Why the best chatbots have a human fail-safe

Want real value from chatbots? Use the 80/20 rule. An interview with Zendesk's VP of Engineering & MD Singapore on customer service automation.
24 June 2020

Over chatbots already? Quantum computing could change that. Source: Shutterstock

  • Chatbots have helped automate customer service functions amid the difficulties of the pandemic
  • The technology is becoming increasingly advanced, offering customers a convenient and natural channel for enquiries
  • For all their innovation, they shouldn’t replace human agents

Chatbot technology has become a customer service staple. Customers want to reach businesses on channels that they are already on, and chatbots help bridge that gap on websites, messaging apps and social media.

According to estimates by Invesp, two-thirds of customers worldwide used a chatbot for customer support in the past year. Amid the COVID-19 pandemic, the technology has seen further uptake; as customers actively avoid crowded places like shopping malls, supermarkets and even workplaces, many businesses have turned to digital self-service solutions to help support agents work remotely.

Easily accessible and quick to deploy, organizations both large and small have embraced these tools for customer service, as well as for generating leads and supplementing sales process.

As a result, businesses enjoy cost savings, improved customer satisfaction, and increased conversions. These tools have become so popular, that 85% of all customer interactions could be handled without a human agent this year.

Mimicking human conversation

Aside from the convenience, one of the key draws of chatbots is that the conversation flow and style mimics those they’d have in ‘real life’.

In many cases, the customer won’t even realize they’re not speaking to a human agent, especially when they’re paired with a live chat function, making them a powerful 24/7 extension of the customer support team.

Good chatbots will be fully personalized to the company using them and, being plugged into the business’s knowledge base, will use the necessary context in their conversations, such as past enquiries, purchases and other customer records.

“Advancement in AI and machine learning capabilities have made it easy for businesses to integrate chatbots for efficient, real-time customer engagement,” Abhishek Deshmukh, VP of Engineering & MD Singapore at Zendesk, told TechHQ.

“This means they can learn from any past mistakes and provide more accurate, in-context recommendations in the future, while factoring in colloquialisms.”

The 80/20 rule of chatbots

This technology has undoubtably changed customer service for the better, adding efficiency and new value for both the business and its customers, and the tools will continue to advance in their capability rapidly. But for all these benefits, we shouldn’t expect them to replace human agents – yet.

“[Chatbots] can handle the dull, the dirty, and the data – such as addressing common inquiries, tracking packages or directing visitors to FAQs – but predictions should not be confused with decisions,” explained Deshmukh.

For chatbots to be really effective, they need to be ‘intelligent’ enough to know their limitations and know when to escalate them to a human agent with the necessary context.

Deshmukh recommends applying the “80/20 rule” which, for many businesses, has shown to be a good gauge to ensure a “healthy balance” between chatbot and human support.

This is because roughly 80% of customer tickets are repetitive, high frequency questions, while the remaining 20% typically require human intervention. By taking on this menial workload, those enquiries that must be elevated can receive a more attentive response from agents, whose jobs have been made a lot more interesting.

“For example, chatbots empower customers to self-serve, reducing wait time for simple inquiries like tax policies,” said Deshmukh. “This then frees support agents from having to explain the same question 20 times a day, allowing them to focus on higher level work.”

Deshmukh continued: “Humans will always have a part to play in customer support. Moving forward, I expect to see chatbots becoming better at intelligently handling complex customer interactions over to support agents.

“While chatbots will improve in function with more breakthroughs in AI and machine learning, many situations in customer service still call for empathy and the human touch.

“In fact, our research shows that while 35% of people prefer an instant response from a chatbot to a delayed reply from a person, 60% want to be able to speak to a human at any given point during an engagement with AI.”

The future of chatbots

So what’s next for chatbots? Deshmukh hazards that businesses will start to realize that the rise of conversational business in “less about the channel and more about the conversation.”

Chatbots will be one component of a broader omnichannel customer experience strategy, where customer data can be collected in order to create more personalized conversations over time. This will be a boon for the customer service consumers receive, but it also become a necessary evolution and customers’ expectations continue to rise.

“Customers don’t care which department is behind the chat window – in fact, [most customers] expect companies to collaborate internally so they don’t have to repeat themselves regardless of the channel they use,” said Deshmukh.