With the growth of cryptocurrencies in recent years, LiteBit, one of the top cryptocurrency brokers in Europe, has had to scale its customer service response strategy.
When the cryptocurrency industry saw record-breaking growth rates in 2017, more people started to click onto LiteBit's website, inquiring about the market.
The Netherlands-based broker helps customers buy, sell and store more than 45 different types of cryptocurrencies. With only two people in its customer service department, the company realized it would need more people to answer inquiries.
Although it had an integration with Zendesk, that wasn't enough.
Turning to Ada
This led LiteBit to partner with Ada, a customer service chatbot vendor based in Toronto.
Founded in 2016, Ada provides a chatbot-based platform that helps enterprises automate customer experience. The platform automates the interactions between brands and their stakeholders, said Ruth Zive, senior vice president of marketing at Ada.
"Ada is actually underpinned by machine learning models and natural language processing," Zive said. "We're able to recognize intent and really deliver AI-powered experiences."
She continued that the chatbot anticipates customer needs and interacts with them in a customized fashion.
Ada is a no-code platform, so users do not need to depend on their IT department. However, IT teams can go into Ada and tailor experiences and integrations to make it personalized for users, Zive said.
The platform's multilanguage functionality appealed to LiteBit, a European company that serves customers who use localized languages, said Bas Lucieer, director of data and automated customer experience at LiteBit.
LiteBit was also impressed by Ada's interface and simplicity. LiteBit needed a platform where customer service agents could create answers themselves, which Ada allows, Lucieer said.
The cryptocurrency broker first implemented Ada by putting the bot on its FAQ page. When people clicked on the bot, they were able to see the top five frequently asked questions and quick responses in short text format.
If those questions didn't fit the technical problems that they had, different flows could be created so that the customer's information could be gathered and sent directly to the Zendesk platform, Lucieer said.
"Instead of ping-ponging back and forth three times asking for more information, [we can now] get all the information directly from the customer [once]," Lucieer said. He added that this reduces the agent handling time significantly.
Ada enabled LiteBit's agents to go from full-time customer support to working part time on the Ada chatbot and part time on customer support.
While some customers still choose to call customer service, others are happy to receive their answers more quickly using the chatbot.
The training involved
It took LiteBit about three months to fully integrate Ada into its websites. One challenge the company faced was determining how to fit large articles into short texts for customers.
"You might think 'OK, I want a lot of text balloons coming up,' but you sort of overwhelm the customer," said Lucieer.
Customer service agents also had to learn how to tune machine learning to improve Ada. They were trained on how to flag keywords so customers receive the right answer.
LiteBit has shifted its focus from using Ada to help solve customer service questions to using it to proactively get messages to customers.
"If something is wrong, for example, with an order or someone needs to do a customer protocol or anything like that, we want to give a signal on the website," Lucieer said. He said the company now uses the platform to ask customers for more information and give real-time data streams on LiteBit's website.
Ada's new features
Recently, Ada introduced new features on its platform, including language-agnostic natural language understanding (NLU), languages that are read from right to left, omnichannel extensions and A/B testing.
The NLU features improve the understanding and intent recognition of multilingual customers. It helps LiteBit to recognize customers' questions better, Lucieer said.
"Ada is trying to use more machine learning to help out our agents," he continued. This helps LiteBit improve their answers to customers' questions. Ada gives agents suggestions about which answers are best for customers.
LiteBit also uses the A/B testing to track which answers fit which customer age group or location.
Ada, which raised funding in May at a $1.2 billion valuation, said it doesn't publicly disclose how much it charges users, but it tends to be a higher-cost platform compared to other chatbot tools in the market.
Other chatbot customer service platforms includes Zendesk, Kustomer and Intercom. Technology giants like Google and Amazon also have chatbot technology in their customer service offerings.