Science project – HSBC’s Peter Serenita and BBVA’s

Rapid advances in big data analytics could open up entirely new frontiers for financial institutions, allowing them to process the vast amounts of data they have at their fingertips better, faster and cheaper. Not only will this allow them to service their customers more efficiently and in a more personalised way, but everything from cybersecurity to fraud prevention is set to benefit. The question is: can the banking world keep up with the astronomical speed at which data science tools are evolving?

The amount of digital data available to us today is unimaginable and fast on the rise; a 2011 McKinsey Global Institute (MGI) report on big data predicted that the amount of information and data across all sectors of the global economy was predicted to increase at a rate of 40% annually. The problem, of course, lies in the impossibility of ever having enough labour to manually sift through it all, let alone make any sense of it.

Enter big-data analytics, or as it's also known, data science. So fast has this field advanced over recent years that the latest and most powerful analytical tools allow their users to not only capture, process and extract actionable insights from structured data - the sort you might find in a spreadsheet - they can do the same from the unstructured kind, which includes social media interactions, internet searches, phone calls, emails and GPS locations.

When all analysed together, the insights that can be gained into everything from customer preferences to the behaviour patterns of hackers are unprecedented, and increasingly - and most excitingly - they're being used to build predictive models, answering questions like 'when might a fraudulent transaction take place?' and 'when should I send a new offer to this particular customer?'.

Opportunities for banks

For the banking industry, which is sitting on more structured data than any other (every banking transaction is a nugget of information) - and that's before one even starts thinking about unstructured social media, mobile or GPS data - the implications of being able to analyse this vast resource better, faster and cheaper are enormous.

"Big-data technologies will have a significant impact on the way financial services firms can use data. The amount of data and the sources of data continue to grow at a rapid pace, and this trend will continue. As with any statistical analysis, the more data used in the analysis the more accurate the results - assuming the correlation of the data is appropriate," says Peter Serenita, HSBC Group's chief data officer. "This will allow financial services firms to better service their customers by understanding them in a more holistic way. Financial services firms can customise their offerings or provide more targeted advice to their customers based on the customer's needs and the changes in the marketplace."

One source from which banks can gain valuable insight into customers and the wider economic situation is social media. For example, a study by the European Central Bank (ECB) in 2015 found that Twitter has "a statistically and economically significant predictive value" when it comes to international share prices, while research published in 2014 analysed 19.6 million geolocated 'tweets' from across Spain to produce an accurate picture of regional unemployment.

Social media can also offer insights into which competitor products are popular among customers, as well as giving banks instantaneous feedback on the effectiveness of their own marketing campaigns and new products. The key to success with this type of analysis is for banks to take their cues from the social media platforms themselves and act quickly on the insights they gain.

Yet improving the customer's experience isn't the only way big-data analytics can benefit banks; data science can also help strengthen risk management in areas ranging from fraud detection to cybersecurity. Indeed, the most advanced analytic tools can detect subtle patterns and associations in the behavioural and transactional data left behind by fraudsters - from emails to social media interactions, call centre notes to agent reports - and use this to build predictive models. Similar techniques can also be used to discover, prevent and counteract cyberattacks.

Rather than focusing on just one area though, many banks are experimenting with the gamut of data analytics opportunities.

"There is not much room to differentiate yourself by focusing on single areas. We drive our efforts by projects instead and these typically have a cross-area reach - from user experience to pricing through security, risks, corporate and investment banking. We put a lot of emphasis on improving the analytical capacities of any area of the bank. The 'datafication' shall be organic and largely homogeneous across the bank," Marcos Bressan, global lead for analytics and quality at BBVA remarks.

A central component of BBVA's data strategy is a global programme to transform data-miners into data scientists. "The target was to provide our teams of commercial intelligence with world-class capabilities and resources (data, skills and infrastructure) to help the bank and clients make the right decisions," Bressan says.

Balancing act

With so many opportunities for banks to use big-data analytics across the business, entire departments devoted to the discipline have become essential - although it's also important to make sure there is some level of federation according to Serenita.

"In big-data analytics, there needs to be a fine balance between centralisation and federation," he stresses. "The sourcing and movement of data within an organisation needs to be well managed so that the data landscape doesn't become overly complex and redundant, but it is important to not treat all data the same. Access to external data can be managed more locally if the data will predominately be used on a local level. Lastly, the use of the data in analytics needs to be close to the business (or end users) so the consumption and use of data should be federated.

"The big-data technologies make it relatively easy to develop a business hypothesis, test that hypothesis, and then determine if it is useful, needs tweaking or did not pan out. This iterative nature works better when close to the business and therefore federated across the organisation."

Of course, with data science being such a young field, banks' approaches are also likely to change over time.

"We started centralised, to attract and build skills and are now in the process of decentralising this knowledge across the whole bank," Bressan notes.

Challenges ahead

While the potential for big-data analytics in the banking industry is unquestionable, it isn't proving an easy task to get it right, not least because of the rapid speed at which data-science tools are continuing to advance.

Indeed, for Bressan, the biggest challenge he and his team are facing is ensuring that the company's centres of excellence or "hubs of data science" keep evolving at a faster pace than ever, whether they were started up internally or acquired. "We need to maintain an excellent level of analytical quality," he stresses.

Moreover, more tools and experts don't necessarily make for better data analytics capability, Serenita believes. "The ability to leverage the best tools without creating a proliferation of tools and developers experienced in these tools makes for an intricate balancing act," he remarks.

"The way to address this is to recognise that not all tools are created equal. The tools near the bottom of the software stack (for example, file system, security and so on) require more standardisation than the tools near the top of the stack (analytics tools) and the tools 'in the middle of the stack' make for an interesting discussion. It is always about the balance between business benefit versus cost to the organisation from a total cost of ownership perspective."

Meanwhile, another challenge operators are facing is how to communicate their data analytics strategies to customers effectively. "We need to convey a clear and transparent message to our clients in terms of what information we manage, how we extract value from it and, most important, how this is of value to our clients," Bressan notes, adding that regulation is also an important area for banks to consider. "We need to help regulators shape the legal framework reasonably. They need to understand the pros and cons that any decision that they could make might have on society."

Finally, the issue of how to make the data science tools of 2016 work with the older legacy systems in which much of banks' data is stored is one that's proving difficult for many operators. At BBVA, resolving this fully will be an ongoing process. "The bank has invested in its core platform, and has solid and global access to its core operational systems, and we have in place an ambitious plan to update our data architecture towards a full XaaS approach," Bressan notes. "The combination of both systems has already proved its capability to enable data science and capture the value of data across most of the bank's business units."

Crucially, Bressan is keen to stress that the effort required to overcome the challenges associated with big-data analytics will be more than worthwhile in the long run. "Banks unable to use data will understand their customers and context less, and will be unable to provide relevant experiences, or to facilitate opportunities for their customers, clients and employees," he believes. "Furthermore, the efficient use of data is what enables the levels of automation required to compete in today's environment."

Serenita agrees that, if used right, big-data analytics will be a win-win for the financial services sector. "It is hard to
say whether banks that don't use big data will be left behind. As with everything, it is about the business outcomes," he concludes. "Some companies using big-data technology will use it well and some may not, but I do think that those using big-data technology will have the opportunity to make advances that will help their customers and their businesses."