Bots Can Boost Banks’ Efficiency — But Watch for Pitfalls

Robotic process automation offers many benefits for financial institutions, but you need to prepare before you jump in.

Robotic process automation (RPA) is quickly being adopted in the banking industry, trickling down from the biggest players to community banks and online-only startups whose customers may not have traditional accounts. Before you dive in, gain a better understanding of the many ways bots are being used in banking, the future potential, and the implementation considerations you shouldn’t overlook.

While technology disruption is nothing new, RPA advances are growing at a dynamic rate, enlarging the set of tasks that can be automated. In general, RPA is used to mimic repetitive, rules-based routine tasks and typically works best with structured data sources. Deeper behind the scenes, RPA may harmonize data among various digital systems.

Examples of RPA use in banking

One of the most visible uses of RPA in banking is interactive customer-service chatbots that assist with common requests (e.g., helping customers change passwords or access account information). Applications of RPA in banking range from these simple types of customer service interactions to multipronged anti-fraud investigations and many processes in between, such as file management, account updating, and account renewals for time deposits. Other common uses include scanning paper documents and entering their information into databases, and “scraping” websites for useful data.

Recent advances mean RPA can now collect the kind of big data that can transform a company’s understanding of its clients, customers and processes and is key to improving decision making and gaining insights into increased efficiencies. For example, RPA software can be a useful tool when it comes to compliance, due diligence and anti-money laundering requirements. Well-deployed RPA can also screen multiple lists and sites, such as those in the Office of Foreign Assets Control, to detect those on sanctions lists or even identifying politically exposed persons.

Humans versus bots?

Bots are faster than humans at tasks that are repetitive and deterministic — for example, gathering data from searches on individuals, accounts and transactions in various systems. They work around the clock, 365 days a year, and a properly designed and implemented bot can reduce the probability of typographical errors or overlooked results. They are also excellent at differentiating between normal and abnormal patterns of behavior.

As a bot gathers more data, it can apply the amassed information to new situations (i.e., machine learning). For example, it might note when a prospective client is in the same niche business as a politically exposed person. Automating these kinds of tasks frees analysts to review results, make judgment calls and monitor investigations. Federal Agencies Encouraging Banks to Innovate

More innovations will be coming. In December 2018, federal bank examiners urged banks to pursue new ways to meet BSA/AML obligations with the goal to “further strengthen the financial system against illicit financial activity.”

The joint statement from financial watchdog and oversight agencies (which include the Federal Reserve, the FDIC, FinCEN, the OCC and the NCUA) encourages banks to consider, evaluate and, where appropriate, responsibly implement innovative approaches aimed at combating money laundering, terrorist activity and fraud. The release notes that AI technologies banks are already using can both address BSA/AML compliance requirements and improve transaction monitoring. And, the statement noted that BSA/AML gaps exposed by pilot programs will not necessarily lead to supervisory action.

In addition to helping with BSA/AML compliance, RPA has become an important tool for banks that originate or refinance mortgages, according to a recent white paper by Infosys. When guided by the correct rules, RPA can help to streamline the mortgage process. The key, says Infosys, is in “identifying work processes that are repeatable, definable and rule-based.” Software can then be programed to complete such tasks. As mortgage lending often involves multiple players and processes, the Infosys paper notes, the ability of RPA to query across multiple data lists holds the promise of increased efficiency, as well as fraud detection.

RPA considerations for banks

While RPA holds much potential, as with any technology change, proper planning before implementation is crucial to help you avoid pitfalls.

• Examine your process. The rule of thumb is that RPA will magnify, automate and accelerate existing processes. A bank that already has good governance will amplify it through RPA. But robotic process automation may also highlight less-than-sound processes. Don’t just aim to automate the current process — bring together everyone involved to determine the most effective way to approach the process. Then, automate that.
• Guard against algorithmic bias. Testing for bias may require gathering data, building statistical models, and analyzing those models. The last thing you want is for an algorithm to unintentionally discriminate in its work. Be sure to document your efforts to guard against bias, as well as to establish proper governance.
• Align technology with business strategy. Today, most technology projects are also business projects, and RPA implementations are no exception. In addition to understanding the goal of the process you’re seeking to automate, it’s important to assess how the technology aligns with the bank’s overall business strategy and goals. Also, make sure senior management is committed to the plan.
• Test your automation. Start as small as possible with a pilot study to identify and comb out problems. Post-implementation testing of any automated process is key, and involves walking through a sample with the bot, making sure its results are as expected.
• Measure results. Comparing the time it takes for RPA to complete a task versus the time required for manual completion of the same task will help you start to quantify efficiency gains.

Realize the full potential of automation

As you may imagine, RPA holds as much potential for smaller financial institutions as for large ones. Although some smaller banks may not yet find it cost effective to adopt this technology, they will eventually benefit from emerging solutions that can help level the playing field in the financial services industry.

The use of artificial intelligence, machine learning and robotic process automation in the industry will undoubtedly continue to grow in the coming years. As the technology evolves and becomes more available, it will also become more sophisticated and enable banking professionals to focus their time on analysis, judgment, strategic thinking and innovation, rather than on routine processes — we’ll have bots for that.