Hello BBVA! CallMiner’s artificial intelligence helps companies to better understand their customers

The phone rings. Over the next few minutes, the adviser of Company X has several things to consider – understanding the customer’s wishes, providing the right response and ensuring that they are satisfied. Now, thanks to innovation, companies can have additional capabilities to help them meet those goals more successfully and efficiently. Technologies such as artificial intelligence (AI) and new strategies including collaborating with the innovation ecosystem can be used to get the most out of every call.

This has been demonstrated by a proof of concept (PoC) carried out by the Human & Physical Touchpoints team from BBVA Spain’s Enterprise Client Solutions department. Thanks to an alliance with CallMiner, a company that specialises in speech analytics, the potential of AI to study users’ telephone communications with the bank’s remote managers has been confirmed.  “Technology gave us the opportunity to find out everything that is happening in conversations from a customer experience perspective”, notes Daniel Ordaz, Global Head of Seamless Experience at BBVA and sponsor of the PoC. Unlike existing satisfaction surveys where you have to rely on the customer taking action (answering questions or filling out a form), this solution allows the content of dialogue to be analysed comprehensively, with the customer’s consent, and infers that evaluation without the need to contact them.

Internal and external cooperation

Some 7,000 calls received by 27 of BBVA’s approximately 2,000 Gestores Contigo (Managers With You) were chosen for the test. These managers are responsible for facilitating and streamlining customers’ relationships with the bank through direct, personal and individualised communication. They help customers to perform complex financial operations and carry out their day-to-day transactions, so that they don’t need to go to a branch.

Amaya Gorricho, Seamless Experience Manager at BBVA and product owner of the proof of concept, explains how this selection enabled data protection requirements to be met: “With incoming calls, we ensured that the customer was informed of the conversation being recorded, so that we could access and analyse it.”

From the outset, those responsible for the PoC worked side by side with other bank professionals and consulted in different areas to fine tune the test. On the one hand, the business team helped define the use cases they wanted to study, such as automatic categorisation of the reasons for those 7,000 calls being made, customer sentiment analysis, and the degree of final resolution. On the other hand, the engineering team helped to make the test a reality.

“The first option is always to do it in-house, but this solution would have needed a lot of time and too much effort”, explains Luis Mayorga, Global Tech Lead in Client Solutions Engineering Holding and also tech lead of this PoC. They therefore carried out “research work for several weeks” to determine which technology partners could support them along the way.

When the AI is all ears

The first of those chosen was CallMiner. Established in 2002, this US company offers solutions to analyse calls, chats or emails with customers using technologies such as machine learning, in order to improve business intelligence and the performance of customer service centres.

In total, this company has raised financing to the tune of $142 million (around €120 million), with Goldman Sachs among the investors. According to a 2018 Forrester report they are considered to be one of the most relevant companies in the area of speech analysis driven by artificial intelligence. For the proof of concept, CallMiner’s cloud platform specifically took care of speech to text, i.e. transforming calls into text format, and providing descriptive analyses of the resulting texts.

They needed one more piece to complete the puzzle. Consulting firm Accenture managed the relationship between CallMiner and BBVA, contributing both its advanced artificial intelligence models and its experience in similar projects applied to the Spanish financial sector. In this way, the strengths of both partners were complemented. “That level of specialisation and speed when generating a proof of concept could not have been achieved without them”, Mayorga stresses.

A first step in understanding and serving customers better

The PoC, undertaken over nine months, made it possible to obtain notable conclusions on the defined use cases. Eventually, if conversations could be analysed with a broader scope, the main reasons why a customer contacts BBVA could be identified. “This kind of information would be extremely useful, since it could help us to develop solutions to allow customers to help themselves, with all kinds of functionalities available on their phone or online before they have to make a call”, says Daniel Ordaz, sponsor of the PoC.

The results also indicate that shorter calls achieve a higher level of customer satisfaction, so making conversations more brief could be a useful recommendation to improve the service. The product owner of the PoC, Amaya Gorricho, also suggests that analysing the calls received by those managers who achieve higher satisfaction levels could help “to identify good practice and ways to focus the conversation that could be useful for other managers”.

Having demonstrated the feasibility of using technology to extract knowledge from calls, thanks to the effectiveness of CallMiner and Accenture models, BBVA is weighing up whether to run a pilot or project with a greater scope, and is carrying out actions aimed at improving the customer experience, optimising business results and implementing actions that promote operational excellence.

In the not too distant future, artificial intelligence should be capable of dissecting any customer interaction to better understand their wishes, give them the best solution and ensure that they are satisfied every time they pick up the phone.