Working with Google Cloud, HSBC is automating the review of its contact center sales calls from a quality perspective, significantly reducing the time required to do so and enabling it to improve customer experience. The bank is also working on providing faster access to cloud resources, fostering innovation and product development.
Google Cloud Results
- Google Cloud results
- Significantly speeds up the time to review the quality of sales calls, helping to highlight any opportunities to improve the quality of conversations and compliance with customer needs.
- Helps to improve the overall customer call center experience and service by quickly feeding back opportunities to improve quality at an agent and center level.
- Reduces the time needed to run machine learning models from one week to one hour.
HSBC is a digital bank, with over 90% of its retail transactions globally and in Hong Kong being conducted through digital platforms.
However, while digital banking is increasingly popular, call centers remain an important service and business channel. The Hong Kong Contact Centre for HSBC, which celebrated its 30th anniversary in 2020, serves over 1.5 million customers through over 20 million annual interactions.
- Customers continue to appreciate the “human touch” of call center services, especially when they seek assistance.
- But handling high call volumes efficiently and effectively is challenging, particularly in a market like Hong Kong where many people speak a mix of Cantonese and English (often incorporating literal translations) that requires a high degree of experience to handle.
- Quality assurance for this style of speech tends to be a particularly cumbersome and manual process.
This was one of the challenges the HSBC’s contact center teams faced, and a particularly challenging AI problem, because there were no existing open source data or machine learning models to work with.
Access HSBC’s Intelligence Hub with Google Cloud
- The HSBC Intelligence Hub is a team of data scientists, engineers, and architects dedicated to migrating HSBC data and analytics workflows onto Google Cloud and looking for ways to leverage the data using AI and machine learning.
- This team works closely with different departments and business units within HSBC to design and build end-to-end data and AI solutions.
“People in the team come from a broad range of disciplines and backgrounds, from mathematics, business, computer science, behavioral economics, and engineering. You don’t necessarily have to be an expert in statistics. The most important thing is for our people to stay curious and have a drive to learn new things,” says Richard Bates, Global Head of HSBC’s Intelligence Hub.
“Google is one of our strategic partners,” continues Richard. “We’re investing in machine learning and data capabilities with them because their cloud solution is approved to handle personal information securely and reliably.”
“As we continue to build a customer-centric culture, data empowers us to conduct a more scientific and quantifiable assessment of our services,” adds Yusuf Demiral, Head of Data and Analytics, Asia Pacific at HSBC. “We see AI as an opportunity to better understand customers’ needs, hear their voices, and alleviate the risks for customers and our bank.”
An Automatic Quality Management System
- Within the Intelligence Hub is the Data Science Squad, led by Mary Chu, Senior Analytic Architect, who took on the Contact Centre Language processing challenge.
“We turned to AutoML Natural Language and Speech-to-Text to train machine learning models to classify, extract, and detect customer sentiment,” says Mary. “Using Google Cloud computing resources and BigQuery as a data analytics warehouse, we applied Speech-to-Text powered by Google’s AI technologies to the data, which accurately converted spoken combinations of Cantonese and English.”
- The result was an Automatic Quality Management system (AQM), the first in-house voice-processing solution powered by AI that helps to identify areas of improvement in customer conversations, a key milestone for HSBC’s Hong Kong Call Centre.
Better Agent Service
The HSBC Intelligence Hub Data Science Squad completed this development in just three months, and, working closely with 10+ HSBC teams, launched the solution in September 2020.
“With Google Cloud, we could run a machine learning model in an hour, rather than a full week as would have been required in an on-premises environment, which considerably shortened our rollout time,” says Mary.
HSBC now processes through this solution every Cantonese-English call into the contact center that involves presentation of information on terms and conditions.
- The solution helps the institution quickly identify sales agents with room for coaching and improvement. The monitoring team automatically receives alerts to such calls, which enables them to follow up with agents in a timely manner.
- It also allows HSBC to capture more insights to sharpen the way the bank serves its customers. “These insights also become valuable resources for us when we design our future customer journeys or work on staff training and reinforcement.
Theoretically, it will involve 1,200 man-hours to monitor 100% of our sales calls. It is now unnecessary to spend this amount of time for monitoring,” says Richard.
HSBC plans to extend AQM to other calls within Hong Kong and eventually, globally. “We want to cover service calls as well, so we can track customer sentiment and their experience of the service,” says Mary. “In addition, we expect to cover languages such as Mandarin Chinese, English (including accented versions), and Spanish.”
Streamlining ATM Cash Inventory
Over the last six months, the HSBC Intelligence Hub has used its Data Science Workbench to develop capabilities in addition to AQMs. For example, HSBC has launched an internal tool, known as iCash, that predicts the stock of cash in ATMs and enables optimization to ensure they never run out of cash. iCash also provides strategic solutions for cash forecasting and planning for further development and rollout.