Tuesday, September 26, 2023

A Grassroots Initiative to Bridge Practice, Education, and Research.

Commonwealth Bank: Amplifying Customer Centricity with AI

Steven Randazzo,
Warwick Business School and Harvard University

Jin H. Paik,
Harvard Business School and Altruistic

Yael Grushka-Cockayne,
Darden School of Business, University of Virginia

Steven Randazzo, Jin H. Paik, and Yael Grushka-Cockayne describe how Commonwealth Bank used data and AI to maintain a competitive edge by bridging gaps between retail, call center, and digital services. By making data more available and standardized, the bank improved customization and enhanced its interactions with customers.

Success in retail banking hinges on retaining customers in order to capture their lifetime value. Until now, many banks have succeeded in attracting new customers through a one-size-fitsall marketing approach. The efficacy of this traditional approach has led retail banks to also apply it to retaining customers, investing marketing dollars in both traditional and digital channels.

Like other banks, Commonwealth Bank, or CommBank, Australia’s largest retail bank, with 15 million customers, took this approach. “It was a fairly heavy sales-oriented culture, with a largely homogenous approach to customer interactions,” explained Matt Malady, general manager for channel and customer at CommBank.

But with a changing customer landscape, CommBank decided to rethink its approach rather than simply funneling more resources into the existing strategy, especially since it had access to large amounts of data. “In the future, our customers will increasingly be able to serve themselves rather than coming to us, so we have got to find a way to be more relevant and personalized in meeting their needs than we ever have been before,” said Angus Sullivan, group executive of retail banking.

The leaders of CommBank decided to shift to an artificial intelligence- powered approach to creating new value. This approach would use data to augment the abilities of employees, transforming every frontline worker into a customer expert, saying the right thing, at the right time, to every customer.

The CommBank Approach

To fundamentally shift its approach to customer engagement, CommBank tapped into the data and business expertise of the organization, using existing components to create a new framework.

But the new approach did not match the existing technical and operating structures of CommBank. Its three primary channels – branch, call center, and digital – were poorly coordinated, which led to duplication and confused messaging and supported the one-size-fits-all approach.

Yet these three channels included more than 1,000 branches and 48,000 employees, 2,000 of them in customer contact centers, making and receiving calls. CommBank’s digital footprint consisted of its website and popular mobile app.

AI-based digital transformation meant changes to infrastructure, workforce, and organizational structure.

AI-based digital transformation meant changes to infrastructure, workforce, and organizational structure. And cooking up the secret sauce of creating value was difficult. “How do we, over thousands of interactions, try to generate the same outcomes as from an in-depth, one-to-one conversation?” Sullivan asked.

To develop this new approach, CommBank tapped specialists who had knowledge of the different pieces that make up an AI system, as well as leaders and experts with architectural knowledge of the existing framework and what needed to be changed. This team was charged with linking all the components together into a new, coherent whole.

The result was the Customer Relationship Banking Program (CRB), an internal program intended to redesign the customer experience throughout the bank. Its flagship project was a customer engagement engine (CEE), an AI-driven customer- experience platform.

The Sweeping Impact of CEE

It is no hyperbole to say that the CEE transformed how CommBank employees interacted with customers. The CEE makes 35 million decisions per day, devoting less than 300 milliseconds to each. It provides full cross-channel integration using 450 machine learning models that learn from a total of 157 billion data points.

For example, when customers go into a branch, the CEE suggests options for a next best conversation (NBC) to the workers who greet them, ranking and scoring possible messages. The staff can then use these conversation starters to improve the customers’ experience.

The system first assesses whether a customer should receive a specific message. Then it evaluates the NBCs, determining which are appropriate to the context. Finally it presents its selections to the frontline staff as options. Employees then use their judgment to select an NBC. After the interaction, employees record feedback on how the customer responded to the NBC. The bank has applied this process across all channels, and customer behavior determines which NBCs are most relevant. (See Figure 1)

Figure 1. Sample Next Best Conversations (NBCs).

The success of an NBC is measured by the NBC acceptance rate, that is, what the customer does after being presented with an NBC. For in-person interactions, tellers note in the system how the NBC was received; online, the NBC acceptance rate is measured by what the user does next. CommBank uses this acceptance rate to modify its NBCs to better fit customers’ needs.

In one example of the value an NBC can have, a teller was prompted to wish a customer a happy 90th birthday. The customer was delighted. “How did you know it was my birthday? Since I live alone, you are the first person to wish me a happy birthday.”

The CEE and Customer Engagement

The CEE eliminates the ambiguity of manual assessments as well as the expertise necessary to overcome it. Applying principles of behavioral economics, the bank sent various NBCs to customers to test whether the CEE could help them make better decisions about using their tax refunds than previous campaigns. It achieved a fourfold increase in NBC acceptance over years past.

In another experiment, Comm Bank offered a ‘fuel finder’ to 250,000 New South Wales customers, providing current information on the cheapest fuel in their area as well as a price map. Customers could choose to receive NBC alerts to help them save on their preferred fuel type. The system sent messages by using past fuel spending patterns to predict when the specific customer would need to refuel their vehicle. Over two test months, the bank’s acceptance rate was ten times that of the standard message suite.

For the contact center, the CEE identifies customers who regularly call and sends them messages offering more convenient options. While on the phone, staff members receive a push notification helping them to direct the customer into chat messaging. More than 40 percent of call center interactions are now performed via messaging, including those performed in the bank’s app.

The CEE allows CommBank to make best use of its data, identify customers instantly, and offer immediate support that matches their needs. The CEE’s groundbreaking ability to improve the performance of frontline staff, making everyone an expert to improve the experience of customers, puts it far ahead of the systems of comparable banks.

The CEE also raised the bank’s annual net promoter score (NPS). In the 2020-2021 period, the home lending teams saw a tenfold increase in leads that were 300 percent better than those of previous years. This improvement resulted in more conversions and an increase in the NPS score among mortgage customers by 16.4 points, to a score of 10.4. Improvements in NPS also spilled over to CommBank’s digital channels, with 1.2 million more customers logging in daily and more frequently, up from 32 to 34 times on average per month.

In 2019, the CEE delivered its 50 millionth NBC to a frontline employee, transforming their interaction with a customer. The bank’s leaders have determined three factors, which other companies can readily adopt, to be crucial to success.

  1. Explainability: CEE managers include a message on the NBC screen that explains why the suggested message is relevant to the customer and situation. Organizations can replicate CommBank’s approach by making it clear why users are receiving a particular prediction, which reduces the uncertainty often associated with AI ‘magic.’
  2. Training: CEE managers developed a training program for staff to prepare them for the drastic change in operations. Managers made more experienced employees advocates of the process, asking them to coach other team members. The goal was to integrate the messaging into everyday speech without sounding scripted.
    CEE managers also set up a feedback forum to give employees confidence in their modified duties while also allowing them to share their feedback about the approach and the NBCs. Managers used this feedback to improve NBC development, honing the suggested messages. Organizations should take staff feedback into account to ensure that their algorithms make correct predictions and to establish new pathways for data collection to further inform the AI.
  3. Content: The CEE team carefully considered which messages would resonate with customers. Should they start with sales messages that could generate revenue or with informative messages? The pilot program included some sales messages, but ultimately the team decided to favor service messages, such as notifying people of their anniversary with the bank, thanking them for their fidelity to the bank, and so forth.
    Customers responded well to this approach, which helped frontline staff to feel confident in the technology. As one CommBank manager put it, “It is an active way to show that we care.”

Since that 50 millionth message, CommBank’s use of CEE with customers has continued to evolve, moving beyond service and sales messages to new programs that help and support customers. The CEE can now connect customers with financial relief programs during difficult times like the COVID-19 pandemic, wildfires, and other disasters, as well as during economic shocks like persistent inflation.

CommBank continues to use the CEE as a core piece of its customer experience strategy. With the development of the CEE, the bank moved from a siloed, generalized approach to one that was data-driven and personalized, translating millions of customer data points and combining them with the expert knowledge of the frontline staff.

The success of this transformation was not a given; it hinged on key investments in infrastructure, talent, and a culture that embraced learning and AI. Human plus machine created an improved experience that focused on creating value for the customer by assessing, predicting, and responding to their needs while capturing value for the bank.

Author Bios

Steven Randazzo is a PhD student at Warwick Business School and a visiting research fellow at the Laboratory for Innovation Science at Harvard. Steven’s research focus is on the use of AI in firms and how AI is creating new value and capturing opportunities in complex knowledge work. Steven is also Head of Innovation for Altruistic, an AI and data science consultancy.

Jin H. Paik

Jin H. Paik is a research scientist at the Harvard Business School and a cofounder and managing partner of Altruistic, a data science consultancy. He was the head of labs at the Data, Digital, and Design Institute at Harvard, where he developed the institute’s strategic vision and directed project and research activities. He earned a bachelor’s degree from the University of Michigan and a master’s degree from Harvard and is a doctoral candidate at New York University.

Yael Grushka-Cockayne

Yael Grushka-Cockayne is a professor of Business Administration, the Altec Styslinger Foundation Bicentennial Chair in Business Administration and Senior Associate Dean for Professional Degree Programs at the University of Virginia, Darden School of Business. Her research and teaching focus on data science, forecasting, project management and behavioral decision making. She is published in numerous academic and professional journals and is a regular speaker at international conferences on decision analysis, project management, and management science.