Data

“We don’t want to be in a situation where the customer is ‘creeped out'”: How Barclays US is using AI

  • Barclays uses customer data to recommend products to customers and better understand customer sentiment
  • Rigorous beta testing is needed to ensure customer comfort with data gathering as a basis for personalized interactions
close

Email a Friend

“We don’t want to be in a situation where the customer is ‘creeped out'”: How Barclays US is using AI

The race to become the customer’s mobile digital banking ecosystem is getting tighter, and Barclays wants in.

Barclays — the bank behind popular co-branded cards including Uber, American Airlines, JetBlue and a host of other retail partners — is using customer data to suggest products and understand the root causes of customer complaints. It’s an approach large banks like Capital One and JPMorgan Chase as well as startups like Credit Karma and MoneyLion are using to push insights and recommendations that are most relevant to the customer’s spending behavior and product preferences.

The transatlantic consumer, corporate and investment bank’s U.S.-based business is also moving beyond credit cards, savings accounts and loans to launch a full-service digital bank later this year.

Tearsheet spoke with Mona Jantzi, managing director of strategic analytics and customer experience, to learn more about how Barclays is using data to create personalized experiences for customers. The answers below have been edited for length and clarity.

How do you use data to shape customer experiences?
We’re trying to leverage all the information about the customer so we can to make experiences as seamless and effortless as possible. With the Uber card, for example, we’re working closely with Uber to make use of appropriate information they may know about customers, and recommend the right type of offers to the right type of customer at the right points in time.

How does use of customer data go beyond product offers?
One of the technologies we’re using is unstructured machine learning technology that involves natural-language processing to do text analysis on complaints. By using natural-language processing, we get to know the underlying reasons for complaints. For example, we have some complaints about account closure. But when you look at what the customer is concerned about, something else that happened earlier hasn’t met their needs. This technique also allows us to do sentiment analysis of how distressed the customer is.

Do you use artificial intelligence to sift through the data? And what do you mean by AI, exactly?
AI is sort of a broad term that’s used for different [machine-learning] techniques. We’ve talked about unstructured machine learning, where you’re trying to categorize it [for example, using natural-language processing to analyze complaints] and structured machine learning, where we’re using data from customers and partners to determine what offer would be most interesting to a particular person.

Is there a danger about data quality and the data not providing an accurate picture of the customer’s behavior?
One of the most important things that we do is to make sure we are respecting the customer’s data and using it correctly. We have a very robust system of checks and balances in place to protect customer data and make sure there are no breaches. We also track outcomes to ensure the customer experience is exactly what we intended. We’re making sure that customers are actually able to go through a digital channel and achieve what they’re trying to achieve and that they’re in a good place financially, and that we’re not providing a credit limit that’s inappropriate for a particular customer.

Is data part of your strategy to scale across the U.S.? 
It’s a massive part of our strategy. We did a pilot last year with a mobile app called “My Personal Banker” where customers put in some of their information from other institutions to provide a full picture of their financial health.

With all this data gathering, how do you avoid creeping out the customer?
We do a lot of testing, we are very thoughtful about products we develop and ways we communicate with customers. We don’t want to be in a situation where the customer is “creeped out”. We try to be very sensitive to that. We do a lot of data analysis and co-create experiences and products with customers to get a sense whether something will resonate or not. It’s an interesting time. It becomes straddling between convenience and relevance for the customer and going too far, [the sense of which] is different for each customer.

0 comments on ““We don’t want to be in a situation where the customer is ‘creeped out'”: How Barclays US is using AI”

Announcement, Data

DataDay Conference 2022: All session videos

  • Tearsheet’s DataDay Conference focuses on financial data and the ecosystems currently being built by data aggregators, banks, payment firms, and fintechs.
  • Here are the full videos from the sessions we held on June 21, 2022.
Shabih Rao | June 24, 2022
Data, Sponsored

Cash flow underwriting: A BNPL solution

  • BNPL is undergoing a serious growth spurt. But with legacy data reporting systems galore, the new payments trend is putting lenders in a tough spot.
  • Using cash flow and cash data to determine credit worthiness could be a solution to the problem.
Nova Credit | June 01, 2022
Data, Sponsored

How we can stop using the term ‘financial inclusion’

  • Our current credit system has led to decades' worth of errors and financial exclusion.
  • But as consumers get more autonomy in their financial decision-making, lenders may have to finally change their ways.
Nova Credit | April 26, 2022
Data, Sponsored

Our credit system is broken — now how do we fix it?

  • Our current credit system has led to decades’ worth of errors and financial exclusion.
  • But as consumers get more autonomy in their financial decision making, lenders may have to finally change their ways.
Nova Credit | April 12, 2022
Data, Sponsored

How data is helping B2B fintech lenders step up to address loan application fraud

  • Alternative data leads to predictive models for fintechs -- risk modeling, fraud detection, and lead scoring.
  • In the third part of this article series, we discuss how alternative data is a solution against loan application fraud.
Explorium | April 04, 2022
More Articles