Lydia Chatbot
Spellz solution now integrated into Lydia’s Android and iOS mobile apps
200
Employees
12
Agents
Fintech
Industry
Support chatbot
+100
users requesting help through the Lydia application
+60%
find an answer with the Spellz chatbot
+30%
of questions generate a ticket with human intervention
+70%
of understanding
” We were looking for a powerful technical solution. We were able to integrate the chatbot into our application and create scenarios that could be easily modified by our business team. ”
Flore
Product Manager, Lydia
Who’s Lydia?
Lydia is a mobile payment application that offers a digital wallet. It enables users to manage all their bank accounts, make transfers, pay by SMS and email, all in a single mobile app.
The company raised €18 million in 2018 to expand its business, and has since experienced monthly growth of 15% in new users.
Why implementing a Chatbot
Victims of its own success, Lydia was not only gaining new users, but also receiving a growing number of tickets every day. As the startup had no desire to expand its support team any more, the team decided to automate responses to recurring requests. After comparing the prices and features of several platforms, they chose Spellz to automate their customer support.
Lydia’s team had a clear objective when they began their search for customer support automation: to respond efficiently and accurately to as many requests as possible.
The solution to their problem statement
The startup was looking for a solution to support its pace of growth, and opted to automate part of its support function as the answer. Lydia mainly deals with sensitive topics related to money and finance, requiring fast and accurate answers. Lydia’s team chose Spellz because of its straightforward usability, making it easy to master and facilitate the transition between teams. Spellz’ platform enables non-technical users to implement the chatbot and generate the support scenarios they need.
Once the project had been validated by Lydia, development of the chatbot began immediately. The aim is to guide users to the right scenario when they formulate their questions in natural language via the chatbot, then direct it with the help of buttons. In total, the building process required three weeks of full-time engineering with one project manager and one developer. Following the creation of the scenarios, the chatbot was integrated into Lydia’s mobile app via the Spellz API. The first version was released on IOS, followed by Android.
60%
Of users have their questions completely answered by the bot
35%
Of users get partial answers and are redirected to the help center
5%
Of queries are irrelevant or bounced (the user leaves)
A quick look at the results
Following the integration, there has been an increase in user requests without overloading the support team. Lydia receives the same number of tickets per month, although its user base continues to grow by 15% per month. They receive 15,000 support tickets per month, across all channels (Facebook, Twitter, e-mails, calls, etc.).
The chatbot filters 100% of requests before putting the user in touch with an agent from the support team if necessary. This process has reduced the time spent by an agent on problem resolution by 60%. Since the bot was implemented, the number of tickets has not increased.
The process of improving the bot is not over, as the chatbot learns and adapts every day to increase its ability to provide more answers over time.
In July, Lydia recorded 1 ticket for 60 users. After automating support, by September they had 1 ticket for 65 users.
600+
Minutes saved per day
3X
Less time to resolve tickets
200
Tickets avoided per day
Overall, inquiries with the chatbot have a resolution time of 3-4 minutes. The virtual help center also automates several operations that can be carried out in-app, so there’s a significant time saving compared to numerous email exchanges.
When the bot fails to respond to a request, it qualifies the query and forwards the data to the support team. The chatbot also features a suggested wording tool in the search bar when the user types in his request, enabling a better understanding of requests.
