CIMB tests acceptance of chatbots in Malaysia

• CIMB makes online banking services accessible through a chat interface, providing a two-way messaging system
• In less than a year, CIMB Enhanced Virtual Assistan (EVA), a chatbot-based mobile banking app, generated around 130,000 downloads and 300,000 transactions
• New self-service capability aims to reduce cost-to-service and extend banking services through new channel

In an increasingly digitised and disrupted world, the cross-industry user experience is dramatically changing customer expectations from their financial institutions. The proliferation of social media and messaging apps, coupled with new technologies that pave the way for new channels of engagement and connectivity, have redefined customer demands when it comes to interaction and services. Technology disruptions like robotics-based chatbots, a blend of machine learning, artificial intelligence, and natural language processing are emerging as popular tools recently, aimed at enabling the personalised customer service capability.

In Malaysia, accessibility to smartphones has propelled the usage of online banking. According to Asian Banker Research, internet banking penetration stood at 69% at the end of 2016, and we expect that mobile banking penetration within the population will exceed 30% by the end of 2017. In line with the increasing use of digital channels in the country more banks adapt and monetise key features of social media and messaging platforms to gain competitive advantage in the digital space.

Facilitated by these new technologies, CIMB Bank, a leading banking group in Southeast Asia, launched its chatbot-based mobile banking app called CIMB Enhanced Virtual Assistant (EVA) in Malaysia in December 2016.

“We believe that the future of digital banking belongs to conversational user interfaces (UI) as the core, constructed with products and services built into or around it. CIMB EVA will enable us in delivering digital banking products and services to a much wider demographic of consumers in the Southeast Asian region in a truly revolutionary manner,” Samir Gupta, chief executive officer of CIMB's group consumer banking, explained.

Samir Gupta, Chief Executive Officer, Group Consumer Banking, CIMB (Copyright: The Star Online)

Challenges

More banks are experimenting with chatbots to provide 24x7 availability and improve personalised interaction. However, the features and capabilities vary as many chatbots only cater to basic inquiries while the "smarter" ones seek to assist customers in transactions, payments and advisory. For instance, American Express’s bot on Messenger for customers targets to provide real-time information about card benefits or services, Bank of America’s “Erica” aims to help customers with services including making payments, checking balances, finding educational videos, and eventually, even receiving one-to-one advice. Many other banks such as Wells Fargo, HDFC Bank and DBS Bank are implementing chatbots in hope of improving customer experience.

Though the application of chatbots across banking is rising, there exist challenges that they need to meet towards achieving the greater standards of ‘trust’, and accuracy, to provide secure services needed in financial services whilst developing the knowledge and communication skills at par with a human , especially for complex transactions. The process is evolutionary as most chatbots are still learning and growing.

CIMB saw the significant potential of mobile banking and messaging applications to optimise the bank's ability to reach a wider customer base, personalised experience, driving payment convenience and deepening the relationship with customers. The bank currently has 50% digital channel users among its total retail banking customers, where almost 70% are active users. As customers’ channel behaviours evolved, the bank increasingly realised the need to move beyond a “click-and touch-based” experience towards a "conversation-based" experience across multiple channels, akin to social media engagements.

CIMB EVA also sought to improve CIMB Bank’s efficiencies and reduce operating costs. The bank had a high reliance on telecommunication networks and significant messaging costs to provide real-time transaction alerts to its customers for security and compliance purposes. Through this project, the bank wanted to explore a new channel for fast and cost-efficient banking, which can also be used to communicate with and alert its customers.

Solution and functional features

Claimed to be the first banking chatbot in Southeast Asia, CIMB EVA was designed to make the bank’s online banking services accessible through a chat interface. This project provided a secure two-way messaging system that acts as a virtual advisor to help customers manage their daily banking needs. Its user interface was designed to be simple and intuitive, mirrors a typical chat app to reduce adoption barriers in navigations and transactions. This is a cross-channel platform implemented as an app which is initially scheduled to be available across branches, the internet, and via mobile. Built on the CIMB Clicks platform, EVA was initially launched as a minimum viable product, and additional functions are planned to be added to it.

Natural language processing technology has limitations and the chatbot capabilities need time to evolve to be able to process complex transactions and conversations. The chatbots, especially in banking not only need to "contextually" understand and respond to the question effectively without any error, but also need to have a personalised touch in the conversation. Most chatbots are still learning to handle “real life scenarios”.

CIMB implemented EVA as an initial experiment and is now gradually expanding its capabilities to include more features. As of now, EVA is able to answer general enquiries on account portfolio, track finances on-the-go, transfer funds, pay bills including for JomPAY (Malaysia’s national bill payment scheme), and reload mobile phone prepaid packages. Among these transactions, account enquiry remains the most popular feature, and fund transfers accounted for more than half of total transactions.

EVA can perform transactions through simple texts and chat, as compared to many other chatbots that are focused mainly on inquiries. It also comes with a messenger feature, allowing customers to receive notifications on their account activities, real-time alerts, as well as the latest offers from CIMB. The bank plans to add other features such as a comprehensive personal finance management function by the end of 2017.

Technology

The new system, which comes with transactional features, targeting seven million customers, was awarded the "Best Self Service Banking Project" during The Asian Banker Technology Innovation Awards 2017 and “Internet Banking Product of the Year” for Asia during The Asian Banker International Excellence in Retail Financial Services Awards Programme 2017.

Built in-house without an external vendor, CIMB EVA’s technical architecture is designed to modularise the message types and adopt a service-based architecture. Built on an open application programming interface (API) platform, the technology allowed the bank to enhance functional features and service capabilities by adding layers of technology and enabling easy integration with financial technology.

It is currently deployed in Malaysia and is planned to be implemented in other countries within the CIMB Group. There are certain limitations to its applicability across devices as only the bank’s online and mobile banking customers with devices operating on iOS 8.1 or Android 4.1 platforms and above can use this facility. The system works only in "online" mode and leverages existing internet and mobile banking security and anti-fraud measures. Internet banking login is required to access customer accounts and a one-time password (OTP) is needed for each open transaction. For fund transfers, the bank uses a two-factor authentication using short messaging system (SMS).

In a unique approach during the project development phase, the bank merged its technology and business teams to ensure that the solution being offered is customer-centric. The development process of CIMB EVA took about four months and was piloted across a closed group of internal users in August 2016. In the subsequent months, the bank sought regulatory approvals, trained its support teams, and developed marketing collaterals. EVA was formally launched to the public in December 2016.

Business impact

Less than a year since its launch, CIMB EVA generated about 130,000 downloads and around 300,000 transactions, yet to drive significant adoption among retail users. The platform still has to gain significant traction in Malaysia’s huge digital banking market as the bank alone has over 3 million digital banking customers. The chatbot would need dedicated resources that can redefine rules and shape its capabilities and performance over time to expand its application.

The bank believes that its current performance translates to higher customer engagement and loyalty. “CIMB EVA helps us respond to basic customer needs such as account balance enquiries and the ‘24/7’ mode of the system helped improve customer satisfaction substantially,” Gupta shared.

The adoption of CIMB EVA has increased customers’ current and savings account (CASA) balances with the bank. Customers who adopted both the CIMB EVA and CIMB Clicks digital platforms increased their balances by 14%, while customers with only CIMB Clicks increased their balances by 7% (Figure 1).

“EVA has the potential to improve the cost-to-serve for customers by providing a self-service platform for account balance enquiries, servicing accounts, new account opening, and identity authentication. With this, we can further improve service excellence and efficiency, in addition to our existing physical call centres and branches,” Gupta added. Additionally, through instant messaging, CIMB EVA has been able to directly push internet or mobile banking transaction alerts to customers, with estimated savings of around $930,000 (RM4 million) in 2017. The real-time alerts also have the potential to improve cross selling capabilities through targeted-based offers using analytics.

Since EVA's inception, the total number of active mobile banking users has increased by 6.3%. The mobile banking active user rate improved to 41% compared to 38% prior to the launch of the project.

CIMB believes EVA will become a more popular channel than CIMB Clicks, its internet banking arm, and combined with analytics, will result in better product offerings. In addition, it has helped banks reduce costs and relieve its workforce from basic tasks such as answering common or frequently asked questions.

Global comparison

Both banks and fintechs are racing to offer chatbots as part of their financial services. In transforming customer experience through this innovation, below are some other comparative cases from around the globe:

Bank of America: The bank unveiled Erica, an AI-based chatbot, to its customers in May 2016. As a digital assistant, Erica was designed to advise its users in terms of financial decisions and opportunities. Leveraging on artificial intelligence, predictive analytics and cognitive messaging, Erica is capable of directing individuals to look up their FICO score and check out educational videos and other content. It also enables customers to make payments, check balances, save money and pay down debt.

Santander UK: In December 2016, the bank opened to a select group of account holders the ability to enquire and transact using voice commands with its chatbot via the Santander SmartBank App. In February 2017, it released the voice biometric technology to its customers enabling them to conduct banking in a more seamless manner.

Nordea life and pensions: At the end of June 2017, the bank launched in Norway its AI-based chatbot Nova which has been trained to handle customer queries relating to pension and insurance in Norwegian. The bank also previously introduced virtual assistant Liv to life and pensions, which is claimed to have brought about 80% faster process lead times, a strong reduction in errors, and helped boost compliance.

OCBC Singapore: The bank developed a mortgage chatbot named Emma together with CogniCor – one of the start-up from the bank's fintech accelerator programme. Emma was built to converse with the bank’s customers and at the same time perform on-the-spot calculations for loan approvals. As of April 2017, the bank has converted more than 10% of the 20,000 chat session to mortgage loan prospects since its launch at the start of the year. Meanwhile, the bank just also introduced ‘Buddy’ to its employees as a human resource (HR) chatbot in May 2017.

Wells Fargo: In April 2017, the bank piloted an AI-based chatbot to communicate with its customers and guide them in making informed financial decisions. Back in May 2016, Wells Fargo already started using Facebook Messenger as a channel to answer common customer queries and issues, recognising that millions of its customers are using the platform regularly. The bank is yet to leverage on the chatbot to enable customers to do banking transactions.

Testing the potential

CIMB Bank’s new system facilitates customer engagement by responding to queries, simple transactions and real time messages. However, it is a young chatbot, still learning and testing capabilities to gain traction among digital banking users in the market.

Chatbot technology is still at a nascent stage of development, with various players experimenting with its functional and contextual communication capabilities. As its functions expand, chatbots could potentially bring cost efficiencies to banks. But it still remains to be seen how effective it will be in adding true value to the customer self service experience. It is also imperative for banks like CIMB Bank Malaysia to keep in mind that technology is only as effective and as secure as the team behind it. 

Note: This article was written by The Asian Banker in partnership with CIMB.