A proof-of-concept using AI and natural language processing (NLP) found that more than half of all email enquiries could be resolved instead by guiding members towards digital channels.

In the PoC, analysts identified the root cause of each email enquiry, whether the query was resolved, and how many emails were exchanged in the process. Sentiment analysis from SAS helped Nationwide to detect the member’s mood; for instance, unsurprisingly, analysts identified that people’s moods worsen as the number of emails rises.

It established that over half of the emails could have been avoided entirely. Some 10% of members who sent a secure email said that their transaction had not been resolved first time around in another channel. Another 19% were not resolved the first time they sent a secure email, and a further 26% of requests concerned transactions that could be moved to an online digital process.

Nationwide is now using the information to make changes to its website designed to reduce the need for customers to contact it with enquiries in the first place.

Graeme Reed, senior manager, analytics, Nationwide, says: "Using SAS text analytics, we established several concrete ways to make our service better. For example, many members contact us for proof of the travel insurance gained with the FlexPlus current account. The documents are available on our website - if we make them easier to find, more people will access them on a self-service basis.

"We also found that members often ask for a paper statement, because our online statements did not contain sufficient personal information to function as proof of address. By tweaking the statement format, we can eliminate these requests. And if several members contact us about transactions that cannot be completed online, we can consider building a digital capability to meet their needs."