Chatbots won't solve your company's problems

About 6 weeks ago I left my job to start a new live chat software company. The number one question I've been asked since I've started is what my opinion on chatbots is.

I've replied the same way every time: chatbots for customer service have potential but are a long way off from being able to replace actual human beings. I've asked 100+ people for examples of chatbots in action in a customer service setting and I have only seen one that seems to have some potential so far (I'll get to that one at the end). The rest of them just package up FAQs and present them in the form a conversation instead of search results, which I'd argue is almost always less efficient for the end user.

Here are the main issues chatbots currently face:

They lack empathy and don't bend the rules

I don't expect bots to have empathy, but if you've ever had a truly great customer service experience, there was a good chance that it may have been because the customer service rep you were dealing with actually felt for you and your situation and genuinely tried to help. They may have bent the rules to fix your problem or just really connected with you to make sure you felt like someone actually cared about your issue. Chatbots are often just complex decision trees, and are just designed to add a self-service layer between you and the customer service rep you’re trying to talk to. This expectation of human interaction is compounded when a user isn’t exactly sure that they’re talking to a bot. Some bots on the market are leveraging things like machine learning and artificial intelligence to make it seem like you’re talking to a human, but that frequently makes the interaction more frustrating.

They create too much friction

Chatbots may be able to surface answers to common questions to you, but those can usually already be easily found in the Help Center of any company's website. Many companies testing chatbots use them as the first line of defense (I don't blame them), but then make it a pain in the ass to escalate your issue to a real person if and when the chatbot fails to find an answer. If you're dealing with a chatbot that can't solve your issue and you don't know where to turn next, you'll probably end up feeling more frustrated than before. If you don't have somewhere else to turn, you'll feel helpless which is the absolute worst thing a customer could feel if you’re trying to deliver quality customer service.

They can't handle complex situations

Each company has unique challenges that only their company has to deal with. Every company has different software/systems, policies, decision makers, etc. This makes it extremely difficult for chatbot software companies to build software that can specifically address the complexities that YOUR company has. Sure they can make software to handle predefined questions, but do you really think a chatbot can go back and forth with a customer for 30 minutes until they can finally figure out what the answer to their question is? Bots may be able to solve more complex customer questions some day, but for now, real human interaction will continue to dominate high-quality support interactions. Small or medium sized businesses already have a hard enough time creating and maintaining internal and external help documentation, let alone developing a customized chatbot with all the information and decision trees needed to arrive at every possible outcome.

The bright side

The good news is that most of these issues are addressable in some way. The lack of empathy can be overcome if expectations are set appropriately and bots make it clear that they aren’t human. Contact friction can be fixed by making it easy to switch between a chatbot conversation and a conversation with a real person. The issue of knowledge complexity will take the industry longer to develop because it's a REALLY hard problem to fix. The good news is that there is no shortage of extremely intelligent people working on it. I'm sure it will get there...someday.

To date, the most promising use of chatbots I have seen is from a company called Directly. Directly started as a company that allowed you to leverage your "power users" to answer questions instead of your support agents. They're now starting to leverage the responses their experts have sent to customers to power a very simple, more empathetic chatbot (this could also be classified as a smarter auto-reply) that also makes it incredibly easy to talk to a real person if needed.

For example, the chatbot sets expectations properly saying that it is a response that was previously sent to someone else. It also provides the option to easily get help if that doesn't answer the question. It's super low friction, helps increase deflection rates, and doesn't create a user experience where customers feel trapped if the bot doesn't answer the question properly.

While I am bullish on customer service chatbots as a whole, I think it's way too early for the vast majority of companies to significantly invest in them. In the next few years, we'll see the AI powered chatbots get much better and we'll find out the best workflows that provide the right ticket deflection to customer friction ratio that still maintains a great customer experience. Once those key elements get sorted out, I think it will make a lot more sense for companies to take a much closer look at using chatbots for support.

Derek Homann is a co-founder of Median, a company that has a new take on live chat software that is designed to dramatically reduce the time it takes to solve customer issues. We are currently about to launch in a private beta. If you are interested in seeing a demo of Median or just want to keep up to date with development, sign up for updates at HelloMedian.com.