all of the work."
-
About the Author(s)
You May Also Like