When it comes to chatbots for business, many imagine primitive, button-based bots used to order a pizza or to send a sticker. This field, however, is much larger. A chatbot is a dialogue interface that can be integrated into software, websites and even messengers.
There is a big amount of misconceptions around chatbots used in call centers, customer service and customer service. Stanislav Ashmanov, Nanosemantics CEO, has addressed the most common ones in his recent speech at Russian Tech Week. This article gives a short summary of his report.
Myth 1: The market is not ready
Contrary to that opinion, only the laziest of companies do not use chatbots. Most of the large businesses are faced with contact center overload, and they are finding it increasingly harder to deal with. Every day more and more companies come to the realization that introducing a chatbot will allow for optimization of the most important resources – time and money. Based on our own experience, we can confirm that the market is indeed ready and the interest keeps growing.
Myth 2: Everyone needs a chatbot
Unfortunately for us, this is not the case. The criteria for chatbot necessity, in fact, are very straightforward.
First, if you don’t have to deal with a high number of clients, a chatbot will not save you any money. From a cost efficiency point of view, it would be cheaper to hire some interns to handle the queries.
Second, if you receive mostly nontypical requests that are hard to predict ahead of time, the best chatbot can do is redirect the customer to the customer support officer.
Third, the dialogue with the client may be absent altogether. There may be no communication even in cases with a large client base.
Fourth, not everything can be accomplished with text or voice. For instance, in cases of vast item catalogue, the chatbot may respond with a list of 20 items to a request of a single specific one. Navigating through such a list would be inconvenient for the customer.
Myth 3: Chatbots are useless
There is an opinion that chatbots are overhyped and inefficient. While hype is definitely there, it has been declining as of late. As mentioned before, chatbots save significant amount of money for large companies with vast incoming traffic. Clients benefit from this as well, as a conversation with a chatbot is shorter and the response time is reduced to nothing since the bot is available 24/7.
Myth 4: People are uncomfortable talking to chatbots
This statement is only true for primitive chatbots that fail to resolve client issues. If the chatbot is implemented well, people are eager to chat with it and enjoy the communication.
Myth 5: A chatbot is a set of buttons in a messenger
Chatbot functionality is much wider and depends on its type. For example, our chatbots that are rule-based include the following features:
- Keeping the conversation going
- Customer system integration
- Dialog analysis
- Connecting to a customer support officer
- Voice recognition
Myth 6: Chatbots can do anything
The bot cannot answer all of the questions. It can only answer within the frame of what it was taught, what is already present in its knowledge base.
Myth 7: Creating a chatbot is easy
There are open platforms which can be freely used by anyone to create a basic chatbot. Those of our customers who do, come back to us a year later due to their chatbot lacking required functionality. Creation of a more advanced chatbot requires more time as well as professional input from data engineers, developers, and computational linguists.
Myth 8: Machine learning is everything
There is a recurrent idea in the field of chatbots, “Why don’t we fetch the recordings of call center conversations and train a neural network? We’ll get a superchatbot that will continue learning from new dialogues!” It doesn’t work that way. First, the database may not exist to begin with. Second, the database would require labeling and cleaning, due to a large amount of personal data (names, telephones) contained within. Third, such chatbot will not be able to competently follow the topic of conversation and consistently give relevant answers.
Myth 9: A chatbot must pretend to be human
Users must be informed that they are talking to a chatbot straight away. Otherwise, once this becomes known, anyone would become uncomfortable and would think they’ve been lied to. As a rule, people are much more patient if they are aware that they are talking to a emotionless machine as opposed to a human employee.
You can watch the full speech on our Youtube channel.