mannequin head leaning against a wall
Editorial

Don't Let Chatbots Drag Down Your Customer Experience

5 minute read
Phil Britt avatar
There are definitely some areas where chatbots hinder rather than help the customer experience.

Global Market Insights predicts the chatbot market, estimated to be over $250 million in 2017, will grow at a compound annual rate of 31% between 2018 to 2024.

One of the biggest drivers for the growth is the hope of businesses that chatbots can improve customer experience. While that may be true in many cases, there are definitely some areas where chatbots hinder rather than improve the customer experience.

Chatbots Lack 'Humanity, Humor'

"There is an argument that interactive chatbots have a place in the market as many customers like the option to self-service,” said Lindsay Willott, CEO of Customer Thermometer. “According to Zendesk, 50% of customers think it’s important to be able to solve a problem themselves, and, according to Forrester, 63% of customers are happy to be served by a chatbot — if there is an option to escalate the conversation to a human.”

But it’s critical to maintain humanity at the core of chatbox usage considerations, Willott added. “Chatbots cannot yet deliver humanity and humor. If you look at the examples of live chat that have gone viral, it's often because they are really funny and have a shared experience (Harry Potter, RPG, etc.) — no chatbot could do that.”

Related Article: A Good Chatbot Is Hard to Find

Need for Trust

Trust of the technology is essential, particularly in industries such as health care and financial services, which routinely deal with sensitive personal information.

According to a recent UserTesting study that focused on chatbots in health care:

  • Chatbot apps still need to earn customer trust. Participants in this report noted unease in trusting unfamiliar brands with their personal information as well as the lack of HIPAA compliance with many chatbot apps. 
  • Chatbot apps do a poor job of handling complex conditions. All of the apps were inconsistent when diagnosing complex conditions. Some of the simplest symptoms tested would often lead to alarming and inaccurate diagnoses.  
  • Basic usability issues detracted from the overall chatbot experience. Loading issues for some participants, for example, made some chatbot apps unusable at times. Chatbot apps with a poor CX may lose potential users, revenue or worse. 

Though this study was focused on the healthcare sector, where trust is a larger factor than it might be in purely transactional interfaces (such as in retail), Janelle Estes, chief insights officer of UserTesting, said the following inferences can be drawn for chatbot cx across industries: “Most companies are investing in chatbots for the business benefit and return. They are touted as a solution to reduce supports costs, decrease the time to purchase, and increase conversions. However, these favorable metrics can't be met if the chatbot experience is not designed with the customer in mind.”

Learning Opportunities

Related Article: What Makes a Chatbot Tick?

Be Prepared to Train Your Chatbot 

Tom Hutchinson, principal product director at Acxiom offered a few other considerations before deciding to launch a chatbot — regardless of whether you buy one from a vendor or develop one in-house.

Are you willing to spend the time and manpower to train it? “AI is like a two-year-old child. It can think, but you have to tell it what it needs to know and how to evaluate it,” Hutchinson said. “It takes time and effort to train the system to recognize data and react to questions in the way you want it to. Even with machine learning capabilities, you have to create a baseline, otherwise its perceptions of the world be random or nonsensical and your users will be frustrated.”

While chatbots can provide information without additional support or staff, a company has to be prepared to deploy them in the proper way, Hutchinson added. “Simple data, with straightforward questions and pairs can be supported with a rules-based solution. More complex data requires an AI, but you have to be willing to provide it with data and train it to understand the data and how it ought to evaluate it.”

Related Article: Machine Learning or Linguistic Rules: 2 Approaches to Building a Chatbot

Tips for Chatbot Success

Estes recommended the following strategies to help ensure customers have a positive experience with a chatbot:

  • Offer a just-in-time discovery or prompt: Companies are finding the areas where people typically struggle or have questions and either prompt customers or make sure the option to engage is front and center at that time. This is a much more favorable experience than being constantly prompted by a chatbot, especially when someone has just arrived on a site. With this type of experience, the chatbots becomes an annoying roadblock that almost all customers will dismiss.
  • Use a conversational approach: Informal, conversational language that implies the customer is talking to a real human is much more engaging than an interaction that seems pulled from a standard script. Pulling this off is not easy, but it drives customer satisfaction and reuse.
  • Escalate complex conversations or scenarios: A chatbot cannot handle every conversation, especially interactions that evoke strong emotions — such as anger, fear and sadness. For example, the 1-800-flowers chatbot asks to connect a customer with a customer service rep when asking for recommendations for sympathy flowers. Knowing when to offer these escalation paths are critical for providing a great customer experience.