ChatGPT and Customer Support — A Changing Ecosystem

Gladly Team

Read Time

7 minute read

Unless you’re tech clairvoyant, you weren’t expecting the name “ChatGPT” to become the most talked-about tool on the planet so far in 2023. Since the mega-popular chatbot launched, Microsoft has thrown a significant investment behind it, while Google has announced it’s cooking up its own language model. The powerful AI seems limitless to the general public, and its expansion into support centers is inevitable.

ChatGPT and Customer Support Teams

Support centers are unique enterprises — they need to be ruthlessly efficient to serve as many customers as possible, but they can’t sacrifice the relationship-building interactions that encourage shoppers to stay loyal to their favorite brands. A tool like ChatGPT could theoretically be a bridge between automation and personalization if brands implement it in a useful way.

As a brand leader, you have a responsibility to understand the nuanced applications — and yes, even the consequences — this technology could have. The retailer of the future needs to be able to handle what’s coming — being innovative and capable of finding ways to incorporate next-gen tools that improve the way they serve their customers.

Gladly Chief Marketing Officer, Jason Finkelstein, sat down with CX leaders from Crate & Barrel, Support Ninja, and Thankful to talk about ChatGPT and how it’s changing the customer support ecosystem, and what CX leaders should do about it.

The consensus is that while ChatGPT can undoubtedly increase efficiency and make automated service better, it still requires a human touch to create an exceptional customer experience.

Check out an excerpt from the webinar for key insights from the panel.

Will ChatGPT Change the Future of Customer Support Teams?

Pat Tamondong, Senior Manager, Ops & CS, Crate & Barrel: It [will], if utilized correctly. But there are things that you still won’t be able to do with it. And I think it’s more about time because it’s getting smarter, and it will continue to get smarter. But we still have to operate that with certain limitations. So really, the effectiveness of automation, a lot of it depends on data quality; what you put in is really what you get. So solutions require everything to align, to get to the right solution, and you may not have all the pieces.

So that’s when the imagination of an associate comes in, and they can fill in the blanks to help you get to those places. It’s great for simple tasks and processes. Things like replacements that are basic, but when it starts to get complicated, multiple things, multiple addresses, you know, things that require a lot more action and have more dependencies, that’s when again, you may need an agent to do it. It’s a great supplemental technology because agents have a great grasp of the little things that happen throughout the day; things are changing every moment, every hour, every minute. So outside factors such as weather system outages, things that happen to the internal business, agents can understand that a little bit better than AI because the data might not be there for it to make an educated resolution or educated guess.

The personal touch and emotional connection are hard to replicate. Data can help personalize the experience, but things that happen during an interaction are as important as the data that’s in their personal profile. So, you know, hearing about a person’s rough day in the middle of the conversation and showing that empathy, making that personal connection. Now, those are skills that are really hard to replicate with AI in general. So you still need that human agent to be able to do those things.


Find mentorship opportunities, free courses & more in Gladly Connect.

Join now

Integrating AI With Other Systems

Evan Tann, CTO, Thankful: We’ve seen a lot of tech vendors in this space start by providing suggestions to agents of what to say but not actually communicating directly with customers. And I think that if you’re just using ChatGPT on its own, that’s the best you can do. But when you integrate it into all of your policies and all of your processes and tell it exactly what you want it to communicate, we found that it does an amazing job at adapting your macros to the specific situation. Without those hallucinations, you just need to give it all the context it needs. So really pulling integrations from all of your backend systems and presenting it in a way that is ingestible by ChatGPT.

ChatGPT uses this thing called a prompt, where you’re describing in human language what you want it to do, just like you would if you’re talking to a fifth grader. So you need to describe it in such a way that a fifth grader could accomplish that task exactly as you want them to. So you’re going to describe everything about the customer, everything about the package, everything about the shipment, and how it’s gone so far. Here are our policies and SLAs, and solve this problem as a customer service agent with this particular brand voice. And it tends to perform extremely well in those scenarios.

Connor Tomkies, CEO, Support Ninja: So if you’re connecting like your knowledge base, and your fulfillment machine, Zendesk instance, and your Omni channel support, and you’re providing all that information, I can make some good decisions potentially.

Evan Tann: Exactly. Yeah, but lacking that, the best you can really do is summarize or rephrase things and suggest to agents how they could reply, but it’s not going to be able to get it right, you’re really going to need a human for that last mile unless you can do deep integrations at all your systems.

The Loyalty Economy and ChatGPT

Jason Finkelstein: Great customer service drives customer loyalty. So let’s talk for a moment about what this loyalty economy idea means and how ChatGPT fits in. In the last decade, there’s been this Cambrian explosion of brands — it’s never been easier or cheaper to create a brand, and the barriers to entry have never been lower. It’s never been easier to copycat products. Cookies and third-party data are disappearing. And this is all to say that competition and customer acquisition costs are going to continue to rise. This is all creating this growing necessity for these brands to drive customer loyalty. Loyal customers retain and grow revenue and also lower customer acquisition costs, right? When brands delight existing customers, average order values, increased repeat purchasing, and purchase frequency increase. So yeah, revenue grows, but delight also increases customers’ love of the brands and, therefore, word-of-mouth. So all that together lowers customer acquisition costs. Layer on top of that, this growing tension between the number of channels that brands need to communicate with customers, that’s on one hand. And then customer expectations, on the other hand.

Consumers ultimately don’t care about channels. They care about the relationship with the brand holistically, independent of the channel that they use to communicate. So you have all these exploding channels, right, more and more channels, making it harder and harder to meet these changing consumer expectations that I have as a consumer. It’s my expectation that a brand just talks to me like they know me without channels getting in the way. Which brings us to this big question, how does ChatGPT play into this loyalty economy? And how can brands make this an advantage? Let’s assume that ChatGPT allows for better automation more broadly and creates opportunities to change this mix of human and automated service. In the backdrop of the loyalty economy, should brands use the greater efficiency of their customer service teams in service of cutting agents and saving costs as an option? Or, as an alternative, should brands use the greater CX team efficiency and service of empowering human agents to make deeper connections with customers, drive loyalty, and thereby realize all those benefits?

Pat Tamondong: So you can’t do both. I know for us, one of the important things is our brand, right? We love our brand; we think it’s super important. And people connect with it. So being able to create a really high-level brand experience is also very important to us. So for us, I would want to repurpose it, I want to handhold you throughout your entire delivery. You know we’re helping you pick out a swatch; we’re helping you place an order. And we’re communicating that it’s ready to get delivered. You know, we’re checking on it and making sure that the delivery went well. We’re checking on you six months later to be like, “Hey, how’s the sofa that you just got? And guess what, we just got a new picture that would be great in your living room.” You know, being able to see how we could really create a great relationship that lasts over time, I think, is something important that we want to do. And again, just spend time on saving things, you know, saving cancellations, making sure people are consistently happy and don’t walk away with a bad feeling overall. So I think for us, we want to really repurpose our agents to do more and spend more time with our customers.

Want to learn more? Watch the entire webinar, How ChatGPT is Changing the Customer Support Ecosystem & What CX Leaders Should Do: