Table of content
Serasa, one of the main references in financial analysis and information in Brazil, offers in its Help Center the possibility of 24/7 assistance for several important services, such as Score consultation, debt negotiation, registration update, access and information security. As it is an important service for the financial health of Brazilians, the number of assistance and support requests is proportional to this demand.
Bel, Serasa's virtual assistant, was born with the implementation of their conversation bot, an even more efficient tool. With planning processes, continuous analysis and information feedback, it was possible to reduce the number of tickets created and increase the quality of automated assistance.
Lígia Braga, Head of Product-Led Sales at Aivo, talks with Lucas Silva, Customer Experience Coordinator at Serasa, and Daiane Oliveira, UX Writer at Serasa, about the strategy and details of this initiative.
Always focusing on the consumer - Serasa Platform
Ligia Braga: What was the focus of the strategy to improve the customer experience this year?
Lucas Silva: We believe that the consumer is the center of the business and the chatbot was one of our actions to improve the experience in 2021, as we realize that it is becoming more and more demanding and we need to invest in this area. We want to offer more channels so that you can always talk to us and always improve that relationship, with less friction and offering a satisfying journey.
Your bot's interactions with customers increased exponentially in 2021. How does the bot help you manage growing demand?
Daiane Oliveira: Bel, our character, is in our Help Center, a contact channel for consumers to ask questions and get to know the products better, to talk in a different way and speed up the resolution. Sometimes the person seeks a question which is explained in a long article, so they ask Bel and get a straight answer. Even when the doubt is about the Score, it is not necessary to read the complete content in the Help Center because Bel indicates where this information is.
This technology support was also important to reduce the volume of open tickets by around 30%, a very expressive number for our service. This result demonstrates that people are already able to solve their doubts more autonomously with the presence of Bel, it is not necessary to always talk to a human, gaining agility in the process.
Lucas Silva: Serasa's Help Center has a monthly average of 3 million visits, this is the audience that Bel handles today. And from these visits, we receive around 10 thousand tickets created per day, and Bel's retention is around 30% of the volume. This achievement in recent months is the result of a construction process, first "in house", thinking and planning how we wanted it to be and how this chatbot would be, and then seeking Aivo for this joint development.
Our goal is to bring all possible improvements to the consumer. Customer Experience (CX) is the consumer at the heart of the business, it's no use developing a strategy inside a bubble and the consumer needing to look for us. The intention is always to take the channels to where the consumer is making his journey.
You might also like: Discover all the featured cases of Aivo Awards 2021!
Smart and efficient service to support high demand
What is the biggest difference in your service before and after the entry of Aivo's Conversational AI solution?
Daiane Oliveira: As we updated Bel for the new version of Aivo’s solution, we went through processes of studies, tests, and structuring. And a difference from the previous Bel, in addition to language adjustments and optimized text volume, we also included a guided flow for when the person starts the service, a hybrid system with the AI that understands the natural language.
This was important because at times, for example, the consumer would ask about the weather forecast — today we have amusing answers to these types of questions. Then we've put together a list of the main topics she talks about, you can select them with buttons or directly type what you need. And these ramifications exist according to the types of products and services, which helps direct those who have doubts to receive the appropriate solution.
We've also included frequently asked questions in our script, managing to thoroughly analyze the results and key user interactions. We use triggers for specific events, like our Clean Fair Name, which is working well to highlight this intent in the bot. These are changes that make it more dynamic and easier to talk to on a daily basis.
Lucas Silva: We understand that people want solutions, so we already have the conversation part to answer questions and, soon, we will also include product transactional services to deliver more and more to the consumer.
To make the customer satisfied in the financial sector, what would be a good practice applied at Serasa suitable for companies that want to develop strategies with chatbots?
Lucas Silva: For starters, technology goes hand in hand with good customer service. It takes good tools and strategies to use them in the best way. The second point is continuous improvement: tracking KPIs, processes, thinking about chatbots, content feedback, reviewing, evaluating NPS, taking advantage of insights, and always in a short amount of time.
Related Article: The challenge developing content for a virtual assistant
It's not just thinking about long-term plans and actions, because in the meantime the consumer can give up. So, the idea is to invest in technology, continuous improvement and agility in the next year
Our area is called Customer Care because we believe a lot in consumer care and that is why we always invest in these strategies to effectively achieve this result.
Daiane Oliveira: Launching a chatbot is not simple, we need curation and evaluation on a daily basis. If not, the results will not be effective. When we add an intention, we alter the cognitive knowledge of the AI, we joke that we put it in “time out” and explain that maybe what it's responding to isn't quite what the person is looking for.
Recently, we put our intentions about Feirão Limpa Nome, but we also have a product called Limpa Nome, which left Bel a little confused. So we went to “time out” and differentiated which were the moments when each group of responses was compatible. But, if we didn't verify this, people would still be receiving information about Feirão today, even when the question was about the product. So it's not just a matter of including lots of intentions, but being careful with everything that's done.
The Training area of the Aivo platform gives us a lot of insights to know what’s going on in the Help Center, with some questions that weren't mapped, for example. Some questions, as we have a high volume of queries, end up not being relevant to our work – there are 3 million questions per month! – and others are very interesting, and may even be directed to existing intentions, but with another way of asking. Continuous and daily curation is key to having a resolutive chatbot, one of our greatest wishes with Bel.
Want to achieve these results?
If you haven't already, you can check out the full list of featured clients in Aivo Awards and discover the story behind their conversational bots.
Or, if you'd like to implement our technology in your customer service strategy, please don't hesitate to schedule a call with one of our representatives.
.svg)
