Kustomer is the only true omnichannel SaaS platform reimagining enterprise customer service to deliver standout experiences – not resolve tickets.
October 28, 2020
So, you think you’re ready to invest in conversational automation, but you want to avoid the hype and nonsense? Great, then we are here to help! Keep these seven points top-of-mind as you build out your chatbot program, and you’re sure to see value in no time.
“Containment rate” (the percentage of total conversations fully handled by the bot), or its alternative name, “deflection rate”, is a key metric to track when trying to figure out how well your bot is performing. Customer satisfaction is also important. Keep in mind how the introduction of a digital assistant could alter existing performance indicators. For example, will average handle time increase now that agents are only handling more complex inquiries? Ultimately, a well-defined bot program will be able to communicate increased agent efficiency and customer satisfaction, which equals a reduction in the cost of care.
Your first bot does not need to be elaborate. In fact, we recommend against it. When you are first getting started, pick one or two simple, but useful, use cases to automate. Then, you can learn and iterate as you discover how your customers prefer to interact with a chatbot. No one gets it totally right out of the gate, so avoid wasting time by trying to build something “perfect”.
We have seen countless chatbot programs fail to engage the existing front-line customer service team when designing an automated conversational experience. It’s great to learn from data and prevailing user experience research, but your agents are the ones who know how your customers are interacting with the bot. Treat the bot like another agent: when you need performance feedback, use its peers.
Not all chatbots are “conversational AI”, because not all use cases require machine learning. Very effective bots can leverage rules and simple conditional logic, it all depends on the use case. Similarly, natural language processing is great when you have a bot with many different skills and a large corpus of knowledge — why make your customers trudge through structured flows when all they should do is ask the question directly? In both cases, we recommend leveraging buttons, quick replies, and other conversational templates that help the user move through the conversation quickly and efficiently.
A chatbot is not a replacement for a human agent. Often you need to give the user a way to bail out of tough conversations and difficult questions, and that’s alright. Chatbots are excellent at fully resolving low level queries. However, just because an issue is complicated does not mean a chatbot cannot be helpful. Consider how you can use the bot for information gathering and light triage before routing to the right agent. In these cases, the chatbot helps reduce handle time and expedites the customer’s support request.
Chatbots get a lot of attention when it comes to automation. Often it’s the mental model in our heads for intelligent customer service. Consider other ways you can streamline the customer support experience with a bot, and leverage additional intelligent services: automatic tagging, routing, and prioritization for the agent, just to name a few.
At the end of the day, the success of your chatbot comes down to how well it fits into the support journey and cadence strategy you have outlined for your customers. Consider different segments of customers that might prefer automation to that “direct human” connection. Perhaps automation can be more helpful at the end of an interaction than at the beginning. Take a good look at your customers, and we’ll help you find out the right size that fits.
Want to learn how to get started with intelligent chatbots? Find out more here.