When executives at Norway’s largest financial services group DNB Bank met for their quarterly strategic meet up back in 2017, they all agreed that the importance of data-driven customer insights had grown significantly and that this should become a main pillar of their overall strategy.
In order to lead this new pillar, DNB Bank wanted to hire a chief data officer (CDO), and looked globally for the right candidate. They found it in Aidan Millar, who had previously looked after business intelligence and data management for Bank of the West, and was in charge of corporate strategy, change management and enterprise business intelligence at Central Pacific Bank.
The bank, which has offices in the UK, the US and Singapore wanted Millar to draw up a three-year roadmap to get the organisation where it wanted to be with data.
“We split this into four different tracks: cultural transformation, implementing processes, implementing technologies and then finally displaying and demonstrating the value we can get quickly from the business. Without credibility, it’s very hard to sustain what we’re doing so we’re showing the business quick wins with the value we’re getting from data,” he says.
All four of these tracks are running in parallel, with the organisation going live with a big data appliance in the cloud, and using data science tools from the likes of Amazon Web Services.
“We’ve invested a lot of time and effort in reskilling, and it’s a new area for most people to start thinking about. We don’t have a lot of data scientists but we have very capable tech individuals who are reskilling to use some of the data analytics skill sets we’ve deployed. We have 60 data scientists live on our big data appliance in the cloud,” Millar explains.
But the progressive use of data meant that DNB Bank had to first take a step back and consider data governance.
“Before you’re exploiting data, you really have to make sure you’re doing it the right way: to ensure you understand the content in your data landscape, where the data resides, how it’s used and how you should be allowed to use it, the quality of data and that’s where [data governance company] Collibra comes into play,” Millar says, adding that when it was evaluating the types of data governance tool sets, Collibra stood out as it provided both a systematic view of data across the bank, and was not overly technical.
The use of Collibra and other tools and processes helped the team to address all of the regulatory requirements for GDPR, as well as for anti-money laundering.
As well as AWS machine learning tools, the organisation is using DataRobot, which effectively cuts down the time it takes to develop data models.
“It’s a fascinating tool because data scientists spend a disproportionate amount of their time doing data modelling, and with this tool you can put in transaction data attributes from a specific set of customers and press a button to start and it consumes that data and builds a predictive model for you,” he says, adding that his team found that using these systematic machine learning tools can do things more effectively than data scientists.
“I try to rephrase it positively to data scientists that it is going to relieve you to do higher value added type of work; clearly they are not redundant as they have to evaluate those models and make sure they are reasonable and test this against their own human made models but it also releases them to think further up the value chain so rather than building the model they can actually take the output and think of how they can apply that to something else,” he states.
In terms of using data to add value to the business, Millar has several examples, including the ability to see how customers are progressing on their digital journey whether on a mobile or laptop. This means if they drop out of the customer journey at a certain stage, and there are many of these cases, the bank can check whether there is a design fault, or whether there is an issue in understanding the next step, and it can act by triggering a customer service agent to contact these customers. It is also using advanced machine learning models to better price products, and predict customer behaviours.
“We can predict pretty accurately whether a customer is likely to leave based on their average account balance over a 90 day period – there’s all kinds of ways to predict customer behaviours so you can add more value to what they’re seeking to do,” he says.