There was a time - like around 2015 - when chatbots were all the rage. Facebook, Microsoft, and Google all suggested chatbots might be the future of consumer interaction, and chatbot-focused startups like Digit were raising millions of dollars in funding. Now big tech companies have scaled back their use of chatbots, and even Digit has jettisoned them entirely.
What’s happening to the bots?
A growing mountain of evidence suggests chatbots just don’t work most of the time—and they might be doing more harm than good to your brand.
A chatbot fields customer inquiries and attempts to provide solutions with AI, either by voice or over a text-based chat.
Some chatbots simply follow a conventional question-and-answer tree, while others use AI that makes them more responsive to customers.
Chatbots might save some money—but there are big caveats. The main reason brands were invested in the idea of adding chatbots is that they see bots as a way to cut the cost of their customer service operations. Chatbots Magazine claims you can reduce your customer service costs by 30% with chatbots.
But the problem is, no chatbot resolves a majority of customer issues. In fact, according to Gartner, self-service solutions like chatbots solved only 9% of customer queries without needing to loop in a human. With such low success rates, you may be risking customer satisfaction for some cost savings today.
Plus, there is a cost to all of those failed customer interactions: namely, your customer’s satisfaction. 82% of U.S. customers report in surveys they want more human interaction from companies.
One big reason is that talking to a chatbot tends to be a major waste of time. After a customer already tried and failed to speak to a chatbot, they’ll be much more impatient by the time they get to speak to a human agent. And many might not reach out to a human at all: 53% of customers say they will abandon a purchase if they can’t quickly find an answer to a product-related question. That’s why giving them a live person from the start is always the right move.
Even the biggest, most sophisticated tech companies have tried—and failed—to turn chatbots into a staple of their business. In 2016, Facebook went all in on promoting M, its AI chatbot assistant for Messenger. But the company quickly discovered that even with its sophisticated programming knowledge, it couldn’t create a chatbot capable of fielding complex human questions. The Messenger chatbot failed to solve 70% of all questions that users asked it, meaning in 70% of all cases, a human had to step in to answer the customer question.
In a 2021 academic study, nearly ~35,000 customers rated their satisfaction on a scale from 1 to 5 after speaking to a chatbot. Roughly 66% of customers gave the chatbot a 1 out of 5, the worst on the scale. The overall average satisfaction was just 2.16 out of 5.
And in a 2019 study from Forrester, more than half of U.S. e-commerce shoppers thought interacting with a chatbot even had a negative impact on their shopping experience. Let’s dig into the areas chatbots can’t compare to human customer service.
A recent academic paper identified “affective issues” as one of the leading problems with chatbots in customer service settings. Chatbots, in other words, can’t respond effectively to a customer’s emotions. Whether they realize it or not, many customers “expect an element of sympathy and personalization within the interaction,” the authors wrote, and chatbots don’t pass the test.
Just as chatbots can’t personalize a conversation to your customers, they also can’t cool the tenor of a customer’s emotions. If a customer enters a conversation already angry, a chatbot, unlike a human customer service agent, won’t know how to meet them where they are. That is bad for brands because that’s about 20% of the population. For these people, chatbots are especially unhelpful. A survey in 2021 found that frustrated customers were extremely dissatisfied with their chatbot interactions and their desire to buy from that brand again fell dramatically after interacting with the chatbot.
There’s nothing worse than when customers go through the work of trying to spell out what their problem is, and the chatbot just doesn’t seem to get it. Many of us have had those experiences. You type “return to the main menu” and “I’d like to speak to a human” in the hope of fine-tuning your interaction with a chatbot, and the chatbot doesn’t recognize the command and instead responds as if you’re having a different conversation.
Even the most advanced chatbots are built only to handle two kinds of scenarios: live chats in a messaging app like Messenger, or an automated voice response system on your phone line. But customer service is migrating into new platforms. Around 70% of customers report using social media to solve their issues, most often by tweeting at a brand or by commenting on their posts. These interactions usually involve questions too complex or too contextual for a chatbot to be able to answer.
Most companies choose to use chatbots because they’re on a tight budget, and they think that chatbots will help them save money on their customer service experience. But next-generation customer support solutions lets you have both human customer support and affordable pricing. Chatdesk provides flexible customer support staff who can engage with the nuances of every customer question. Chatdesk starts at just $99 per month - and majority of our Experts are based in the U.S. To learn more, you can schedule a demo with us today.
Having some automated messages is not a bad thing, as long as automation isn’t the default way customers can interact with you. Plus, there are some experiences where you won’t need a human on the other line. For example, nail care chain MiniLuxe uses automation to help customers schedule appointments faster. When a customer calls the MiniLuxe phone line, MiniLuxe deflects the call to text. A customer gets an automated text offering them a chance to book an appointment in the MiniLuxe app, rather than over the phone, where the wait time isn’t as long. The company now saves ~$2.85 per appointment booked.