Improve Customer Service with a Smarter Chatbot
By Ken Dulaney
Why Chatbots Are Failing To Deliver On Their Promise
Vice President of Research at Aragon Research, Ken Dulaney recounts his subpar experience with a chatbot.
He observes how chatbots may be able to provide a better, more efficient customer service by leveraging customer data and content analytics that incorporates a company’s specific top performing attributes to deliver superior customer satisfaction.
The other day I needed some help with a service. Got onto the website to search for help, saw a chat option and clicked on that to see if they could help me. It didn’t take much to determine that I was interacting with a chatbot–an artificial intelligence (AI) software that mimics human speech to simulate a conversation with a real person.
The chat started with a cordial greeting and asked me what help I needed. Typed about 3 sentences knowing that it was an issue that likely wasn’t covered in the Frequently Asked Questions (FAQ) which I was sure the chat would try to direct me to. The chat from there ended quickly by telling me that I had to call in because my question couldn’t be answered. My reaction: not happy. A human would have easily known by just counting the explanation points in my reply. Was this an indication of a generational problem with today’s chatbots or just a poor implementation?
Yet, in looking over the past year when I needed support and was directed to a chatbot it became clear that most of the ones I had interacted with were nothing more than a fancy front end to the supplier’s FAQ. AI was used to map my natural language query to a specific or a set of FAQs. The hype and promise of last year about chatbot’s ability to replace support staff while simultaneously improving the customer experience seems to have been significantly overstated when examining chatbots in the aggregate. It is clear that implementation is the foundational issue today causing so many chatbots to underperform. So how can an organization understand the weakness of an implementation and know what to improve?
Later that day, Aragon Research had a conversation with AmplifAI, an AI-based coaching application targeted today at call center employee improvement.
The application gathers data from various sources to help exploit top performer attributes and to guide average performers toward better skill levels. So, it seemed obvious to me that chatbots need a coaching application too. Why not loop them in with the learnings about other elements of an organization’s support organization. Data should be maintained from surveys, metrics, and the percent of problem resolution. Yet, it is doubtful that my supplier had done much of the above as have many others who jumped on the chatbot idea but didn’t really treat it as digital labor
Furthermore, chatbots could benefit from a greater use of content analytics to determine the mood of the person using the chatbot, a greater use of the content that exist throughout the organization instead of just the FAQ repository. And this should be married with a real-time assessment of the customer’s sentiment to divert the call to a better solution, preferably a person through a call out scheme to make it as easy for the customer as possible.
The best examples of AI chatbots identify and consider the customer’s prior interactions and journeys so as not to always start from square one for every interaction, and many now recognize and adjust responses based on sentiment analysis. There are examples where chatbots can bring a smile to one’s face because they were done with a strong technology foundation implemented extremely well (pleasant voice based and very interactive).
As important as customer service is to the long-term relationship for any vendor, it is amazing how poorly chatbots have been integrated into so many organizations. It likely stems from the ongoing focus of customer service organizations: cost. Chatbots seemed like a way to reduce costs, so they were adopted. We have argued that customer service should be seen as a revenue opportunity and measured as such. That would drive more rapid improvements in the technology and ultimately in the satisfaction of the customer. Better implementation must be highlighted by the industry and organizations must put forth a more comprehensive effort to make their chatbots pay off.
For more insights on chatbot AI trends, schedule a free analyst inquiry.
RELATED
Aragon Research Releases Top Ten Technologies for 2022 and 2027 |
Collaboration Technologies Are Quickly Emerging |
3 Emerging Digital Business Technologies to Watch |
Check out our blog to stay up to date on the latest research and trends!
Have a Comment on this?