The man who put the buck back into banking

Barclays’ switch to response-based marketing resulted in a new data platform that predicts customers’ needs. No easy task, says David Reed, who talks to the man who made it happen, head of CRM Richard Zanetti

phone

If, like Barclays head of CRM Richard Zanetti, you have spent a year working at the nuclear research facility CERN, the complexities of predictive modelling in the marketing world probably don’t look that daunting. On the other hand, projects in the commercial world tend to move rather more quickly than in the scientific community, requiring a different set of skills.

“What you find in the commercial world is that there is a wider variety of work and a faster turnaround,” says Zanetti. With a PhD in Physics, an academic path might have appealed, working on projects that could have an enormous impact – or end up fizzling out.

Instead, he moved into the commercial world and has managed to produce his own big bang, not once, but twice. For Barclays, he has doubled the sales generated by the response-based marketing team in the past two years through better targeting. Meanwhile at Sainsbury’s, he delivered a step-change in performance when sales increased by a factor of six, from £50m to over £300m, through the aggressive use of personalised promotions.

Little wonder that Zanetti was named Data Professional of the Year at the Data Strategy Awards in October. At the heart of that performance improvement for Barclays was the use of data to drive offers to customers based on predicted need, rather than product push. It is a concept much talked about in financial services, but one that is often hard to deliver.

“We have tried to dynamically link individual customers’ needs to products and identify the most important ones to tell them about. It has allowed us to be product-agnostic because we model it purely based on those needs,” explains Zanetti.

While the business has the flexibility to prioritise certain products to meet specific sales goals, it is customer insight that drives the messages an individual receives. Some of this insight will come from trigger events, such as a customer telling the bank they are moving house, while others will be modelled.

The crucial thing is that decision-making about marketing and budgets focuses on what’s right for the customer, as opposed to where the budget is – or who shouts loudest

Critically, the response-based marketing team that Zanetti works in has leveraged these models and triggers into reserved space in the bank’s statements. “That is very powerful because the statement is the one bit of material the customer is most likely to read,” he says.

Research by the bank had already shown that customers wanted a more customised approach – “show me you know me”. At the same time, the performance of direct mail was under pressure as costs were rising.

A suite of messages on statements, personalised letters or selective inserts has been used to execute the strategy. As a result, Barclays has reduced direct-mail activity – and saved £1.1m – while increasing cross-sales to customers.

Using customer data to drive messages can be quite exposing for a business, so data quality and integration programmes were undertaken first. The success of the approach has led to investment in a new production platform next year which will give the bank the ability to personalise in near real-time.

Zanetti notes that data and production are not the hardest things about this customer-centred approach. “The biggest challenge is building a library of marketing collateral. Somebody has to get that created and signed off,” he says.

Anticipating customer needs is nothing new to Barclays, which facilitated the needs-based strategy rather than one driven by product sales.

“The crucial thing is that decision-making about marketing and budgets focuses on what’s right for the customer, as opposed to where the budget is – or who shouts loudest,” he says. “We are product-agnostic.”

We’ve tried to link individual customers’ needs to products and identify the most important ones to tell them about

“We are looking for triggers both unpredictable, like every day occurrences, and predictable, like the end of a product cycle or end of a loan term,” he says. His scientific background has come into play through applying a logical filter to the lead selections that may then get made, such as targeting customers who have a competitor product.

This is particularly important to sustain belief in the leads his team generates for the front line. A call-centre agent needs to understand why a customer should be told about a particular offer, for example. Logical selections make the reasons clear and avoid the need for complicated debriefing of analyses.

Richard Zanetti

This is one way in which Barclays is very different to Zanetti’s previous role at Sainsbury’s. “In a supermarket, there is a limit to what the cashier can do,” he notes. Personalised coupons at the till were one of the initiatives he drove for the retailer. He was also able to transfer his knowledge of the high-volume Teradata analytical environment from one to the other.

“They have different types of data. In a supermarket, you have millions of transactions and rows of data which are used to send personalised offers, like money off. The bank also has lots of transactions, but not many purchases. The customer is also more engaged,” he says.
He believes that CRM is much more at the heart of financial services than in retail.

Pushing the tool down to branch level is the next goal for Barclays, allowing cashiers to be given prompts onscreen once they have identified the customer. Zanetti’s team writes those prompts, as well as the models and selections. “We position ourselves as there to help staff serve the customer better, but the prompts are not overly prescriptive. People don’t like to read from a script,” he says.

To the customer, the interaction should feel like they are being helped, rather than sold to. It is part of Barclay’s strategy, “making the hard things in life easier”. “We’re a big part of that. It is about talking to a customer at the right time, not just selling stuff, but having a real conversation,” says Zanetti. Services are another dimension to that, so if a cashier talks to a customer about an insurance product, they are able to complete the application process right there in the branch, rather than having to go into another channel.

Customer insight may have been a key resource for large organisations in recent years, but putting its findings into practice has been carried out less often. Barclays has shown what can be achieved if customer data is properly leveraged. Its results in the past couple of years compared with high street rivals suggest that putting customer needs over product sales does really pay off.

To do that requires the linking of back room analytics with frontline actions, something Zanetti is keen to ensure happens. He says/ “You have to get out there into the business. You can’t just sit in a bubble.”

CV

2006-present Head of CRM, Barclays Retail Bank
2006 Head of research and insight, Barclaycard
2000-2005 Senior manager, customer targeting, Sainsbury’s
1999-2000 Commercial marketing manager, 24-7 Home Shopping
1998-1999 Concept development manager, Somerfield
1994-1998 Associate, Mercer Management Consulting

Q&A

What are the major differences between working at Sainsbury’s and Barclays?
Retailers and banks are not that different – you see marketing people moving between financial services, retail and telecoms, too. Having said that, there are important differences in the products and what we are trying to do with frontline staff.

Who do you support in the business?
We support many functions – frontline branch staff, call centres and online. We supply leads for email, SMS, direct mail, direct communications – all the channels we operate in. We also feed into product teams and influence what they do. We also have links to Barclaycard – although it sits autonomously, we have connections with them.

What issues take up most of your time?
Half of my time is spent with stakeholders, like the product teams, understanding their problems and how we can help. The other half is spent with my own team talking about how to prove and deliver concepts.

What are the key measures for your role?
We get measured against all the usual things, but one interesting metric is assisted sales. That is about supporting the frontline with inbound prompts saying, “talk to the customer about this need”. It might be something service-related, rather than sales.
It is about allowing agents to have the right conversation with the right customer and is a big part of our role.
In branch, when a cashier logs on to the first screen, when they have identified the customer, they get a prompt driven by us. We identify the customer need, create the individual leads for inbound channels and do the scripting, too.

Is there senior-level buy-in to data at Barclays?
Absolutely, and it goes all the way to the top. That is one of the reasons for the success we have had. It also means I can’t baffle them with science because they understand it.

Where does Barclays sit on the maturity curve for its use of data in marketing?
We are probably about three-and-a-half out of five – we have made great progress and next year we are aiming to be at four. Nobody ever rates themselves as a five. One of the good things we have done is provide staff-level management information so people in branches can manage their own performance, rather than just being managed from the top. That is significant.

We don’t have those arguments with product teams about who has the most accurate set of profitability figures, because the business is customer-centric. It is less about how much money we made from an individual, it is about whether we met their needs.

my biggest challenge

We don’t have clickstream data in the data warehouse. There are projects in hand to get that. We can get some feeds, so we know what people buy online, but we don’t know what pages they visited and in what order. That information is held by a third party. So we have a blind spot or fuzzy view of the customer there that we are looking to resolve.

Our transactional database is good – it has to be to send out bills and statements – so we don’t have a problem there. Our problems tend to be around contact details. There is also the concern about appearing to know too much. That goes to the way we use selections and what we know about people. There is no point talking to people if they are not interested.