Blog

To Chatbot or Not to Chatbot? The Conversation Continues

By Mandeep Kwatra, HGS VP of Global Client Solutions and Capabilities

Earlier this year, HGS released a white paper on this year’s top 10 trends in customer service. Over the course of the year, we’re dedicating a blog post to each of these CX game changers. Here, we address our Trend 6, Chatbots, as we delve into the evolving expertise that supports a continually optimized chatbot experience—one that drives higher CSAT and NPS scores for today’s businesses.

In the past few years, there has been a lot written about chatbots and their future, abilities, and benefits. Amidst this hype-and-hope debate, many organizations have launched bots—to varying degrees of success—to optimize their customer experience management. However, it’s important to understand why research shows that many chatbot deployments fail to meet customer expectations.

First, most customer experience professionals don’t truly understand the possibilities of what a chatbot is and can do. So what is a chatbot?  Most know that a chatbot is a computer program that can manage a conversation using natural language and implied methods to understand the intent of the user and solve for customer queries. But it’s important to understand that today’s second-generation chatbots have come a long way, employing artificial intelligence (AI) and natural language processing (NLP) capabilities to speak to and understand language in a way that feels, to customers, almost human. Second, most of our colleagues get caught up in the whole idea of conversation management. They’re concerned that customers may not like the idea of talking to a bot. But they’re missing the point, because today’s customers don’t really need a conversation. They just need issue resolution at their convenience. Don’t believe me? The research supports it. In fact, Harvard Business Review recently found that 84% of customers would prefer a straightforward solution to their problem. And further research supports consumers’ comfort with chatbots.

According to a recent Statistica survey, 34% of customers are comfortable having their questions answered by intelligent chatbots (text or speech based) to help with their e-commerce transactions. And, surprisingly enough, healthcare customers’ acceptance was at 27% according to this survey. Even the public sector and government customers responded positively in double digits.

Now that we’ve established the chatbot’s unlimited potential and the fact that customers care most about resolution, some key questions still remain. Like, what does it take to launch a successful chatbot program? What are the limitations and dependencies for success? What is the secret recipe to make it successful? With these questions, I met up again with HGS’s chatbot friend to continue our conversation from a year or so ago.

Mandeep: Hello, Chatbot!

Chatbot: Hi Mandeep, how have you been? How’s the world of optimized customer experience?

Mandeep: You tell me. Your clan seems to be making all the right strides.

Chatbot: Yes, it was a rough start in the early years. You see, trust is very important in managing customer experience. And, finally, I think people have started to trust that we can solve their questions faster than our human friends. Today’s customers are ready to engage in conversation with a chatbot. We are seeing high demand for us across different industries.

Mandeep: I am sure! So what do you think is the likelihood of interactions that can be automated across different industries?

Chatbot: You will love this! Chat partners LivePerson did an analysis of different industry use cases with high likelihood for chatbot application. Results showed that retail, consumer packaged goods (CPG), and travel aggregators had a higher number of use cases, along with the credit cards unit of financial services. 

Mandeep: Interesting! So do you think that there is a shift in the customer mindset? About a year ago, everyone talked about how conversations were important. But now customers seem more adjusted to talking to a chatbot. That makes me wonder if convenience supersedes the conversation aspect.

Chatbot: I won’t say that the customer mindset has changed. It’s just that the expectation has become more real. Let’s be honest, humans are always in a rush. You can wait around on your smartphone on either a call or chat to have a buddy answer your question. Or, you can just ask us and do better things with the time you save.

Mandeep: So yes, it is about convenience. But I feel there is still apprehension among companies when it comes to questions like, “How fast can we deploy it?” or “How do we demonstrate ROI?”

Chatbot: All valid questions and important ones. Let’s look at the speed to deployment. A lot of people might be surprised to know that a simple chat bot can be deployed in as little as 60-90 days. It depends on a few things. Is your data structured? A simple chat bot can solve for the most common interactions like FAQs, etc. Now, if you want to deploy a claim processing bot or a travel reservation bot or a smarter virtual assistant, that may take a little while longer. Here is something that you may find useful:

 

Estimated%20Time%20Taken%20For%20BOT%20Deployment.jpg

Mandeep: I agree. In fact, one of the best practices I have seen, and one I always recommend, is to start with a smaller and simple problem and run a proof of concept. Once you have tested it successfully, you can keep introducing different scenarios and make the bot more mature.

Chatbot: I like that you talked about maturity. One thing I want readers to think about is that when you are training bots, train them like you would train a human. If you are expecting a simple web chat bot to do multi system based claim processing, you clearly need to realign your expectations.

BOT%20maturity%20and%20Transaction%20Map.jpg

Mandeep: True that! So going back to our earlier conversation, how does a company get going using chatbots effectively?

Chatbot: A bot to manage your customer interactions can prove beneficial rather quickly. But before you jump straight into the deployment, you need to keep a few things in mind. First, strategize according to purpose. Start building a plan based on why you want to deploy the bot. What problem are you solving? How many agents solve it currently? How much time do they take? If there is a bot deployed, how much time would be freed up? Will the bot make the customer journey easy or complex? Depending on the use case for which the bot is deployed, and the volumes that the business gets for it, millions of dollars can be saved.

chart%20three.jpg
MANDEEP: What is your best advice for those who have never used bot?

CHATBOT: Ah, great minds think alike, Mandeep. Here are the steps to bot deployment:

  • Understand your customer journey.
  • Assess the need.
  • Define current use cases and success criteria.
  • Identify the right tool and partner.
  • Create the information bridge within systems.
  • Focus on convenience not just conversation.
  • Define and deploy the right reporting tools.
  • Prepare the bot trainers.
  • Establish learning and try again.

In closing, the best advice I can give is, as you suggested, invest in a pilot. Try us out. Be open to the fact that adoption will be slow. You will have to train customers on our bot solution and ask them for feedback. There’s no one-size-fits all approach for us and how we work with agents to serve today’s customers. Humans like yourself, Mandeep,help to bridge the knowledge and understanding. I think you’re on the right path. Till we meet again!

LEAVE A COMMENT

Plain text

  • No HTML tags allowed.
  • Lines and paragraphs break automatically.
  • Web page addresses and email addresses turn into links automatically.
The content of this field is kept private and will not be shown publicly.
CAPTCHA
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.