AI Is Not The Answer To Customer Service Woes

A lot has been made in the past several years about artificial intelligence and what it’s implications could be for the customer service world. While it still seems like AI technology and AI focused companies have been dominating the news cycle, not much ground has been made in terms of applying AI to customer service.

When AI was first linked to the customer service world, companies had grand visions of replacing call center reps with machine learning bots that would reduce call center costs to next to nothing. These bots would seamlessly be able to handle any customer service request, providing instant answers to all customers, and creating better experiences at a cheaper cost for all companies trying to service their customers.

Best of all for companies that traditionally view the call center as a sunk cost or cost center that decreases the company’s bottom line, AI in theory, could allow a company to provide better service to it’s customers, all while increasing profit margins for the company and increasing shareholder value. If you were an executive at a public company, when news of AI first came across your desk and you envisioned increased productivity at a lower cost, you would have thought it was too good to be true.

It all seemed great on paper. Until it wasn’t.

While the idea behind using machine learning and AI makes complete sense for all parties involved, the execution has been lacking. Even worse, it’s now costing companies millions as customers are becoming increasingly more frustrated with these companies, and churn related to this frustration increased 5% from 2016 to 2017, crossing the 50% mark to a staggering 54%. Think about that for a minute. 54% of all customers are now leaving one brand for another, all due to poor customer service. The main reason for customers leaving is due to the lack of accuracy and speed from the IVR and chatbots, leading to more frustrating experiences.

Facebook M, the AI technology that seemed to be at the forefront by promising customers could interact with their companies, all through it’s messenger platform, recently shut down. When it was launched, Facebook promised a bot that would understand user’s issues, be able to connect a user to a rep at a company, and even be able to help book plane tickets or initiate returns to stores. It seemed like a great idea, but in the handful of times that members of GetHuman interacted with it, the bot had a hard time understanding simple customer service requests and in the end all the bot could suggest was for our team members to actually call the company ourselves. Each chat was a complete waste of time as the reason we were contracting the chat was to avoid calling a company in the first place.

Now Apple seems to be entering the same space with it’s announcement of “Business Chat” that sounds a little too close to Facebook M for our liking. Business Chat is promising to allow customers to connect with companies through chat and avoid frustrating customer service phone calls. As much as we love Apple and all of it’s offerings, we are highly skeptical that this will work. Despite the fact that every year there are studies that show a staggering 78% of Americans prefer to speak to customer service reps over the phone, companies are too tempted by the higher profit margins and lower expenses related to chat and AI bots.

If it feels like we’ve been down this road before it’s because we have. Whether it was the invention of 1-800 phone numbers and the call center in the 1960’s, the creation of the IVR in the 1970’s, call center outsourcing in the 1980’s, email and live chat through the internet in the 1990’s, and now social media and AI in the 2000’s, it seems there’s always the promise of new technology that’s going to revolutionize the customer service industry. Sadly it seems we’re still looking for answers.

In the case of AI and machine learning being applied to customer service, there are still some major issues that need to be worked out.

Customer service needs to learn from it’s failures before it can predict what’s best for the future. When a piece of sand gets into the engine, it can slowly wear down the engine until it stops working. Instead of blaming the engine, the engineers should figure out how the sand got in there in the first place, and then prevent it from happening in the future.

The biggest problems in customer service have been consistent throughout time. These include:

1. Having multiple interactions with a company to get an issue resolved
2. Being transferred to too many people, trying to find someone that can help
3. Customer service reps that don’t understand the issue and can’t help

When looking at that list, chatbots and AI still won’t be able to address these issues. Each one of these issues can really only be resolved by an actual person that knows what the issue is, that exhibits empathy towards the customer, and will do whatever it takes to get the job done, and this includes knowing when to engage a supervisor. Sorry, but it’s hard to imagine a robot being able to do that anytime soon.

Chatbots are too stuck to their decision trees and can’t intuitively understand what the issue is, like a person can. As this is the case, they can’t make decisions outside of the decision tree. This is very important to customer service as often times, some of the best experiences users have with customer service departments come from reps who go outside their job responsibility to provide the user with an incredible experience. We hear from users all the time about great experiences from call center reps that went above and beyond to provide a refund or waved a rebooking fee. Unless someone programs the bot to do that, it’s not gonna happen.

All that being said, there are still some attractive features of AI that should help make some components of customer service service better. There are certain functions within customer service where using AI to create workflow efficiencies makes too much sense, however expecting it to completely replace call centers altogether is shortsighted.

Requests that bog down agents that users can do themselves like making a flight or hotel reservation can easily be handled by bots. Placing a service request to your cable provider could also be handled by a chatbot. Government agencies should also adopt this technology as many requests to the government require a form either being downloaded or submitted. Those are jobs that could be handled by an AI robot.

Simple customer service requests that can be taken out of the IVR should only help to reduce long hold times, and allow users to interact with customer service reps that can actually handle their issue. In theory it should also give these agents more time to dive into the issue, call in a supervisor if need be, and go the extra mile to help customers have better experiences, thereby decreasing customer churn.

At GetHuman, we used AI to collect user information in order to help our customers with their customer service issues. Chatbots are really great at creating a comfortable experience for users when done right, but when it came to actually helping users with their actual issues and not just collecting information, we need the help of actual human beings because lets face it, customer service problems are complicated and time consuming.

We’re all for innovation and disrupting industries that need change injected into them, but lets all stop assuming that AI is the future of customer service. The core issues that have plagued customer service for years, still remain. We’re better off focusing our efforts on fixing those issues first, then designing bots and technology around those improved problems in order to create a better experience for all moving forward.

Make sure to follow our blog and keep checking back for tips on dealing with customer service, and helping you, the informed consumer, save your time and money at gethuman.com.