Unicaja Banco: Increasing customer satisfaction with a virtual assistant

About Unicaja Banco

Unicaja Banco is the fifth largest bank in the Spanish financial system in terms of volume of assets. It is a listed institution, with headquarters in Malaga, and a leading bank in six Spanish regions with a proven track record of being financially solvent and sound. It operates under proximity banking approach with a solid business model based on the knowledge of the customer.

Industries: Financial Services
Location: Spain

Tell us your challenge. We're here to help.

Contact us

Unicaja Banco selected a range of Google Cloud AI solutions to build a virtual assistant that addresses customer queries faster while processing customer behavior data to make informed marketing strategies.

Google Cloud results

  • Accelerates time to production with pre-trained and custom ML models on Vertex AI
  • Secure by design, all data is automatically stored in an encrypted state
  • Multi-disciplinary and technical support delivers answers quickly to aid in the build-and-deploy lifecycle

From proof of concept to production in less than six months

Data shows that the future of customer service in the banking sector is undeniably digital, but striking the right balance between high-tech tools and human advice is what Unicaja Banco is working to achieve.

"We selected Google Cloud because it offers us the data capabilities we need to create applications that extract value from internal and external data in real time and in batch mode to improve the customer care. Additionally, we prioritize energy sustainability, savings, and efficiency, and want to work with companies that have similar values."

Moisés Riestra, Director of Artificial Intelligence at Unicaja Banco

When its Digital and Artificial Intelligence team wanted to build a virtual assistant solution to address customer queries faster, it turned to Google Cloud. "Our mission is to analyze online and offline information that helps us understand what customers' main concerns are, so we can address them in the most efficient way possible," explains Moisés Riestra, Director of Artificial Intelligence at Unicaja Banco. The solution accelerates the search for information needed to solve common issues for customers, reducing the number of calls to the call center and saving eight minutes on average per call.

To build it, the team looked for tools that could increase Unicaja Banco's data capabilities in an easy way, with high interoperability with its existing and future data architecture. "We selected Google Cloud because it offers us the data capabilities we need to create applications that extract value from internal and external data in real time and in batch mode to improve customer care. Additionally, we prioritize energy sustainability, savings, and efficiency, and want to work with companies that have similar values," says Riestra.

Building a comprehensive data pipeline to increase customer satisfaction

Unicaja Banco wants its virtual assistant to support customer interactions while also identifying customer behavior with advanced ML models to inform its marketing through different digital channels, including email and web. To build the virtual assistant, the Digital and AI team used a combination of Google Cloud services that, integrated, complete the full data lifecycle, from ingestion in different formats and across varied time periods, to storage and analysis powered by Google AI.

Dialogflow CX acts as the virtual agent that handles concurrent conversations with the bank's end-users. The natural language understanding module is sensitive to the nuances of human language and translates text to structured data so other apps and services can understand it, too. To train Dialogflow to handle expected conversation scenarios, Unicaja Banco uses Vertex AI tooling for pre-trained and custom models, accelerating time to production.

"Using Google Cloud we built and escalated our virtual assistant from proof of concept to production in less than six months. Without Google support, the deployment, configuration, and development of the virtual assistant could have taken double the time or more in the first release," says Riestra.

"The first KPIs already confirm that we're going in the right direction by working with Google Cloud to meet our business goals. We're modernizing and improving our interactions with customers with the ability to adapt to their changing needs."

Moisés Riestra, Director of Artificial Intelligence at Unicaja Banco

BigQuery is used as a data warehouse for analytics at scale and Google Kubernetes Engine supports the application development with CI/CD tooling that secures and speeds up each stage of the build-and-deploy lifecycle. Pub/Sub allows services to communicate with one another and is used for streaming analytics and data integration pipelines to ingest and distribute data. The resulting insights are then visualized in Looker Studio.

From July to October 2022, the bank leveraged this pipeline to analyze real-time data from more than 700,000 customers navigating its website home page. The virtual assistant also answered more than 21,000 queries from customers where Unicaja Banco was able to pre-qualify a purchase in 9.51% of the conversations.

"The first KPIs already confirm that we're going in the right direction by working with Google Cloud to meet our business goals. We're modernizing and improving our interactions with customers with the ability to adapt to their changing needs," says Riestra.

Shifting to the cloud with security in mind

To get to this stage, the Data and AI team fostered a cultural shift at Unicaja Banco, always considering the security of the processes and environments as one of the most important features when using the cloud. "Security is a key component of our solutions due to the nature of our business," explains Riestra. "But it was clear to us that Google Cloud offers a wide range of security services and components, so we can trust our data will be safe."

Unicaja Banco now uses Cloud Storage, where all data is automatically stored in an encrypted state. The bank also uses Secret Manager to store sensitive data. Because it offers a central place and single source of truth to manage, access, and audit secret information across Google Cloud, Secret Manager helps Unicaja Banco build more secure applications in a convenient way.

"We had to adapt from working with an on-premises architecture to the cloud, where we can leverage new kinds of services with security in mind. But we have full support from Google Cloud to solve all our doubts, providing great examples of real-life projects and demos that mitigate all the potential risks of adopting the cloud," says Riestra.

Current deployments rely on basic security components because the bank does not use Personal Identifiable Information (PII), and all data is encrypted in transit and at rest while in storage. But Unicaja Banco looks forward to leveraging more of what Google Cloud has to offer around security: "We have ambitions to increase the services and use cases deployed in Google Cloud, such as including Google Data Loss Prevention to encrypt and obfuscate data," he adds.

Looking forward to more customer-centric innovation

Unicaja's virtual assistant, Nica, uses artificial intelligence to facilitate digital operations in the mobile banking app, eliminating adoption barriers, helping and explaining to customers how to perform more than 100 operations within Digital Banking. Nica helps find information on available digital banking operations, offers personalized financial assistance, such as categorized income and expenses, and advice on how to improve your financial health. Since its launch, it has handled more than 295,000 conversations. The system architecture is hybrid, combining integrations with internal banking systems through REST APIs with the power of Google AI, including Dialogflow CX, both using traditional AI and new generative AI capabilities based on Gemini and BigQuery. In six months, over 1.5 million interactions have been recorded with Dialogflow, with over 25 conversational flows and detection of 110 customer intents. In addition, more functionalities are being migrated to generative AI, with the main objective of further improving the user experience. This integration and enhancement enables faster and more efficient access to banking services, simplifying financial management and providing personalized assistance in real time.

"We meet often with Google Cloud to discuss ongoing projects and learn about new features we could explore moving forward. The multi-disciplinary team really facilitates our work with answers to all our requirements in days, even hours. We'll continue to work together to bring more value to our customers."

Moisés Riestra, Director of Artificial Intelligence at Unicaja Banco

Tell us your challenge. We're here to help.

Contact us

About Unicaja Banco

Unicaja Banco is the fifth largest bank in the Spanish financial system in terms of volume of assets. It is a listed institution, with headquarters in Malaga, and a leading bank in six Spanish regions with a proven track record of being financially solvent and sound. It operates under proximity banking approach with a solid business model based on the knowledge of the customer.

Industries: Financial Services
Location: Spain