“Dashboards have gone from being a rear-view mirror to showing you what might happen tomorrow,” says Richard Spiegal, head of Nationwide building society’s business intelligence (BI) centre of excellence, as he reflects on the changing role of BI and the rise of predictive analytics for financial services. “They’d tell you what happened yesterday, and a human would pick it apart and try and figure out what comes next. Now we’re doing that automatically.”
Like many of its rivals, Nationwide, the UK’s largest building society with some 15 million members and assets of £254.9bn, is embracing the predictive power of digital systems to try and make life easier for its customers and staff. But this is not without its challenges, as Spiegal explained to Tech Monitor as he discussed the company’s approach to analytics and what the future might hold.
Nationwide’s business intelligence strategy
Spiegal joined Nationwide two years ago and runs a team of 30 people. “In simple terms we build operational dashboards all day long, working for teams across the entire business,” he says. “Anything from the finance function, to audit, propositions and everything in between. We work with anyone who wants to make a decision with data, basically.”
When building a new dashboard, Spiegal says his team employs the agile BI approach, which sees them work closely with the end user to ensure the finished product meets their needs. “If you try and go down the ‘big bang’ route – someone gives you a set of requirements and you spend months building it on your own – you generally get bad results,” he says. “Our approach is to get as close to the business teams as we can, so if they’re launching a new mortgage product, for example, we’ll go out to the mortgage team and really try and understand the product and develop the dashboards and reporting systems side-by-side with it.”
Nationwide uses two BI packages – Qlik and ThoughtSpot – to build tools for its staff, and Spiegal says low code functions, which allow non-technical teams to play a role in the development of software systems, are becoming increasingly important. “Low code allows us to iterate quickly and get the answers we need, so it’s a really important function in the tools we use,” he says. “As a BI team, we’re a team of 30 in an organisation of 18,000 staff, which is why we’ve adopted the centre of excellence approach. We give Qlik as a tool to people and let them build or adapt their own solutions.”
This means the role of Spiegal’s team is changing. “We are there to put the guardrails in,” he says. “The customer data we have is very sensitive, so we have to put in the governance around that. But the plus side is that they’re the people who know what the members want and what they need to achieve that.”
Spiegal adds that low code is allowing Nationwide to get analytics out “to every corner of the business,” and adds: “When you give people the tools you see their skill levels start to rise. They’re doing things that we would never have thought of and showing imagination and speed that it’s difficult to achieve with a central team.”
The role of predictive analytics at Nationwide
Spiegal is a software engineer by background, and since moving into data analysis has worked in the charity sector and as an analytics consultant. “I think it’s fair to say that, for a long time, analytics was a rich man’s game,” he says. “That’s not the case anymore, because tools have come along to democratise things, and financial services companies have been ahead of the curve in adopting these.”
Spiegal says Nationwide is starting to deploy predictive analytics. In the organisation’s anti-money laundering operations, for example, automated analysis is being used to predict which cases should be tackled with the highest priority. “Decisions will still come down to a human,” he says. “We don’t want that side of things to be done automatically. But the system might help guide the human through their caseload.”
He says getting the balance between human and automated processes is important to help Nationwide maintain its culture. “We’re a building society, not a bank, which means our members have to come first,” he says. “In the context of predictive analytics, we don’t want a situation for our members where the ‘computer says no’ in a blind way that no-one can explain.”
The biggest issue is trust. If we can’t explain how these models work and tell a good story then people instantly lose trust and they’re of no value.
IT teams have had mixed success when it comes to extolling the virtues of predictive analytics to staff, Spiegal says. “The biggest issue is trust,” he says. “My team is down in the trenches creating these models, but then we have to go out and explain what we’re doing to our colleagues. I think that’s something that has been hit and miss, not just at Nationwide but in the whole analytics arena. If we can’t explain how these models work and tell a good story then people instantly lose trust and they’re of no value.”
Spiegal says artificial intelligence systems are becoming better at explaining how they reach decisions, and adds that Nationwide has started experimenting with synthetic datasets in a bid to add an extra level of security for personal information. “When you’re doing analytics, it’s hard to do it without real data,” he says. “There are all sorts of issues like that, not just with privacy and security but from an engineering perspective too; moving real data about in the sort of volumes we use can be a real problem.”
“We have a programme within the analytics team at Nationwide which is building synthetic datasets which have no actual customer data in them at all,” he explains. “They’re fantastic for proving that these models work, and playing with new technologies like this can open up a lot of opportunities for us.”
The future of business intelligence for financial services
The evolution of predictive analytics is changing the way staff at financial institutions make decisions, Spiegal says. “The main BI tools which are out there now all have predictive elements baked in,” he explains. “So whereas previously you had a whole forecasting team looking at mortgage sales and making predictions for the future, now the tools do that automatically. This fairly benign use of predictive analytics speeds things up drastically and is huge for us on a day-to-day operational level.”
Again trust is the key factor for ensuring the successful adoption of these kinds of tools across the industry, Spiegal concludes. “Staff have to be able to know how we’ve made a forecast or come to a decision,” he says. “If we can be transparent and explain the process in plain English, we know people will use these tools.”
