Chatbots are quickly becoming a mainstay of modern business. As interactions migrate to digital platforms, more companies are using chatbot technology to create a seamless pathway for customers to connect via artificial intelligence (AI) and real human interaction.
Whether your business is already using chatbots or simply considering whether they will have a place in your CX strategy, you’ve likely encountered at least a few confusing terms. The world of chatbots is full of specialized concepts, and you’ll need to understand some basic terminology to navigate this new landscape.
This guide will help you gain your footing and understand a few of the most common chatbot terms you’ll encounter.
What is the Difference Between Conversational AI and Chatbots?
Before we get into the terms, it’s important to understand a key concept: Conversational AI and chatbots aren’t the same, but they can be used together.
Conversational AI is a type of technology designed to enhance interactions between humans and computers. It helps computers understand human language and its many nuances in order to respond in a more “human” way.
Chatbots, in their basic form, don’t rely on conversational AI but on a set of rules that help them respond to a limited number of user queries or inputs. In this form, chatbots can help answer basic questions, but they fall short when tasked with more complex interactions. However, nowadays, many chatbots are built on conversational AI, which enables them to have much more dynamic conversations with users than they could in the past.
For businesses to truly benefit from using chatbots, they need to rely on conversational AI. And, because we see conversational AI as the future of chatbot technology, many of the terms we’ll address in this glossary deal with chatbots that incorporate it.
Essential Chatbot Terms to Know
The deeper you wade into the waters of chatbot semantics, the more you’ll encounter unfamiliar terminology. A complete glossary would include dozens, if not hundreds of terms. We’ll focus here on the most critical ones to know.
Chatbot Conversation Flows
What Are Chatbot Conversation Flows?
Conversation flows are predesigned series of responses that a chatbot employs based on common user inputs. For instance, if visitors commonly come to your website to find out the hours at various locations, you’d design a conversation flow to guide them through asking for hours, selecting a location and so on.
This concept relates to older, non-AI-based chatbots, which are usually known as flow-based or rule-based chatbots. Because these bots don’t rely on conversational AI, you need to design conversation flows for them to follow. As you can see, conversation flows can only address a limited range of issues.
What Is an Autoresponder?
Autoresponders are designed to provide quick replies to an initial contact or a specific keyword from a user. For instance, chatbot autoresponders can instantly respond with a simple, “Hi, how can I help you today?” when a user initiates a chat. Autoresponders are limited in their use but help keep a conversation moving so a customer isn’t just left waiting in a queue.
NLP, NLU and NLG
What Are NLP, NLU and NLG?
These concepts all relate to conversational AI and chatbots with deeper learning capabilities.
NLP, or Natural Language Processing, is a complex technology that allows chatbots to process human language. That means not simply understanding words and basic grammar but the nuances of language and human emotion. It can involve complex tasks like sentiment analysis, which can help bots interpret the tone of a conversation. NLU, or Natural Language Understanding, is one component of NLP that enables bots to read and interpret human language to discern user intent.
On the other end of this is NLG, or Natural Language Generation. This AI technology converts the natural data-based language of bots into human language so the bot can respond conversationally.
What Is Conversational UI/UX?
This applies the concepts of user interface (UI) and user experience (UX) to conversational technology. Rather than interacting with a static website, customers who engage with a chatbot are having a real, live conversation. Accounting for this, chatbot developers must create a chatbot environment and interface that leads to a pleasant experience for a user. Questions here revolve around ease of use and even the personality of the bot itself.
What Is Machine Learning?
Machine learning is one of the key components that distinguish conversational AI-based bots from basic, rule-based bots. This technology allows bots to learn and grow with each conversation they have with a human. Rather than repeating the same pre-programmed responses, bots with machine-learning capabilities continually develop more nuanced and complex conversational abilities.
One key metric for measuring the effectiveness of a bot’s machine learning is the learning rate. According to Forrester, a company can evaluate this by tracking how many of its bots’ interactions get escalated to an agent. If the bots are learning well, they will resolve incidents they previously escalated.†
† Forrester. “Measure The Success Of Your Conversational AI-Powered Chatbots With These Metrics.” July 29, 2020.
What is an Utterance?
An utterance is anything a user says to a chatbot. For example, if a user wants to check the balance of their banking account, they might ask, “What is my account balance?” They might also ask, “Can you show me how much money I have in my checking account?” or, “I’d like to know the current balance of my account.” Each of these sentences is an utterance.
What Is an Entity?†
Entities are key conversation variables that allow a bot to decipher user utterances and drive the conversation toward clarifying the user’s intent. A financial institution, for example, may have entities such as “checking,” “balance,” and “transfer.” Bots are then programmed with specific responses to those entities. For instance, a conversation might go something like this:
User: “What is my account balance?” [Entities: “account” and “balance”]
Bot: “Which account balance are you looking for?
User: “Checking” [A more specific entity]
Bot: “Your checking account balance is $5,000.”
Because the bot is programmed to respond to specific entities, this conversation flows smoothly and resolves quickly.
What Is Intent Recognition?
This describes a bot’s ability to go beyond basic language processing to clearly and accurately discern what the user wants to know. Can the bot take each utterance and form a cohesive understanding of what the user wants, even when those utterances may be confusing or phrased in unexpected ways? Doing so requires advanced AI technology that allows the bot to understand language in context, interpret emotions and more.
A bot’s ability — or lack thereof — to accurately decipher user intent is a key determinant in CX outcomes. It’s the difference between customer frustration and satisfaction.
What Is a Fallback?
A fallback is what happens when the bot fails to understand user intent. It usually comes in the form of a preset response, such as “I’m sorry, I didn’t understand your question.” Of course, the fallback is designed to get the conversation back on track toward resolution. If this happens enough, though, it may end in a human handover.
What Is a Human Handover?
Also known as a handoff, the human handover is the point at which the bot transfers the conversation to a human agent. It may automatically be triggered as a result of too many fallbacks, or it could be a result of the user specifically requesting a transfer. The latter scenario is called a “human fallback,” and any well-built bot will make this easy and painless for the user to request.
What Is a Trigger?
Related to the above two ideas is the concept of the “trigger.” In chatbots or similar technology, this is any input or series of responses that lead to a fallback or handoff. For instance, a specific set of human responses and a bot’s attempts to clarify may trigger a handoff. Or the bot may be programmed to respond to a simple request like “Talk to an agent.” Either way, the trigger interrupts the flow of the conversation and causes the bot to redirect it.
What Is a Conversational Channel?
A conversational channel is any medium where a bot can interact with users. It could be your website, SMS messaging, Facebook Messenger, a mobile app, or any number of other options. If there’s a chat interface, it could potentially be a conversational channel for a chatbot.
What Is a Chat Widget?
Chat widgets are ready-made chat windows you can add to your website. It serves as a host for your chatbot and an easy-to-use interface for your user. You can usually purchase these and then customize them for your needs, and they pop up to initiate conversations with your website visitors.
Don’t Navigate the World of Chatbots Alone
As you can see, there’s a learning curve when it comes to navigating the complex but compelling new realm of chatbot and conversational AI technology. Even if you’re experienced in call center IVRs, chatbots deal with a different type of human-computer interaction, and it’s important to understand those differences so you can effectively use chatbots.
At the end of the day, your customers aren’t bots; they’re people. To successfully put bots to work in your company, you need people on your team who can ensure everything is infused with a human touch. Cyara Botium can help you ensure your chatbots understand your users so you can save your human agents for the interactions that need them most.