AI customer service and the importance of human interaction

When new technology is released into the world, it’s tempting to dive headlong into the new innovation and incorporate it into all aspects of your business.

In the era of artificial intelligence (AI), you may be tempted to add chatbots and large language models (LLMs) throughout your customer service offering. But here at Memberful we’ve observed that there’s a balance to be found between AI customer service and real human interaction, says Lauren Gilbert, Eyal Avital and Mau Fournier from our Customer Happiness team.

Particularly in the world of customer service, AI can be beneficial but should be used in moderation compared to a true, human, customer service professional. “Many people want to delegate 100% of customer support responsibilities to these shiny new AI tools, but there will always be cases that require actual human interaction. Delegating everything to AI will likely leave many of your customers more frustrated than they were before they wrote in,” starts Mau.

Let’s take a look at how, and when, you can mindfully incorporate AI into your customer service offering.

Essential qualities for a customer service team

Firstly, what does perfect customer service look like? How can it best help a company’s customers?

“We try to embody the five qualities of the PREACH model to maintain a customer-centric approach,” starts Eyal Avital. “We aim to be Proud, Responsible, Empathetic, Articulate, Concise and Human.

The team agrees that empathy is perhaps the most important of these. “Put yourself in the customer’s shoes and show empathy for their situation. Let them know that their business matters to the people behind the company who are helping to run it,” starts Lauren.

“Empathy is often overlooked,” adds Mau. “You need to embody empathy in order to understand the customer's issue, and you need to display empathy in your response to make the customer feel understood, which also helps lower their defenses if they're upset.”

In addition, customers need help from someone who knows enough about the product to be able offer all the needed information – plus the articulation to explain it. “When it comes to a platform that you count on to run your business, getting help quickly from someone who knows the product inside out and can explain how to solve a problem in a way that’s easy to understand and actionable, is beyond important,” thinks Lauren.

Customer service is not just a box-ticking exercise: genuine helpfulness is essential. “Don't just answer their superficial question — try to find out what's behind the question; what they're trying to accomplish,” thinks Mau. “Help them with the core issue, and follow up later to make sure they were able to get it done. People will leave the interaction satisfied if they sense that the person on the other side is genuinely trying to find ways to help,” he adds.

Lastly, don’t overlook timeliness and clarity in your interactions. “You need to provide prompt assistance, communicating clearly and concisely,” starts Lauren. Eyal adds: “You must be timely in your reply to customers. You can give A+ service, but if it's too late in the member's mind, they'll feel ignored and negatively view the interaction.”

Where AI customer service falls short

There is no doubt that AI can be helpful, especially for busy entrepreneurs and business owners, but it's not always the best option. We believe there are a handful of aspects of great customer service that AI can simply not recreate.

Personalisation

In our experience so far, AI customer service is missing the mark on finding solutions that actually meet a specific customer’s needs. We’ve seen numerous examples of help desk software that requires users to answer a long series of questions before getting an automated answer. And, by definition, an automated answer can only resolve a limited number of issues because it’s not a ‘one size fits all’ situation.

“AI customer service can certainly demonstrate expertise and come up with answers to questions – and it will often be better at this than humans, although its expertise might not be as up-to-date and aware of your specific domain as a human might be,” says Mau. “It can be helpful in short and simple interactions, but it won't take ownership of being helpful and it can't follow up like a human could, at least not yet.”

Longevity

Even though LLM (AI) apps have improved their tone when communicating with customers, there’s a way to go. “LLM apps can stick to a respectful tone of voice, but that's not the same as genuine empathy from a real human. The human capacity for empathy and relating a resolution to a specific experience can’t be replicated,” says Mau.

Similarly, AI customer service is all about the short-term. “LLMs have a short attention span – it's built into the technology,” he adds. “They won't remember the conversation you had six months ago about your customer’s challenges, or how your content serves them, or even the fact they like to go on fishing trips with their family. Bringing these details back into new conversations will make members feel like you actually care.”

Where AI customer service can help you

This is not to say that AI customer service cannot be beneficial for customer service. There are certainly clear-cut opportunities to automate, as with any new piece of technology – but it’s crucial to know how to innovate, when.

“We should put AI to work on routine tasks that can be automated, while leaving a human voice as an option for those with more unique requests,” starts Mau. “ LLMs offer a fantastic first draft for replies, but the best result will always come from your own editing of that draft. Bring your own voice into the conversation.”

“What used to encompass the FAQ on a website can now be handled with bots or AI to solve simple questions quickly and 24/7. Anything outside of one-click or one-question solutions should then be redirected to human support,” Eyal adds. “Otherwise, customers might be quick to lose patience. It’s like the old days of getting stuck in a loop pressing numbers to get to the right department over phone support.”

The integration with AI is a gradual process. “There's an evolution (crawl-walk-run) for integrating bots or LLMs,” starts Eyal. “New businesses should dedicate more time to their community with hands-on support. As they grow more popular they can offboard their basic recurring questions to LLM.”

“AI can help you scale as your career progresses, as more members will generate more questions,” says Mau. “Many of these will be simple requests that AI could take off your plate so you can focus on the more nuanced issues.”

“You could get away with relying on AI when you're starting out, but I would advise against that. I would recommend more manual support when you're starting out: the questions you receive in the early stages of your career are a treasure trove of feedback that you can use to learn about your target audience and how you can improve things for them.”

Conclusion: Is AI good for customer service?

It looks like the people-first approach we’re taking at Memberful is working well, combining a wealth of expertise with the empathy only a human can bring. Eyal says the way we work helps to humanize Memberul as a brand: “a mixture of empathy, sharing resources, and adding some fun through emojis and GIFs” keeps things lighthearted and helpful.

Mau shares some recent feedback from a Memberful customer, who noted it was nice to talk “to a real human with such an empathetic way of helping” and someone who could offer personalized guidance for their specific query. “That customer said it was the best support they ever had!” smiles Mau.

Lauren recalls one customer who wrote in with a concern about member retention. Using knowledge of her specific industry, and unique trends in her own business, we were able to offer advice on a pricing strategy that ultimately increased her members’ lifetime value.

“You could certainly use AI customer service for help with this type of question, but when it comes to decisions that could have a huge impact on your business, I’d venture to say that most people aren’t ready to trust AI alone with that – nor should they in my opinion”.