Chatbots are meant to solve a wide variety of needs through artificial messaging, without a human’s help. People nowadays can perform simple tasks, be served or notified about different subjects from anywhere and at any time. According to a business strategy, messaging apps (Facebook Messenger, WhatsApp, Slack, Skype, and more) can be programmed to receive a human input and respond to it with different kinds of automated actions. Among the uses given to chatbots, we can find several.
Different examples of uses for chatbots
- Customer care or support. Virtual help-desk agents are a type of chatbot that can help users to submit and resolve tickets without interacting with a real human being. Although it’s difficult to balance a quick answer with a customized one that would fit the consumer’s demand or expectations, chatbots are a better solution rather than not receiving an answer at all, or even getting a “we’ll be back” message.
- Online merchant interactions. To complete a purchase, a customer sometimes needs an answer to a question or extra information to make the decision. It’s important that he/she can always get an instant answer in order to avoid losing the sale. Automated e-commerce agents provide a solution to this need.
- Other specific functionalities: like expense tracking and accounting, scheduling meetings, or other particular purposes can be solved with a chatbot help.
- For gathering resources. When people can interact with “someone” all the time, they’ll also be sending content that you can later use as “user-generated content” for your social media strategy, while your customers get a feeling of counting on a constant companion that is always present.
But among all of the uses we can think of, we should considerate chatbots as a never-ending resource for creating better experiences for users, no matter the action they are performing. Because of this, we need to pay attention to some clues.
Tips to improve a customer experience through chatbots
- The importance of chatbots as layers, not as a unique, and sometimes, frustrating solution. In scenarios where human intuition or contextual understanding is needed, chatbots have limitations, so it’s necessary to complement them with human contact at some point.
- Measure the chatbots performance and the client satisfaction, constantly.
- Pay attention to analytics, in order to gain insights into customers behaviors and preferences.
Julián Eisenschlos, co-founder & chief scientist at Botmaker was here last month at our intive-FDV’s offices. He talked about an interesting trouble that usually comes out when we process natural language:
“Sometimes we have to figure out what the person wants. If they say ‘turn off the light’, we need to know which light they want to be off”.
He was talking about the difficulty of identifying parameters, and he gave us some guidance:
- Search for the minimum semantic unit within a singular phrase.
- Ask which role does each semantic unit play. Subjects, objects and the verbs that connect them allow us to correctly interpret the meaning of the sentence and act accordingly.
- We can use “language models”, a toolbox that provides you with more probable and less probable interpretations of the user command (in face of ambiguity).
Bots are far away from AI virtual assistants like Alexa, Siri or Cortana, and of course, they are still a long way from human touch. But deployed wisely they can be very helpful when it comes to provide contextual data, keep a user engaged and improve the experience of interacting remotely with a brand. We just have to keep working on getting the minimal nonhuman friction possible.