Data artistry: finding business value in data

Data artistry: finding business value in data

Data artistry: finding business value in data

Data science is the new ‘rock star’ business profession that picks up where big data and analytics leave off. Tim Phillipps argues that, to become more predictive, organizations should recruit those who see the future and others who can visualize it for the rest of us

The ‘data scientist’ is a new breed of analytics professional who takes an experimental approach to analyzing data.

Professor Thomas H. Davenport, visiting professor at Harvard Business School, is one of the world’s foremost authorities on data science and a senio adviser to the Deloitte Analytics Institute. He says a data scientist is “part hacker, part quantitative analyst, part trusted adviser and part scientist.”

Organizations are interested in data science because it has the potential to transform business models and create new ones. It also replaces gut instinct with data driven decision-making and is the basis of a new form of competitive advantage that is difficult to replicate. Consider the following examples.

LinkedIn owes much of its success to the ‘people you may know’ function, an experiment instituted by its former chief data scientist, DJ Patil, who also coined the term that applies to the new profession.

GE realized that the growing amount of data produced by sensors in the firm’s gas turbines, jet engines and other industrial products could be collected and analyzed in real time to improve machine performance and operations, turning unscheduled maintenance into scheduled maintenance and identifying potential operational disruptions before they occurred. The insights it has drawn from its devices are already improving the efficiency of customers’ businesses, saving them anywhere from tens of thousands of dollars to millions for each analytics based service.

Worldwide gaming, resorts and hotel empire Caesars Entertainment analyzes data from slot machines to present real-time offers and marketing promotions to its patrons.

While organizations are building their in-house ability to use data to peer into the future, Kaggle, a platform that allows organizations to crowdsource data scientists, has attracted the backing of Silicon Valley high-flyers who were behind the online payments platform PayPal. Kaggle allows organizations to farm out complex business problems to data scientists around the world in return for cash prizes.

Although data science is the future of decision making, organizations must hurdle two obstacles. First, they must bridge a skills gap and, second, understand and act on the predictive business insights that data scientists produce.


A survey carried out last year by EMC revealed that two thirds of data science practitioners expect demand to outpace supply over the next five years.

Even first-year business students know what happens when demand exceeds supply: prices go up. We’re already seeing this, with organizations willing to pay top dollar for data scientists. But this is a risky strategy involving considerable expense and with no guarantee that an individual will be a good cultural fit.

So where can organizations find the data scientists of tomorrow? Mark Grabb, analytics technology leader at GE’s Global Research Center in New York, suggests PhD graduates are likely to remain a major source of data scientists in coming years.

The private sector, universities and industry bodies like INFORM are also developing tailored data science courses. However, there’s a question around whether these courses are teaching the right skills.

Most courses that are currently being developed focus on the data cleaning, mining and analysis skills necessary for data scientists to do their job. But less attention is paid to the ability to communicate those insights, says Davenport.

“Communication still doesn’t have the importance it should in most programs,” he says.

Bridging the communication gap is of paramount importance. It’s great to have a team of big brains that can crunch data, but their efforts are useless if they can’t explain what they find to decision-makers.

Communicating the message hidden within data requires a rare capability to combine the right-brain analysis with the left-brain creativity needed to communicate with non-specialists. This isn’t just about running experiments with data; it’s about being to paint a picture with the results. In many respects, this skill is more art than science, namely ‘data artistry.’

There are two key elements to data artistry. First, the data artist must be able to speak both the language of data and the language of business. Second, data artists need to be able to tell stories with data, understanding the most effective way to communicate results, and being able to work with different media to deliver those insights. This typically involves using data visualization tools.

In the past, these might have been static reports, PowerPoint presentations, heat maps and charts. Now, data artists are increasingly turning to new and more creative ways of telling stories with data. They are using audio-visual presentations, dynamic charts, interactive games, 3D models and apps to convey meaning.

In the short term, organizations will struggle to find a professional who embodies a data scientist and data artist in the one ‘rock star’ package. A better approach is to create rock bands: data science teams with a breadth of expertise.

By focusing on forming teams, organizations can blend a mix of top skills without staking everything on a single risky hire.

It also enables organizations to fish in a wider ocean of talent, rather than the existing pool of data professionals. You can hire in specialists in particular areas: data scientists to run experiments, industry specialists who can apply analytical insights to their work, or even computer games programmers and animators to devise data visualizations.

Finally, tapping into external sources of expertise can bridge skills gaps, benchmark your own approaches against the latest advances in techniques and technology, and source talent in specific analytical areas. The Deloitte Analytics Institute is one such center of excellence; alternatively, organizations like Kaggle offer affordable access to data scientists around the world on a project-by-project basis.

Insurance and data science

The insurance industry is just beginning to take advantage of this new discipline to improve the way it operates in several ways, particularly around pricing risk:

  • By using GIS information and services like Google Earth alongside information in other databases, insurers are increasingly able to map properties to an individual level, allowing for more effective pricing of risk, especially in areas prone to flood or bushfire.
  • Monitoring equipment—‘black boxes’—can now record every aspect of how a vehicle is driven, from distance traveled to how hard a driver hits the brakes. This information could be used to price risk and assess claims. This technology is not limited to motor insurance either: as common devices such as phones, TVs and even ovens become ‘smart’ and connected to each other, the ensuing tsunami of information could be used for other types of coverage too.
  • Loyalty cards and other data from partner and sister businesses (such as supermarkets) can be used by insurance companies to garner more information about lifestyle choices and to more effectively price risk. This should be carried out with individuals’ permission, however.

Debunking the myths

Data science is a term subject to confusion and misinformation, partly due to its fast-moving and emerging nature. Here we debunk the most common myths.

Myth #1: Data science is just about big data 

Data science is often linked to big data, partly because data scientists often work with extremely large and complex datasets. But there is much more to it than that. Data science uses scientific methods to create experiments that test hypotheses based on any available data. These concepts may glean unique insights from smaller datasets as well as large ones.

Myth #2: Data science is just a trendy rebrand of business intelligence

Data science and business intelligence are very different. Business intelligence and its associated analysis is typically a backwards-looking exercise, intended explicitly for reporting purposes.

Data science, meanwhile, involves running experiments on data, rather than seeking particular outcomes. It’s a forward-looking approach that relies on asking “what might happen in future?” rather than stating “this happened in the past.”

Myth #3: You can buy software that will do the hard work for you

At its core, data analysis involves people. It requires skilled professionals who can speak the language of business as well as the language of data. It requires critical thought, scientific judgement, creativity, pragmatism and common sense. This is very difficult to outsource to software (for now at least).

Myth #4: Data science is too complex for ‘normal’ business people

Data science is only successful if the insight it uncovers can be articulated to non-specialists and key decision-makers. This element of the analytics chain—which we at the Deloitte Analytics Institute call ‘data artistry’—requires a particular set of skills and is one of the most crucial aspects of data science.

Myth #5: Data scientists work best alone

Yes, there are some people who can do it all, but these people are rare and expensive. It is clear that the most successful data science is done by teams of people—scientists, communicators, other analytical professionals, those with domain expertise, and consultants—combining their expertise to derive real insights that businesses can use.

Tim Phillipps is the DTTL global leader of Deloitte Analytics and is responsible for Deloitte Analytics Institute Asia. He is an authoritative voice on the application of analytics as a source of managerial insight and competitive advantage.