Automation helps to solve customer-facing issues faster. However, this benefit can come at a cost. Since automation is a process that executes without human intervention, it can make mistakes that a person would not have made. It’s like the digital version of asking a woman how far along her pregnancy is and finding out she’s not expecting 🤦.
While automation isn’t new, it’s starting to appear in many aspects of customer-facing support. Popular customer-facing automations include:
Inside a support organization, automation can route tickets to the right agent faster, classify, or label customer issues automatically, and measure customer sentiment or likelihood of an escalation.
In these examples, automation helps to solve customer-facing issues faster. However, this benefit can come at a cost. Since automation is a process that executes without human intervention, it can make mistakes that a person would not have made. It’s like the digital version of asking a woman how far along her pregnancy is and finding out she’s not expecting 🤦. Users can end up with nonsensical, or incorrect answers if they input data not precisely as expected, or have an issue that is not accommodated for in the automatic workflow. These errors result in a less personal experience and more frustration for the customer, which is no bueno mi amigos.
Is that cost worth it? In this article, we are going to explore the negative impacts that automation can have on the personalization of your support offering, and then provide some solutions to bring the personal touch back to customer support automation. Come on over, we’ll give you the same attention Lisa Simpson gave precious Linguo and you’ll be alllllll better.
Eivind Jonassen from Omnicus says, "The most important asset needed to truly connect with customers is the ability to empathize – an ability that AI lacks and is nearly impossible to replicate using machine learning algorithms." Empathy is important because customers who need support are often highly emotional (I’M IN A GLASS CASE OF EMOTION). When there are elements of urgency, disappointment, or anger involved, human agents are likely to detect it and respond with a comforting tone or a sincere apology. An automated voice or email, even if it has correctly detected the right customer sentiment, often does not have the same impact. When it gets the situation wrong, the tone will be off, and the customer's emotions will only escalate.
In addition, automation tends to repeat itself on each new interaction or when it doesn't understand something. Being asked the same questions over-and-over or hearing the same message repeatedly when there is a critical situation reduces the personalized customer experience (and is really annoying, let’s be real). It's likely you have had an experience with an automatic message saying, "Your call is important to us, a representative will be right with you." That assurance becomes less comforting (believable) after it is played in the same cheery tone for the tenth time after 30 frustrating minutes on hold. Repetition reduces the personal touch of your support through its lack of sincerity or empathy for the situation.
Automation cannot infer any details. It needs to be given all the information it requires or taught how to find the information reliably. Because automation needs data to make decisions, they often ask more questions to the customer. And everybody knows customers LOVE to answer tons of questions over and over while dealing with something that isn’t working that is frustrating the 💩 out of them. Without automation, an agent would typically guess these answers or not even ask them, since they may not be relevant at all to the conversation. Michael Murchison, of Ada, agrees: "The biggest issue with typical chatbots is a lack of personalization. When you speak with a human customer service agent, she can adapt to the direction of the conversation. A good customer service agent doesn't just read from a script but will offer different experiences for different customers depending on their individual needs." It’s true, human beings can actually listen (well, most of the ones in customer service can).
Besides not changing with the flow of information, you also place a burden on customers to do something agents historically would have done. Some chatbots, for example, ask the customer to select a category for their inquiry or the department they want to speak to. Often, the pre-selects don't fit what they are attempting to get help for, or it isn't clear which is correct because they don't have the knowledge to categorize tickets in the same way your agents do. You know, probably because it’s not their job to categorize the issue...😬
Another common case is where an automation requires a customer ID or order number to gain access to support. The user may not have one or not know where to find it. These gates to support have shifted a burden that’s normally part of the agent's ticket intake process on to the customer, making for a less personal experience, and definitely a more time-consuming one at that.
While many machine learning technologies are trying to solve this problem, most automation, whether it be knowledgebase searching or chatbots, still requires 'clean data.' What that means is that you must use certain words or phrases for the automation to understand your request. It may not work if there are background noises or spelling errors, and even the slightest exception could confuse the automation. Similar to a really bad game of telephone (has there ever been a “good” game of telephone?). Imagine going to a car repair shop and having to describe the problems with your vehicle using only the actual car part names the mechanics know. Some automation approaches do just that. Often this means that customers need to speak unnaturally or that non-native language speakers and customers with differing levels of knowledge will struggle with the automation. Self-help is objectively a vital strategy to employ, but you need to ensure that access is personalized to how your customers actually speak or write.
Some interactions can be made more personal through automation because it works in real-time for the customers; something a person cannot easily do. For example, G-Suite's knowledgebase automatically recognizes which account you are logged in to and provides you with inline, personalized steps within articles. A human would have to lookup an account's details in order to do this task, but automation operates in real-time to personalize the experience based on known metadata.
Automations in product can also detect if a customer is going down the wrong path. Instead of waiting for them to fail and call support, you can pop up a useful message, or a link to a knowledge article. These types of automation help the customer be more successful, and often solve problems before a customer even knows they had one. It personalizes the customer workflow through real-time use of your product.
Automation is routinely used for routing incoming support tickets. Using the information the customer provides, you could send it quickly to a subject matter expert, your company’s giphy guru, someone who speaks a certain language, or an agent who is located in a specific timezone. All of these provides a more personal experience for the customer by reducing the chances that they need to re-explain a problem, or that a less-knowledgeable agent goes down the wrong investigative path and lengthens time to resolution. In case you weren’t sure, most people prefer to have their problems solved both quickly and efficiently, crazy, eh?
It’s also possible to better serve customers through an automation that provides your agents with likely responses. For example, when a customer writes to you, your tooling could automatically highlight customer history, existing open tickets, suggest a response macro, or provide useful articles. This is what Halp Answers is designed to do. Instead of sending an answer directly to the customers, Answers provides the agent with the tools to do their job more effectively. These automated suggestions, along with a touch of human interaction, can help you provide a faster and more personal experience.
When a customer does make the decision that they need a human, you want your agents to be prepared so they do not repeat a question, or provide answers the customer has already tried. Chatbots, for example, typically gather operating system or browser information as well as chat history, all of which can be sent to your agent should the customer request a person to assist them. Your agent can then focus on the service, and not the triage, or worse, triaging a second time. Everybody knows what it’s like to explain your issues 7 different times in one phone call, wouldn’t it be nice for everybody to know what it’s like to only have to explain your issue once?
Towards Data Science found that preparing your agents with automation can have a significant positive impact on your customer’s experience: "The system should incorporate the technology that predicts the next step needed by the agent based on the actions of the customer. It should be able to collect data from each interaction and be ready to deal with the anticipated issues. The collected information can then be used to predict future customer problems, the agent’s response, and follow-up questions, which enhance customer experience by adopting a pre-planned approach." Preparing your agent with as much information about the customer issue before the conversation starts, helps create a seamless interaction that feels more personal.
Most importantly, no matter how much you automate, human interaction will always be required at some point. Personalization requires an agent to be one click, one swipe, or one key away. Whether your voice system listens for a request to speak with an agent, or your chatbot recognizes when a customer is frustrated, transitioning to a live person seamlessly can be the difference between a positive experience and a lost customer.
It's always important to consider what you need to automate and what needs that personal human touch. Start with tasks that already have low human touch or are invisible to the customer: FAQ searching, in product guidance, case categorization and routing, standard email response macros, etc. These repetitive actions remove some burden from your agents while providing a smoother experience for your customers. Conversely, avoid automating complex tasks, or issues associated with high emotions. These include: Billing inquiries, uncommon actions, or system outage notices and responses. Taking the human interactions out of these conversations is a recipe for disaster, and may even be considered a super cowardly move.
Also, be open about your automation and how to get to an agent if needed. Trying to hide that a customer isn't interacting with a person weakens trust, and is pretty creepy, too. For example, adding a phrase such as: "These recommendations were made automatically. If they didn't help, type 'agent' to speak with someone right away" to your chatbot helps guide the customer and lets them know that someone is there when they need it.
Automation can be less personal. Whether it's an encounter with a chatbot that peppers you with irrelevant questions or receiving an outdated set of steps from a vendor's auto-response tool, these impersonal experiences are common, but they don't need to be. As you design your automation plan, ensure your experience is sincere and empathetic, do not add unnecessary steps for your customers, and accommodate the accessibility for your entire customer base. Automation is not a one-size-fits-all solution and requires careful planning to achieve the balance of personalized support and a return on investment for your team. By giving thought to your customers' workflows and testing real-life scenarios, you can offer your customers both an automated and personal experience.
Learn more about Halp Answers here.
Today’s IT experts juggle multiple priorities, and the best leaders do it all while keeping up on the latest industry trends. To find out what trends will shape IT in the near future, as well as get insights into how the next generation of IT leaders should approach starting their careers, we asked two industry veterans, Sean Glang, Senior IT Manager at Recurly, and Jesus Ward, Director of IT, Superfly, to share their perspectives. Here’s what they said.