How RBS's 'data guy' built a bank-changing data lake

RBS's Christian Nelissen talks to Computing about how his 'Superstar DJ' data lake and analytics have become central to the bank's decision making

The Royal Bank of Scotland's data analytics team is headed up by Christian Nelissen - a man with one of the more unusual job titles in IT.

Officially known as "The Data Guy", Nelissen started at RBS in 2010 as the director of RBS's customer analytics and decisioning service, before stepping up to his new role in June 2014. That was also the time, he tells Computing in a conversation at Teradata's Partners 2015 conference in Anaheim, California, when RBS's big data journey really began to accelerate.

"The whole thing got tougher in the middle of last year when the bank had a look at its direction," explains Nelissen. "We had a new CEO, and we decided there were a number of things we wanted to pull out that we'd be really good at. Data analytics was one of those, raised out of a divisional silo where it had previously been."

The siloing had been a data lake, gradually building up over the four years that Nelissen had worked at RBS. It had built up because he'd followed his own theory on the way big data should be collected - en masse, and avoiding talk of use cases with the higher-ups.

Earlier in the day, during a presentation to Partners 2015 delegates, Nelissen said:

"When we build things we always talk about capability, not solutions. I find it difficult to have conversations about use cases for warehouses. Having access to your data is a general public good, and something you need to convince the organisation of early on. As soon as you have a use case, you're reducing functionality down to the use case," he said.

But has this approach worked, four years down the line?

Early on, Nelissen was advised to focus on particular use cases in order to provide some early returns on investment. "But what we wanted to do was just dump the data on the box, as doing that would let us come back and make this more profound data warehouse build. We couldn't have done that at the beginning."

The data is also being collected "for everyone at the bank" who can demonstrate a need - this is not just to charm or sway the board anymore.

"If you have need for data out of the core systems, you will come to my team and we will bring that feed for you. And, unless there's a particular reason not to, we're holding that feed for everybody," he says of the highly regulated pipeline that his division pulls "straight out of the central system and pushes to the EDI [electronic data interchange] in accordance with banking laws".

When asked exactly how much of the data RBS is now habitually throwing into its lake, Nelissen replies: "All of it. Because at the moment, everything that goes into the warehouse is used in one way or another. When we say there's use cases, we're [still] not trying to use for any specific purpose - it's more on-demand.

"But I guess the point I was making was that we've never put the warehouse in a box that said that. People ask me what the use case is, but the use case is everything.

"Indeed, I've always avoided having a conversation that puts the data warehouse in a box because that would make it too easy to dismiss the overall value of it."

The data warehouse, which is using a variety of what Nelissen calls Teradata "boxes", including the Teradata Value Analyzer product, has chalked up a number of success stories in the year-and-a-half Nelissen has been running it.

"We use the data to have intelligent conversations with the customer and to find things that are relevant and might help them," he says, hastening to add that, "the aim of the system is to find relevant points of conversation with the customer, not to sell."

He continues: "I routinely get into conversations about the way we use data, and when I say it's not about selling, people say, 'Okay, it's not about selling, but it really is about selling, isn't it?'.

"Yet we fundamentally don't start from that position. We put things into the system that cost us money, but are the right things to do for the customer, and then we put things in the system that are of no immediate benefit to us," he says.

While it's hard to see the exact benefit of wishing a customer a 'Happy Birthday' on the correct day of the year, it's also easy to see how making them feel valued can be a simple, effective way to carry out customer service.

As well as running the data analytics function and managing RBS's chief data officer ("who used to live in the technology world"), Nelissen is also in charge of "data-driven customer communications", of which the birthday message is just one example.

A couple more examples of what RBS "believes are the right things to do" include a feature in which RBS systems will automatically check whether a customer is paying for two similar services twice, and flag it.

"We'll say, ‘Hey, it looks like you have this service through your package account, and it looks like you're paying for it twice'," he says. But is that always a welcome service?

"We're always trying to use their data in ways that people would appreciate, but not be creepy," says Nelissen. "There are lots of things we've discarded because we thought people might see it as going too far. We like to say [privacy] is the number-one thing we worry up. I see things online and worry about how my data is being used."

But Nelissen insists that, while he is unable to discuss concrete, bottom-line-led examples of positive changes analytics has wrought, the benefits to how RBS deals with customers on a day-to-day level is stark.

"Being able to pull data together to give account-level customer profitability made a big difference in our strategy in the retail bank because, for the first time, it was a transition away from the pure sales focus towards understanding long-term relationships with customers: we could see for the first time what the relationship brings in terms of the customers and us."

Nelissen's approach means that analytics at RBS has become highly flexible and available to many different audiences within the organisation - it's just a question of packaging the data and "selling" it correctly.

"I think there are different audiences in in the organisation and they need different messages," says Nelissen. "So, for us, it's obviously about 'buy-in' at the most senior levels, [and for that] we're engaged in the usual internal comms type messages.

"But we also go out and talk to people about data. So one thing we can do, which I'm particularly proud of, is customise the website by the profile of who is visiting, and we have focused on making things that customers struggle with easier. So we go out to the front lines and ask our contact-centre staff to give us suggestions."

The workers who've reflected on customer feedback are then the ones who test out the changes on the website, in a move that should not just help improve employee satisfaction, but provide a genuinely valuable interaction outside the IT department.

"We have a video that shows a contact centre worker dragging and dropping an element, hitting a green button and then we say, ‘Terrific - you just made a change to the website'," enthuses Nelissen. "And that's a really powerful moment, because we just closed that loop."

But Nelissen is embarrassed to admit that the programme is called "Superstar DJs". "The idea is that the DJ is in control of his environment, and he's connecting with his audience," he explains.

"So you can imagine where this is going - the trip out to the front line is called Superstar DJs Live!"

The Data Guy's Superstar DJs Live?

"Yeah, that'd work with a Chemical Brothers soundtrack," laughs Nelissen.