Photo: Canva
Will Chatbots Ever Drive Customer Service Interactions?
Only 8% of customers used a chatbot during their most recent customer service experience, according to a survey by Gartner. Of those, just 25% said they would use that chatbot again in the future.
The survey comes as businesses continue to invest in chatbots for customer service because of their often faster response times, 24/7 customer support, and potential labor savings. Advances in artificial intelligence employing natural language processing are also promising to elevate chatbot experiences.
Gartner’s survey, however, found that resolution rates vary greatly by issue type and remain a hurdle toward greater adoption. For instance, just 17% of billing disputes are resolved by customers who used a chatbot at some stage in their journey, while resolution rates for customers making a return or cancellation were as high as 58%.
Gartner said that customers’ lack of a solid understanding of chatbot’s capabilities and limitations is making them wary of using chatbots for any purpose. The survey found that customers are only 2% more likely to use a chatbot for a return or cancellation than for a billing dispute, despite a considerable difference in resolution rates between the categories.
“Chatbots aren’t effective for all issue types,” said Michael Rendelman, senior research specialist in the Gartner Customer Service and Support practice. “As generative AI makes them more advanced, customer confusion about what chatbots can and can’t do is likely to get worse.”
Recent surveys from Ipsos and Cyara likewise revealed low opinions of customer service chatbots.
The arrival of chatbots like ChatGPT have at the same time heightened concerns about AI’s potential to amplify misinformation and cause harm. A survey of customer service leaders from Helix found that while a wide majority of respondents are enthused about the potential of generative AI-driven chatbots to propel faster customer interactions, 50% cited data privacy and security as significant concerns.
How consumers want to interact with chatbots is also being researched. A university study found AI chatbots showing positive feelings — such as adding an “I am excited to do so!” or a few exclamation marks — turned off many users. Denny Yin, co-author of the study and a professor at the University of South Florida, said in a press release, “The reason is that people do not expect chatbots to have feelings. People don’t react to chatbots the same way as they react to humans.”
Discussion Questions
DISCUSSION QUESTIONS: How much confidence do you have that generative AI will elevate chatbots into retail’s primary customer service tool? Is the poor reputation for customer service chatbots currently more about subpar technology, customers’ poor understanding of or comfort with the technology, or retailers misappropriately employing the technology?
Poll
Loading …BrainTrust
Lucille DeHart
Principal, MKT Marketing Services/Columbus Consulting
Bob Amster
Principal, Retail Technology Group
Some customer service functions have long since been automated: checking the status of an order, reciting opening hours, looking up basic information. None of this necessarily requires AI. Indeed, you probably don’t want to use AI for it as applications like ChatGPT tend to be verbose and give unpredictable responses when a simple “your order will be with you Tuesday” will suffice. Dealing with more complex queries such as processing a return or placing a difficult order could be dealt with by AI. However, I suspect that guardrails will need to be put in place to ensure AI doesn’t go off track or start making things up – something that has become more of an issue with many large language models.
I hope they don’t! There is room for chatbots to assist. There is even room for some AI to enhance how much chatbots can assist. But I believe there is a limit. It used to be called ‘the human touch.’ This is strictly an issue in psychology and not technology, but “No man is an island entire of itself.” The day a chatbot can sound a like a friendly human, we will all be in line to be duped by computers that sound like your next of kin. The customer should know that it is talking to a chatbot.
The acceptance of ChatBots will be noted, not with surveys, but when, at the end of the interaction, the customer says, “Oh, I thought it was a human.”
…or “This is better than a human.”
Will AI get there? I have no doubt. Now that is a little scary.
Like everything else, AI will continue to get smarter and elevate its contributions to whatever role the tool is applied.
No.
The Chatbot “journey” is going to go the same way as voice response technology. (“Press 1 for the address” “this call may be monitored” etc etc etc) and it is a depressingly boring journey. If it is possible to calculate the time wasted by having to listen to this nonsense the lost productivity would be staggering.
Chat will be just as bad. You read it here first!
I’m only “somewhat confident” in the progress of chatbots until the technology can be refined and improved to solve the most frequent issues that consumers inquire about. There will certainly be situations where human interaction is required, and I prefer a data-driven approach to retailers’ strategy on this topic. If the data indicates the chatbot is indeed solving the most frequent inquiries with relevant solutions…great. So far, most data I’ve seen (and my own experiences) indicate there’s miles to go on optimizing this solution to customer experience management.
Chatbots have gotten better in regards to customer service but still far away from solving really intricate customer issues. The other improvements I have been recently, is that if it senses that they are going into a wormhole, they will move to taking to a live person faster than trying to ask irrelevant questions. I don’t think humans will ever be replaced, but I do believe that they are getting better about solving the simple unfettered questions.
Chatbots have not successfully met customers’ needs in the past, leading to negative perceptions. However, with the recent developments and the introduction of ChatGPT, there is hope for a more positive impact and an increase in customer acceptance of chatbots for customer service.
Chatbots are suitable for self service – Where is my order, returns/refunds, next steps in the job application process. They struggle with more complex tasks – checkout/orders, exchanges, or why is my I9 holding up my job application. They help retailers remove costs by saving the most expensive resources – humans/agents for the most complex tasks & improve the service level for high value orders or tasks.
Large Language Models eventually may be good enough to handle complex customer service transactions through bots – (ie. explain the cell phone contract to me), but I do not think I would be willing to be a first mover to gain that kind of efficiency at the price of missteps that hurt the brand. Model collapse is of particular influence as AI is creating the data that many models are training off of, which does not make me feel good about turning over customer service completely to robots in a consumer driven business.
Chatbots fueled by generative AI (as in ChatGPT) are far different from the digital support chatbots from just a year ago. And it’s an inexpensive digital support solution. However, the retailer/brand must be very aware that it isn’t a total substitute for human-to-human.
AI may not solve chatbots become primary customer service tools, but they will help, if only getting them farther along than the clunky product many of us have used to date.
If retailers and brands are going to funnel initial customer interaction toward Chatbots, then it is incumbent on them, not the customer to figure out how to maximize resolution. If the Chatbots aren’t ready for prime time but still placed on the home page, then clarify only will do, what it’s limits are. Otherwise customers deserve to assume your available service and care channels should be operating efficiently and with their best interests in mind.
Today’s chatbots based on generative AI are very different from the previous generations and training them is much easier. They will continue to grow to a point where there may not be a discernible difference between a person and a chatbot, especially for resolving all of the most common inquiries. I’ve been experimenting with a new chatbot already…this is just the beginning!
I’m not sold on people don’t expect chat bots to have feelings or that the reactions are so different. In many instances, consumers don’t even realize they’re speaking to a chat bot. Rather, I think the more generative AI can help bots develop a more natural language, fewer consumers may be turned off by bots. Consumers do not believe that bots can replace human engagement. Further, consumers primarily believe they need a human to resolve most issues, whether that’s true or not. So their is an immediate reluctance to engage. There is definitely a need for consumer education, but that will take time. Meanwhile, retailers must be careful to bring consumers along and not let technology, even when it’s helpful, get too far ahead of consumers; or else consumers may be turned off by the retailer, not just bots.
Chatbots can help some consumers get access to simple questions, but consumers are overwhelmingly more comfortable with a real person assisting them with complex matters. For example, chatbots can be helpful if a person needs to notify their credit card company of days that they’ll be out of the country, but they are not equipped to help someone navigate fraudulent charges and being locked out of their card while overseas. Chatbots are definitely an asset, but I don’t think they’ll ever fully replace customer service professionals because they are unable to provide a high level of personalized conflict resolution that customers need.