For years, companies have used text analysis to derive useful information about public and media sentiments, as well as customer experiences and interests, from text collections such as newspapers, publications and, most recently, social media platforms.

While the use of text analysis to inform ad algorithms of their target demographics is perhaps the most well-known application of textual analytics, it has many other uses as well. Your bank, like many others, is a repository not only of tangible assets, but also of intangible assets such as data.

You are very familiar with traditional data such as customer information, balance sheets and credit ratings, but you may not be as familiar with text analytics as it relates to the banking industry. In the world of banking and finance, text analysis can help you respond to dissatisfied customers, improve your customer experience and even aid in risk mitigation.

What is text analytics?

Simply put, text analytics converts large, unstructured data into useful information. You can mine data from records within your bank such as overdraft protection services, ATM usage, loans, and checking and savings accounts to determine which services current customers use most often.

You can also mine data from outside sources such as social media, news reports and product reviews to glean insight into intangible concepts such as public opinion regarding your bank or what services the general public deems valuable.

Likely, the data from different sources arrives in a variety of formats. Your loan repayments may be kept in a spreadsheet, but news reports are text documents. Social media is a separate animal altogether.

Not only does the raw data appear in a variety of formats, but it needs to be weighed prior to analysis. You can also gather similar text analytics regarding other financial institutions to see how your products, services and reputation fare against the competition.

From the first step of text identification through categorizing the text, analyzing it on a variety of levels and presenting it in a clear, understandable format, text analytics – done properly – can provide you with a detailed view of your bank, the direction it is taking and whether or not that direction is compatible with customer and market needs and wants.

How does text analytics work?

The basic steps are the same for text and data analytics in that you collect the data, validate the results, prepare the data, evaluate it, make the suggested change, and monitor and analyze the change. Both human and machine learning work together to achieve the desired result.

The greatest difference comes in understanding and weighing the text. You will need robust software and someone experienced in interpreting social media data to analyze the medium properly. For example, slang terms do not necessarily reflect their dictionary meanings, obscuring some data results.

Additionally, some social media account owners are overly critical of everything, skewing the data negatively. Once the text analytics is properly vetted and presented, your bank can use the information to track internal processes and manage its external reputation.

How can your bank use text analytics?

Today’s banks differ greatly from the stone and marble edifices of years past. In fact, many banks do not have branches at all, increasing the customer’s reliance on paperless statements, automated tellers and mobile banking. Just using your internal data provides a wealth of text for analysis.

Text analytics enables you to make data-driven decisions regarding your products and services, reinforce your risk mitigation, and remove bias from your offerings. Other applications could include issues such as segmentation, financial planning and virtually any other service currently monitored by hand.

Text analytics does not completely replace human interface in these activities. Rather, it enhances your bank’s ability to review a broader and less biased data set when making personnel, service offering and budgetary decisions.

Products and services

As a simple example, you could analyze the services that major banks provide to determine whether or not it would be cost-effective for your bank to implement those services as well. Another beneficial way banks can use text analysis is to enhance the experiences of their customers.

By monitoring what people on social media are saying about your bank, you can easily and efficiently create reactive fixes to customer experience issues. Moreover, using text analysis to track the success of past customer experiences and how you were able to mitigate the damage of negative experiences, your bank can create proactive improvements that fix issues before they impact your customers.

A more complex example from within your bank’s own document store is the effect of underwriter bias in loan documents. In a recent study, one bank’s Chief Strategy Officer analyzed underwriter bias and its impact on loan approval.

As the bank makes money on loans that are repaid, it has an inherent interest in the neutral evaluation of loan documents. However, the originator has an inherent interest in loan approval ratio, which is often at odds with the customer’s ability to afford the loan.

Cybersecurity

In 2017, Equifax experienced an infamous hack due to an inadequate cybersecurity infrastructure. This breach ultimately exposed the private information of 148 million Equifax customers. While Equifax has largely managed to recover from this incident – perhaps in no small part thanks to a variety of other high profile cybersecurity failures in 2018 – questions remain for many banks: “What if we’re the next Equifax? How will we manage the fallout? How will we know what steps to take next?”

Lest you think Equifax’s cybersecurity woes are completely behind them, the U.S. Senate recently completed its investigation of the Equifax data breach, noting that Equifax failed to implement strategies that would have protected its customers. Congress is now considering legislation requiring uniform cybersecurity standards and the notification of affected customers should such a breach occur.

To this end, text analysis can be a deceptively useful device. By consistently monitoring social media platforms and popular media outfits for text related to your bank’s name or brand, you can keep tabs on your bank’s public goodwill in good times and bad.

Text analysis can even be used to aid banks in fraud protection when used by cybersecurity experts to assess information on the dark web or other hacker conglomerates for security breakthroughs. Oddly enough, your bank may already have one of the more robust cybersecurity platforms at its fingertips – Twitter.

While not always as accurate as the National Vulnerability Database, Twitter’s algorithm rated the threat level of each vulnerability days earlier than the NVD, and achieved greater than 80 percent accuracy.

How should my bank implement text analytics?

Text analytics are not for the faint of heart. Even though 60 percent of financial institutions recently polled believed that using all of this data could be beneficial and make them more competitive in a tightening market, only 37 percent were equipped to do so internally.

With over 20 years of experience, PKSI‘s (Pegasus Knowledge Solutions) solid financial analytics solutions help your bank make the best use of its data and its social media data. Text analytics drive your smart banking decisions. Let PKSI provide your bank with a user-friendly GPS to guide those decisions.

Bonnya Mukherjee
Financial Analyst, Pegasus Knowledge Solutions