Time to Shift From Traditional Customer Support to Multilingual Conversational AI

welocalize January 10, 2022

Multilingual Conversational AI

A lot has changed with global customer experience (CX) in the past few years. Especially since the pandemic started, there has been a rapid shift in consumer expectations. Customers consider their experience with a brand as important as the product itself. And this includes their experience with customer support.

This is the topic presented by Aaron Schliem, Director of AI-Enablement at Welocalize, at our Let’s Go On Demand Summit. His presentation, “Global CX and Support Through Two Lenses: Localization and Conversational AI,” showed the sharp contrast between the limitations of standard customer support and the capabilities of conversational AI in delivering optimal CX.

Insights from this session are also featured in the Welocalize White Paper: Global CX – The Power of Localization and AI in Customer Support and Online Help [register to download PDF]

The Increasing Importance of CX

CX has become a game changer. According to surveys, consumers are willing to walk away from a brand if they have a terrible experience. “We as consumers have become a lot more demanding as it relates to what kind of customer experiences we expect,” Schliem says.

As a result, brands have made CX a top priority. Most companies now have a Customer Experience Officer (CXO) and CX teams. And many business leaders indicate that CX is their number one engine for growth.

This focus on CX revolves around using AI-powered chatbots. There is a “huge shift in the expectation around how people are experiencing artificial intelligence,” Schliem notes.

Traditional Customer Support

The traditional multilingual customer support model relies heavily on human call center agents. This has plenty of downsides:

  • It’s costly and complex to scale a call center. Brands must maintain teams across geographies to serve different markets because a call center has specific days and times it’s available for support. They must set up local support teams every time they expand to a new language market. This buildup is expensive and complicated.
  • It takes time and effort to localize a knowledge base. The knowledge base of a global brand should be translated into different languages. Changes in the original text, say a new product or a new policy, require updates across those languages. Often, this involves retraining workers by building and localizing a new training program.
  • It’s prone to human error. “There’s also a real risk of incorrect information being communicated through these experiences. Humans are fallible creatures,” Schliem explains. A typical example is an agent sharing outdated information.
  • It can be a frustrating experience for customers. The escalation pathway can be a “roundabout, laborious process with multiple pauses in the interaction and lots of time on hold,” Schliem shares.
  • It involves working with multiple systems. Agents often switch to different software systems to conduct transactions. Human-mediated digital transactions are time consuming and inefficient.

Modern Customer Support

In contrast, Schliem notes, the modern approach to customer support uses “artificial intelligence and machine learning to enable interactions in a much more streamlined and agile way. The use of conversational AI chatbots is compelling.”

  • It’s available 24/7. Companies can provide 24/7, 365-day support. “There is no downtime. The machines are always working,” Schliem says. An AI system can support multiple channels, including phone calls, emails, and chat. “So, you now have an omnichannel approach that is consistent and can learn from all of these different channels,” he adds.
  • It’s easier to deploy multi-language support. AI chatbots can be trained to interact in different languages. Schliem explains, “When you want to add a new language, you are simply going to add that new language as a language model in the background and build some custom data sets to train it. But you don’t require the augmentation of your staff with people who happen to speak these languages.”
  • It’s better at problem identification and prediction. “One of the things that machine learning is great at is seeing patterns that humans don’t see,” says Schliem. Using large data sets, an AI system can predict a particular customer issue in real time and analyze trends at a macro level. “This means that you can be proactive about solving customer problems rather than being reactive,” he notes.
  • It handles the escalation of problems more efficiently. An AI chatbot can answer basic questions faster than a call center agent. And it can assign more complex issues to human agents, providing information to the agent in advance. It can even predict emotion. So the agent will be prepared to handle a call.
  • It makes knowledge management easier. “When you change something in the underlying knowledge related to customer support, that information can be very quickly updated in the machine learning rhythms,” Schliem explains.
  • It can handle multiple systems simultaneously. Digital systems can “talk” directly to one another. AI can engage with different data systems dynamically and in real time. As such, it can handle customer transactions much faster.

AI Limitations

While AI offers a lot of advantages, it’s not a cure-all. As Schliem points out, “There are some cases where AI is not a perfect solution.” These are some limitations.

  • It can’t handle complex use cases. AI may struggle with linguistically complex cases, for example, understanding English spoken with a thick accent.
  • It can’t handle code-switching. This is when someone speaks in one language, briefly switches to another, and then back. “This is a hard pattern for AI systems to recognize and to parse,” Schliem notes.
  • It can’t handle complex problems. Schliem says, “AI is not perfect, and there is some risk where an answer that a customer gets might not just be the wrong answer. It might be nonsense, and that would be a negative impact on the brand.”

Despite these issues, AI-driven CX is fast becoming the new model. “We have already reached the point where this is becoming and has in large part already become the standard paradigm for customer support,” Schliem concludes.

Find Out More About Global AI-Powered CX

Learn more about this fascinating journey toward AI-powered global CX and support. Download our guide “Global Customer Experience: The Power of Localization + AI in Customer Support + Online Help.” And watch Aaron Schliem’s entire presentation at the Let’s Go On Demand Summit.

Key highlights from this session are also featured in the Welocalize Guide, Global CX – The Power of Localization and AI in Customer Support and Online Help [register to download PDF]