Techcombank wields data and AI to drive customer engagement

Vietnam’s Techcombank is on track to generate over $50m in incremental revenue this year from its data and analytics initiatives, thanks to its unified data platform that is reshaping how it engages with customers, according to a top company executive.

Speaking at a Databricks customer event in Singapore, Techcombank’s chief data and analytics officer, Santhosh Mahendira, detailed how the bank’s migration to the Databricks platform has enabled it to create highly personalised banking experiences and extend its reach to a wider customer base beyond affluent individuals and top enterprises.

“If you want to go down the value chain, the easy way is to increase the number of people, but that’s not sustainable,” Mahendira said. “And this is where technology comes in. We apply a formula called data, digital and talent, or DDT, which is core to our strategy.”

Key to that strategy was unifying data from nearly 150 systems onto a single platform. The bank chose Databricks for its data lakehouse architecture, which combines the capabilities of a data lake and a data warehouse, as well as for its artificial intelligence (AI) and machine learning (ML) tools.

“We were not looking for BI [business intelligence] automation or dashboards. We were looking for the AI and ML advantage which Databricks clearly has. Their AI and ML superiority over other providers stood out by a mile,” Mahendira said.

Mahendira said the Databricks platform has enabled Techcombank to develop its own proprietary AI solutions, such as its Lead Allocation Curated Engine (Lace). Instead of blindly assigning potential customers, Lace uses AI to score and explain each lead, telling the bank’s 4,000 relationship managers why a customer is a good fit and what their likely needs are.

“We have more data than what Salesforce and Adobe can use to build models for us. So, we will build the models ourselves, orchestrate them and send the results to those systems, which act more as front-end tools,” Mahendira noted, signalling the bank’s ambition to control its own AI destiny.

Besides deploying AI chatbots that serve different functions such as HR and customer service, Techcombank is also leveraging AI for hyper-personalisation at scale. Using a “customer brain” that draws on over 12,000 data features ranging from transactions and demographics to family ties and community scores, it can deliver uniquely tailored offers to its 15 million customers.

For example, an office worker who frequently uses his credit card might receive a personalised loan instalment offer via a push notification at 7pm, when he is likely to be reviewing his finances. The next morning at 7am, as he heads out for his daily coffee, he may receive an in-app message with a Starbucks discount.

This customer-centric approach has led to a 30-fold increase in the number of marketing campaigns the bank can run, moving from generic mass-market messages to highly individualised, timely interactions.

The results, Mahendira said, are measured in financial terms. “I have a P&L [profit-and-loss] target every year. I need to generate incremental revenue,” he said. “Last year, I generated VND 914bn, and this year I’ll generate close to VND 1.5tn, which is around US$50m to US$60m.”

“We have a relentless focus on value creation,” he said. “Just implementing something and saying it’s live in production isn’t good enough. Is it working? How does it create value? How much do we make? That’s very important.”

Mahendira concluded by stressing that technology alone is not enough. Techcombank has trained over 9,000 employees in data and AI, from tellers to the CEO. “You can build all these data and AI workflows,” he said, “but if you don’t upskill and change business processes, it’s not going to help.”