How can CFOs get the most from data science and data scientists?

25th January 2016

This is one of those questions that immediately prompts other questions, such as ‘What is data science?’ and ‘What is a data scientist?’ So we may as well begin with these two, before we return to the question of how finance chiefs (CFOs) can get the most from them. FSN writer Lesley Meall goes looking for answers.




In broad terms Data science is deep knowledge discovery through data inference and exploration and a data scientist represents an evolution of the business or data analyst role.  However, what sets the data scientist apart is strong business acumen and the ability to communicate findings in a way that can influence how an organization approaches a challenge. 

What’s important is that data science techniques used to analyse holistic data in an automated way, eliminate human bias and our limited capacity for complexity. The data itself can determine the clusters (or categories) that emerge and these can then be assessed for usefulness. 

Perspective is everything

On the question of how CFOs – and other finance professionals – can get the most from data science and data scientists, there may be almost as many answers as there are accountants. Some of you may want to find out how finance can work with data scientists to deliver deeper or sharper insights from available data; some of you may want to know if taking a more data-driven approach will require specialist software or systems; some of you may want to reinvent yourself as a data scientist.   

So, where do you begin? When you first hear people talking about data science it can sound as if you need to learn a new language; and if you want to be a data scientist you will need to do this and learn new skills. If you are more interested in collaborating with data scientists, integrating them into your team, or finding out what they can bring to the finance function or to business strategy planning, then you’ll simply need to acquire sufficient vocabulary to communicate. 

“Don’t get lost in arcane technology or concepts or buzzwords. It’s very easy to do this. When you are talking to people in the IT function or talking to data scientists, it can seem like data science is something that’s very alien to finance, but it is very relevant for the finance function,” says Roger Fried – and he should know. Fried is a senior data scientist at Teradata Aster (a global leader in big data analytics and data-driven business insights), who previously had a long and successful career in finance. 

Masters of big data

Data scientists are fast emerging as the masters of big data and its analysis; they are already well established in data-driven companies such as Amazon, Facebook, Google and Teradata. Yet most organisations have not been shaped by their use of big or (even) not so big data. Most organisations have yet to explore (let alone exploit) its transformational potential. So there is plenty of scope for enterprising finance professionals to take a leading role in the transition. 

The professional services firm EY describes big data as an ‘extraordinary opportunity’ for clever finance leaders, who can see the potential of sophisticated analytical and data science-based tools. “I see the CFO supporting analysis around the whole of the organisation, driving business insight with business information,” says Dorian Redding, an EY partner in Los Angeles. CFOs can become innovation champions who shape the new world where information becomes a conduit to insight. 

Data science may be the next step in the evolution of financial analysis – and an improvement on the statistical tools, techniques and strategies that we have come to rely on. “If you take different data sets with real world complexity and boil them down to a ‘mean’ and its tendencies you immediately have simplification, which allows you to say something about the world and have some degree of confidence. But if you step back, you recognise that reality is not like this,” says Fried. 

Focus on the details

Averages are radical simplification and are inherently limited. “It was fine to accept this before, when it was our reality and was all we could do,” says Fried, “but it’s no longer acceptable, now that it is possible to have a different point of view because we have tools and processes that didn’t exist before.”  Now we can use massive amounts of data, tools and techniques that can deal with detail holistically, and algorithms that can identify clusters, we can overcome human bias and other impediments to analysis. 

These impediments include accounting categories. “Accounting is pretty much the process of pouring bias into transactions then looking to see what you get. The limitations jump out at you when you take a holistic data science perspective,” says Fried. Traditional approaches are still needed in areas such as regulatory reporting, but tasks such as financial forecasting and fraud identification are among those what that may be better managed and communicated using more data-driven techniques. 

If you do want to complement traditional financial and accounting analysis with big data analytics and data science methods, will you need to update your technology infrastructure? Maybe, maybe not. Fried dismisses spreadsheets: “Excel is not the tool for data science.” You need to handle massive amounts of data, massive amounts of slicing and dicing, many process steps and do all of this very quickly, so think about whether the tools and processes you have can do this. 

Data science builds on solid foundations. “You need good data quality and data hygiene,” says Fried, who also emphasises the importance of integration within your systems for business, accounting, budgeting, asset management, employee hierarchy, travel and expenses, customer relationship management and so on. He says: “What matters is that all of the data is brought together and accessible in a relatively rapid analytic framework and you don’t have to wait for IT to help you do your analysis.” 

Integration between products such as Tagetik (corporate performance management; CPM) and Qlik (visual analytics) may offer a way forward. “Together, they provide the ability to effectively analyze data, diagnose a situation and accurately forecast what is likely to come next with no dependency on IT,” says Marco Pierallini, co-CEO, Tagetik. But Fried cautions against getting bogged down in the detail. “When it comes to data science, there isn’t just one specific ‘right’ technical approach,” he says.