A look at data analytics in the financial services industry: use cases and challenges

Of all the issues that modern financial institutions face, there are few are as terrifying as fraud. In 2016, banks lost $2.2 billion in total to fraud. With fraud on the rise, data analytics in the financial services industry is the strategy being used to detect fraudulent transactions. Data analytics can help organizations to scour through customer data and detect unusual activity.

Data analytics is the term given to the practice of examining data sets with software platforms to find insights and information. Within the financial services industry, data analytics solutions can analyze data from transactions and determine whether the transaction was legitimate or not.

The data gathered to make this decision ranges depends on the software in use. However, Bolt suggests that when making a transaction online, your name, email, phone, IP address, time of order, payment method, currency, CVV response code, session ID are just some of the details that are analysed by analytics platforms.

The answer to dealing with fraud lies somewhere within this mountain of data. Every time you make a transaction, data is being generated on your activities and recorded. Having a software platform that can read between the lines could lead to a substantial reduction in global fraud.

Businesses struggle to make the most of data analytics due to skills shortage

Innovating new business models and maximising revenue and profits should be the next set of priorities for data analytics, according to Infosys

The banks using data analytics to detect fraud

The ability of data analytics and custom financial services platforms to detect fraud have encouraged many banks to hop on the data bandwagon. In fact, leading high street banks like HSBC have been using of data analytics for the detection of money laundering. Back at Google Next 2017, Dr David Knott, a chief architect at HSBC outlined a commitment to using data to combat fraud. .

Knot stated that “we’re working with Google to build machine learning models to increase our ability to detect genuine cases of financial crime and money laundering and to reduce the false positives so we can really focus on catching the bad guys.”

Institutions like Natwest have also deployed analytics solutions to great effect. In 2018, Vocalink Analytics and Natwest announced a program called Corporate Fraud Insights. Corporate Fraud Insights combined AI and machine learning to flag suspicious payments. The solution is reported to have prevented over £7 million in losses on behalf of customers.

Even though these numbers are promising, the predicament of data analytics in the financial sector isn’t as stable. Data analytics program are struggling to tell fraudulent transactions apart from false positives. Everyday, the transactions of customers are constantly being blocked by overzealous security measures.

Organisations across the globe lack data analytics maturity, says study

Harvard Business Review Analytic Services report reveals that only five per cent believe that their organisations are very effective at implementing modern data sharing, while 67% want to move towards that approach, showing that data sharing is very much on the agenda of organisations in the year ahead

The greatest challenge in data analytics: beating false positives

The principle challenge of fraud detection with data analytics is striking the balance between scrutiny and false positives. Fraud detection platforms must be calibrated enough to detect fraudulent activity without blocking legitimate transactions. The impact of just one blocked transaction can have have a substantial ripple effect.

For example, if a credit card is blocked, both the merchant and the bank stand to lose a customer. Not only is the customer’s purchase lost but there is also the change that disgruntled customers will be vocal about their bad experience with the brand. The costs of blocking customers can be just as substantial as letting fraud through the net.

Shockingly, 2.5 times the amount lost to fraud is lost fighting fraud. The cost of upholding a data analytics platform that has false positives is astronomically high. Thankfully, researchers and data analytics providers are coming up with innovative new ways to reduce false positive rates.

One of the most promising answers to false positives has come from Kalyan Veeramachaneni and other MIT researchers at the Laboratory for Information and Decision Systems who have developed automated feature engineering. The approach has over 200 features to use to measure every individual transaction. The platform monitors normal usage and can identify when the spender’s habits change.

Automated feature engineering has proved successful, reducing false positives by 54% when tested on a dataset of 1.8 million transactions. At the European Conference for Machine Learning Veeramachaneni concluded on the findings stating that “We can say there’s a direct connection between feature engineering and [reducing] false positives…’That’s the most impactful thing to improve accuracy of these machine-learning models”.

In other words, one of the keys to developing more sophisticated data analytics and financial services platforms is to advance feature engineering. Sustained development of these tools will go a long way towards reducing the rate of false positives. More advanced data analytics will not only aid fraud prevention but will lower the chance of alienating customers.

The role of AI in big data analytics

The prevalence of AI technology among B2B and B2C companies is astonishing as it takes the business world by storm.

Data analytics in the financial services industry: breaking the cycle of false positives in fraud prevention

Though data analytics solutions are well and truly embedded in the financial services industry, the rate of false positives remains a consistent challenge. Banks and other institutions are locked into spending astronomical amounts of money just to avoid falling victim to fraud.

Today’s solutions often make customers the first casualty of overbearing security measures. The anxiety over fraud has contributed to a “better safe than sorry” approach to defense, whereby institutions would rather block a legitimate transaction than risk an illegitimate transaction taking place. Consumer confidence is not only threatened if fraud is allowed to persist but also if normal transactions are blocked.

As data analytics in the financial services industry improve false positives will become less of a problem. We’re already starting to see forward-thinking data analytics providers producing smarter platforms with less false positives. In the future financial institutions will be able to enjoy greater protection with less chance of alienating loyal customers.

Related Topics

Big Data Analytics