Delivering a delightful customer experience (CX) is key for any company seeking to acquire and retain customers in the long run. Can artificial intelligence (AI) help with this?
At LangOps Universe, we hosted a panel to explore that question. Our very own Director of Product Marketing, Phillip Brougham, led a discussion with panelists from a diverse group of companies. These customer support experts included:
-
Rogério Correia, Head of ROI and Optimization, Farfetch
-
Christina Asante, Head of Complaints and Quality at GoCardless
-
Herbert Verschuren, VP of Services, Customer Contact, Air France KLM
Although the guests hail from three vastly different industries with varying customer needs, they share a common thread: All are leveraging AI and machine learning (ML) to both improve the customer journey and maximize the capabilities of existing staff.
Here are some of the takeaways from the discussion.
Fast growth has spurred the need for AI tools
GoCardless quickly grew to become the UK’s leading direct debit provider, and soon found customers beyond its shores in continental Europe. However, the company staffed predominantly English-speaking support agents. So when customers from France, Germany, or Spain contacted GoCardless about its products and services, support teams would reply with the help of Google Translate or they simply responded in English.
Hiring some native language agents helped GoCardless expand support across more languages, but the company struggled with retention because agents spent the bulk of their time fielding translation requests from co-workers, leaving them with little time to perform their core customer support responsibilities. This clearly wasn’t sustainable, so GoCardless turned to technology for assistance.
“We had to look into how we could implement language support automation,” said Asante.
Farfetch faced a similar problem as rapid expansion brought growing pains. The fashion e-commerce retailer has become a global force, but creating a smooth CX and generating customer loyalty when regional demand spikes (e.g., Ramadan in the Middle East and the holiday season in Europe and the US) has proved challenging.
Hiring local language specialists during spikes was not practical because of the time and money it takes to employ temporary contractors. The solution: Empowering English-speaking agents with Unbabel’s AI-based language translations.
“We retained some of the people in our English queue so that they could reply to inquiries from the Middle East using Unbabel,” said Rogério Correia. “We basically didn’t need to scale our team locally in the Middle East region.”
AI/automation helps teams do more with less
In a high-stress industry like air travel, creating positive customer engagement means hiring agents for their skills in customer interaction, not for being multilingual.
“You need a lot of knowledge and at the same time you cannot have that knowledge available in every market in every language,” said Herbert Verschuren.
To streamline customer support, the airline employs most of its agents in just a few locations. The experts are English speakers, but using AI-based language translation allows one support agent to serve all customer touchpoints across five European languages.
It’s similar at Farfetch, where AI and automation maximize existing resources by making it easier to forecast labor demand. Using AI-based translations, the company can make plans to distribute workloads without having to worry about language barriers getting in the way.
Now that teams are structured around support expertise rather than temporary language specialists, Farfetch can gauge team productivity and be more predictive about what teams might need to effectively handle seasonal spikes.
Advice for using AI: Test often and don’t automate everything
When considering AI initiatives, Correia advised companies to “think about automation as a kind of a bridge between a process that is flawed but can be better. And when you’re starting with automation, just start simple.”
Asante concurred about finding a balance between automation and human interaction, and not trying to automate all your use cases and workflows. “Understand what your pain points are and start there,” she said.
Asante also warned that business leaders may resist the use of AI and automation, so be prepared to show them metrics about customer behavior and customer insights. “Try to understand why business leaders don’t want to automate and use your customer data to show that it can actually work,” she said.
GoCardless examined customer satisfaction scores with tickets that used Unbabel translation and tickets that did not need any translation. They found there was either no difference between the two or Unbabel tickets actually scored higher. These metrics are the type that shows the power and positive impact of automated solutions.
Similarly, Air France KLM was skeptical about how AI can improve the customer experience. So it rigorously tested AI functionality for translations by having native speakers check Unbabel’s machine translations word for word.
“We wanted to convince ourselves about the quality of machine translations and would only continue when we knew it was good enough,” Verschuren said.
Verschuren recommended continuously testing AI technology to improve the user experience. But he’s also happy to report that Air France’s latest real-time tests using AI chatbots showed that Unbabel translations were “almost on the same level as native speakers.”
To learn more about how AI can improve the customer experience and watch other LangOps Universe sessions, check out the event on demand:
