Exclusive Interview with Donald MacDonald, Head, Group Data Office, OCBC: “Good data is the lifeblood of good analytics and you cannot do good analytics without that foundation layer in place”

TDB: Please share with us the importance of OCBC becoming a data-driven organisation for all its stakeholders and the challenges faced across this journey.

Donald MacDonald, Head, Group Data Office, OCBC: We deal with a lot of data in the bank and many things we do in the bank is based on data and using it to make the right decisions. Which customers do we onboard into the bank? Who do we lend money to with what interest rate? Do we charge them? Which transactions do we approve? What offers do we give to a customer? Data is really critical for OCBC to compete, whether that’s helping the business managers make better decisions or embedding AI into the key processes.

Banks, in many ways, are largely undifferentiated. We see data as a lever that we can pull to differentiate us. For example, developing products that are the most relevant to our customers’ needs or using data to give them the best experience. Personalising the experience, the offers and the interaction they have when they deal with OCBC is important. Protecting customers from fraud and being able to detect scams in a timely manner is another key area where we can differentiate ourselves.

Likewise, using data to ensure that our employees have a good experience, where processes are streamlined and powered by data, is advantageous. Employees feel that they are working on something meaningful, and they can be productive, which adds significant value and plays a big role in differentiating OCBC from other peer banks. We want our employees to believe that we are more efficient because data is helping them perform their roles.

OCBC did not complete its data journey overnight. We have been on our data journey for over 20 years now. Banks are rather complex organisations with many different systems and integrating those systems into one central platform takes some time.

Furthermore, we are unlike digital banks with just one channel. We have internet, mobile banking, ATMs, branches, relationship managers, contact centre and chat bots, making that seven different touch points.

The other challenge is on the people side. Banks historically have not been seen as a place with advanced data analytics and were not perceived as a destination where the top data talent would want to go, compared to the Big Tech firms. It took some effort to reposition OCBC to attract the best talent, and now people know that we have a modern data stack coupled with an advanced team working on interesting use cases.

TDB: What steps has OCBC taken in ensuring the maintenance of high-quality data, which is both secure and accessible for generating deeper business insights, and form a basis for advanced analytics?

Donald MacDonald: Good data is the lifeblood of good analytics and you cannot do good analytics without that foundation layer in place. OCBC was the first bank in Asia to invest in analytics and we had our first data warehouse in place back in 1998. Today over 300 different systems feed into our centralised data platform, and many of those systems are coming in real time. We have more than 900,000 different columns that feed into our database and with access to over 40 billion transactions.

What it takes to get there, it takes time. To do the good analytics, we have what we call the ‘Triple-A’ framework. Data needs to be available, accessible and accurate before you can do any analysis. In terms of data availability, we have 300 systems that feed into the data lake on a daily basis today and we do not process it in an ad hoc manner. We have a structured data strategy, which we review each year to assess the gaps in the data platform. Within my team, I am responsible for owning that data strategy, working with the business units to understand what we think the most critical data gaps are. I can proactively decide to bring in that data, if I think that data is going to be useful in the future.

We are systematically filling all the gaps in the database that we think might be important. Once the data is available, we then make it accessible to employees. In addition, we have what we call pre-approved access policies. This is where we look at all of the systems and for each system, we see what each team in the bank might need to allow them to do good analysis. Basically, we have built a pre-approved access framework that allows people in OCBC to get access to the data that they need to do their job immediately.

So rather than going through lengthy approval processes, we can centrally provide the required data. For example, if you join the risk analytics team, on the very first day, you can apply for access to data and you are granted access to all the systems that are relevant for your job.

There is an enterprise data catalogue that sits on top of the data platform where people can just query the type of data sought. We make data discoverable as well, which I think is a key thing when you have 900,000 different columns in the database.

Finally, how do we ensure it’s accurate? Data governance is a huge focus area for a bank. We have a team which is focused on data governance. Every system has a data owner and we have automated data quality tracking and designated people responsible for resolving any data issue. That in a nutshell is how we effectively manage the data quality.

TDB: OCBC has recently announced the rollout of a Gen-AI-based chatbot for the Bank’s 30,000 employees. What key productivity gains are you expecting to realise through its deployment?

Donald MacDonald: OCBC GPT is just one pillar of our Generative AI strategy. We have over 10 other applications that are already deployed in production. What has made generative AI famous is ChatGPT and so your strategy has to have something comparable to ChatGPT.

OCBC GPT for us is effectively ChatGPT for employees. It is available within the secure OCBC environment. Fundamentally, we wanted to understand how we can harness the power of ChatGPT and bring it to an environment where we can manage what people are using it for to minimise the chance of data leakage. We also wanted to give them an interface where it is much easier for them to use it within their day-to-day job.

We started piloting OCBGPT back in May 2023, starting initially with 50 people and then expanding it as more people asked to get on board. By the time we finished the pilot we had over 1000 users. On average, employees were telling us that it was improving productivity by up to 50% in some of the job roles. It was being used for completing different tasks such as writing new job descriptions, responding to customer complaints, translating documents from Chinese to English, and summarising investment research reports etc.

Employees reported that it had saved them up to 50% of their time and based on this, we were able to go to the senior management highlighting its productivity potential. We also managed to govern the information risks and were able to roll it out to all employees across the region. We believe that these kinds of tools can transform the way that our employees work by taking away a lot of mundane, or ‘brain fatigue’ tasks. This enables employees to spend more time focusing on the things that really matter and areas where a human can truly add the most value.

TDB: What other use cases do you envision in leveraging Gen-AI across the Bank and how will this inform your data strategy going forward?

Donald MacDonald: We think generative AI is about helping our employees be more productive. It is about augmenting the employees so that we take away some of the mundane drudgery of work and they can focus on the higher value-adding tasks. The way we think about it is that we are building the tools at two layers. We are building a layer of what we call productivity assistants that help everyone in OCBC with the most common tasks like ideation, drafting content, translating content, turning the speech into text, and summarising documents etc. Those are capabilities that everyone in OCBC gets value from.

The second layer is on building role-specific co-pilots such as for the IT team, the contact centre team, the legal team and the compliance team, by looking at their specific processes and say, “what can I build for you that’s going to take away the specific drudge work that exists in your area.” An example of that is OCBC Wingman, the coding copilot we built for our IT colleagues. Basically, it plugs into our IT development environment and we can predict what the coders are doing and what they are trying to do next.

Impressively, Wingman automatically produces the code allowing our IT developers to be about 20% more effective when they are developing software. But it also has other use cases such as when I want to write documentation, when I want to do unit tests, when I want to debug the code, or when I want to refactor the code. All of that is automated with a simple click of a button, with Gen AI taking away those tasks.

The contact centre is another area where Gen AI can help. We are building a co-pilot for the contact centre agents where the copilot can listen to the calls and turn speech into text, while also summarising the call in 200 words and categorising it. Now we can automatically figure out the sentiment of the call and we can extract pertinent information from the call. Thereafter, we can share it with the customer relationship management team, relieving the contact centre agents of lower end data entry work and enabling the agent to pick up the next call faster.

We do similar things with compliance. Before, the compliance team would have to listen to every call and transcribe it to try to identify any mis-selling. Now, using generative AI, 100% of the calls can be scanned and AI can highlight the call requiring further investigation.

TDB: How is OCBC embedding responsibility and ethics within its Gen-AI framework?

Donald MacDonald: When we talk about AI, we also have to ensure that the AI is built in a responsible manner. OCBC contributes to relevant industry forums including working with MAS to build the MAS FEAT Principles based on fairness, explainability, accountability and transparency in AI. Whenever we build models, we make sure that they adhere to the FEAT Principles.

We were one of the key partners with MAS on the Veritas initiative to develop testing algorithms for model Fairness. OCBC open-sourced our own enhancements to the Veritas framework so that other people in the market could leverage our code and also augment their own capability. We have also collaborated with MAS on the project MindForge initiative, which is about ensuring Generative AI models are also deployed in a responsible manner. Being able to understand what these models are doing is critical if we want to build trust in these capabilities.

Naturally, it is still an emerging area and even though it is early days, on our side we have already built our own evaluation framework. We can test using a library of 500 OCBC-specific questions to check if the model is accurate. Importantly, we are building these tools ahead of the market standards being implemented, although we do expect that probably in the next 12 months the market will also coalesce around some real standards, and then we will adopt those as they become available.

Shopping Cart

Media Kit

    Data Protection

    Personal Data (“Data”) submitted for Media Kit (“Media Kit”), and/or collected in the form of first name, last name, email address and other contact details may be used for the purposes of inviting you to future events and for reaching out to you with content which may be of interest to you. For these purposes, The Digital Banker will share the Data with our associate companies (including event and content sponsors) to promote their products and services. You will also be automatically subscribed as a user on www.thedigitalbanker.com. If you would like to opt-out, email us at [email protected].

    By clicking Submit, you acknowledge that you consent/ have sufficient informed consent to the collection, use and disclosure of Data as set out above.

    The Digital Banker Summit

    Moving on from FTX: is 2023 the year of CBDCs?

    Indonesia, Jakarta

    Thailand, Bangkok

    Philippines, Manila

    Contact Us

      Data Protection

      The information you provide will be held on our database and may be used to keep you informed of our and our associate companies’ products and for selected third party mailings. Please tick the box if you would prefer not to be contacted for these purposes:

      Request Nomination Pack

        Data Protection

        The information you provide will be held on our database and may be used to keep you informed of our and our associate companies’ products and for selected third party mailings. Please tick the box if you would prefer not to be contacted for these purposes:

        Registration Form

          Data Protection

          The information you provide will be held on our database and may be used to keep you informed of our and our associate companies’ products and for selected third party mailings. Please tick the box if you would prefer not to be contacted for these purposes:

          Registration Form

            Data Protection

            The information you provide will be held on our database and may be used to keep you informed of our and our associate companies’ products and for selected third party mailings. Please tick the box if you would prefer not to be contacted for these purposes:

            Registration Form

              Data Protection

              The information you provide will be held on our database and may be used to keep you informed of our and our associate companies’ products and for selected third party mailings. Please tick the box if you would prefer not to be contacted for these purposes:

              Registration Form

                Data Protection

                The information you provide will be held on our database and may be used to keep you informed of our and our associate companies’ products and for selected third party mailings. Please tick the box if you would prefer not to be contacted for these purposes:

                Registration Form

                  Data Protection

                  The information you provide will be held on our database and may be used to keep you informed of our and our associate companies’ products and for selected third party mailings. Please tick the box if you would prefer not to be contacted for these purposes:

                  The world’s preeminent Private Banks and Wealth Managers are demonstrating a committed drive in innovation, advisory, new products and services to meet the sophisticated needs of their clients.

                  COVID-19
                  Amid economic activity revival on the back of the Covid-19 vaccine program, organisations moving from business continuity plans to stable working environments, together with the slightest improvement in unemployment numbers, forced the world to adjust to new realities. Coming to terms with the “new normal”, global investors are now on the look-out for attractive and stable investment opportunities.

                  Needs of Private Wealth customers and families worldwide have drastically changed due to the pandemic and banks have had to accelerate efforts to deploy a multi-channel service strategy and safeguard clients’ businesses and wealth against negative impacts of economic uncertainly.

                  The Global Private Banking Innovation Awards will recognise the world’s best private banks, wealth managers and asset managers that are championing innovation across advisory, service, products, customer experience and more.

                  Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. 

                  Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

                  Request Nomination Pack

                  Error: Contact form not found.