BUSINESS TECHNOLOGY MEDIA COMPANY
Companies
Sectors
Features

Turning data into rands

In an increasingly competitive environment, a digital strategy can’t be forged without strong data foundations.
Read time 9min 00sec
Simon Marland, Nedbank.
Simon Marland, Nedbank.

Despite having been at Nedbank for nearly two decades, Simon Marland doesn’t come across as a traditional ‘banker’. In our conversation about the future of banking, he talks of an in-house machine learning model called ‘Adam and Eve’, a robot called Pepper, virtual reality, and Facebook being his biggest competitor.

Marland is Nedbank’s chief data commercialisation officer. “Data is an asset, and it’s my job to turn it into rands and cents. We're commercially pushing out some data products, and beginning to make money out of those, obviously balanced between the regulatory stuff and how we commercialise data.”

As South Africa’s banks start to digitally transform, Marland’s role will become crucial. To give some indication of how important the bank’s focus on growing digital and data is, he highlights that two years ago, there was one person within the UI/UX function; today, there are 100 people.

Perhaps the most banker-esque thing he says during our interview is: “Everything we do, we try to make money out of.”

A good example, he adds, is the use of robotic process automation (RPA).

“We've actually had automation for 10 to 15 years; we've just upped our game a bit with it in the past two years. Before, we had simple RPA, screen scrapes and copies of macros, and we used to have a room of about 80 computers, where you'd physically have to go to each computer to start it up. But, in the last two years, we've got additional software, including Blue Prism and UIPath, and we've virtualised things so that you can sit on one terminal and have a few hundred robots running at the same time. To date, the most robots we've had running at any one time is 265, but we seem to have a constant flow of about 80.”

By merging AI with robotics, the bank is looking to improve automation so that parts of its customer journeys don’t need to rely on humans, and making processes smoother and faster. An overall evaluation of the system and the processes means the AI can determine which robot and process to call to suit a particular challenge.

Marland claims that Nedbank is probably one of the only banks in the world that actually makes money from robots. “Worldwide, I haven't seen anybody else other than Nedbank make money from robotics. Everyone has intent, but it's expensive, you have to put in the processes, you have to build everything out. It's not a 98% or 99% job. If you don't get it 100% right, it costs you money,” he says.

Adam and Eve in AI

Marland estimates that Nedbank has 35 machine learning models running currently. One project he seems quite proud of could be revolutionary in changing how banks, and indeed many companies, personalise communications with customers. It’s called Adam and Eve.

Adam is a machine learning element, which works out the propensity and what products clients should be consuming. “Every day, we're running machine learning algorithms to work out what our next best three conversations should be with clients; it’s not necessarily a product sales push, it could be about retention or moving them to the app.”

Then Eve is the delivery of that propensity model, he explains. A copywriting and visual medium creation tool, Eve determines how best to push that product in a way that’s engaging, likely to appeal and is unobtrusive.

“Adam is the data science, machine learning stuff, and Eve determines 'this image, in Facebook, is the best medium to reach this customer'. One knows that you need the product, and the other knows how best to push it to you.”

“Adam is the data science, machine learning stuff, and Eve determines 'this image, in Facebook is the best medium to reach this customer'. One knows that you need the product, and the other knows how best to push it to you.”

The bank is currently experimenting with data to picture (D2P) and picture to data (P2D) models. D2P involves the machine looking through a customer’s card swipes to try to build a profile of them, and then compiling an image from composite elements that will appeal to the customer. Marland admits the imaging capability isn’t quite there yet.

“It might see that you buy beer frequently and have just spent on some football tickets – so it compiles a unique picture of your team’s soccer stadium and a beer; then, if you’re on Facebook, where everyone is marketing to you, that image might resonate more than the others.”

The flipside of that is P2D, whereby a picture from a customer’s Facebook page is analysed, with the resulting data then added to the customer’s profile. “We augment the picture as though it was a piece of script, which reveals, for example, you’re at a soccer stadium, and then when we run the analytics. Instead of looking at just the text and swipes, we've effectively turned the pictures into text as well to get a better and deeper understanding of the customer.”

Personalisation

Marland says that as competition heightens within the retail banking sector, pricing will become more commoditised. He believes the key differentiator will then be understanding the customer better than competitors, and knowing which of the multiple engagement channels best suit them.

“Even in an email, talking to a customer in a language they prefer about relevant stuff, supported by relevant visuals, is important,” he says. “It’s not just understanding the customer, but also knowing what to offer them. In the end, it’s taking all the things we know, all the things we think you want and score you correctly for, and package it so that it suits you. A lot of that will be AI-driven.”

As retail banking becomes an increasingly competitive environment, with major new players launching last year, how does he view the landscape?

“There are three threats –piranhas, sharks and whales,” he says. “Piranhas are all the fintechs, which come in and eat little pieces of the value chain and you become disintermediated with the customer. Sharks are the new, large challenger banks. A whale is both a shark and a piranha, and would include the likes of Facebook, Apple, Google and Alibaba. They're diversifying, doing fintech stuff, maybe crypto, or they'll come into the value chain, or they’ll do full banking.

“From a data, AI and digital perspective, I should be looking at the whales to compare Nedbank’s capability and competency. Rather than looking at how our AI and machine learning stack up against Standard Bank or Barclays, it should be, how do we stack up against Facebook?

“I would see Facebook being the biggest competition, not the other banks.”

Cutting-edge tech

Other technologies that Nedbank is using as it tries to expand its digital prowess include virtual reality and physical robots.

VR is being used by Nedbank’s business banking unit to help potential clients or merchants visualise what an ATM might look like on location, as well as how it could work.

“The other thing we've done, which delivers a bit of a saving when Nedbank creates a new branch, is to convert the architects’ plans into VR, then the Nedbank team can go into the branch, as though it was there, change colours, settings and give it back to the architecture team who designed it. Before VR, there would have been multiple engagements with the architecture team, with anything up to 12 to 13 iterations. We're not necessarily making money out of it, but we're beginning to use it from an efficiency perspective and building a good understanding around where VR might go in the future.”

Then there’s Pepper, the cute little humanoid robot that can provide clients with all manner of basic information around Nedbank’s products and services.

“Even though Pepper is stock standard, we've rewired her a little, and we're using her more for presentations and to get people excited about Nedbank going digital. She does have a ‘total AI’ mode, so we could connect her to virtual robots; imagine we had an army of virtual robots that can open current accounts, run customer surveys and so on; all Pepper really is then is a UI.”

Contrasting the somewhat fantastical idea of a branch staffed by robots and Standard Bank’s announcement last year that it was closing branches and retrenching employees, due to customers transitioning to digital channels, what’s Nedbank view on the future of the branch in SA?

“From a financial literacy perspective, especially in the entry-level banking market, Nedbank needs to push there, and a lot of that will need to be face-to-face. Look at the segment beneath R100 000 per annum, a lot of their transactions are cash,” he says.

“As soon as you have cash societies, you need more interaction. What I would prefer is that people walk into a bank, get digitally activated and have a frictionless service. If clients are looking for the interaction, they should still go into branches. There's definitely still a need for bricks and mortar branches. When you add a new customer engagement channel, you tend not to lose the other channels; the number of transactions seem to go up, and maybe one or two of the other channels come down, but you still need those channels.”

From a digital channel perspective, Marland admits that the bank is experiencing a significant migration from web to the app. “Mobile is becoming where people do their sales and service transactions.”

And, of course, underlying all of these technological developments is data. “Data is the lifeblood to digital. The better you are at data, the more you can do with the digital offerings,” he concludes.

Login with