Nowadays, the term chatbot has become an umbrella term, as there are so many software that have this name. However, from one chatbot to another, there are significant differences. I will therefore help you to differentiate them through this article. In the car!
The first chatbot is called Eliza
First, let’s get back to basics. A chatbot is a software that tries to have a conversation with a person for a few minutes. Its objective is to make it feel like it’s talking to a human. The first chatbot that saw the light of day was called Eliza and was created by an MIT professor, Joseph Weizenbaum in 1964. Well, it’s the most famous one but it happens that there is still an older chatbot dating from 1780, I let you read our article about it to learn more.
Today, chatbots do not necessarily try to pretend to be a human, they essentially try to provide a service by answering our questions, by helping us to look for information or by helping you to collect information (for your CRM for example).
4 different types of chatbots
1) The conversational chatbot to imitate a conversation
The first type of chatbot we will discuss isthe conversational chatbot. This is the best known: the user connects, the chatbot reacts and proposes a path within a decision tree. It is a priori the “Ferrari” of chatbots: it is impressive but it costs a fortune. The most efficient on a straight line but it is not adapted on a winding road.
This means that when the user’s question is foreseen in the decision tree, no worries, he will get his answer quickly (straight road) but if it is not the case, no solution is foreseen (winding road)… “sorry I didn’t understand but I learn every day” (same answer 6 months later)
As mentioned above, it is expensive to deploy and to maintain, and each evolution in the answers requires a lot of back and forth with the IT department, making the update slower. All this to get a 40-60% automation rate of the questions because the chatbot only handles the most frequent cases and is not well designed to easily handle the conversations that did not succeed.
The investment is therefore not very profitable and poses a serious maintenance problem. Keep reading to find out which chatbot can counteract these drawbacks.
2) Fill your CRM with the formbot (CRM-bot)
In the same caliber of chatbot with predefined fields, we have the formbot. Beware, it is not intended to answer users’ questions, but to retrieve visitors’ data in a more pleasant way through a conversation that actually fills a simple form.
It is indeed more pleasant because the formbot gives the feeling that a conversation is going on. This is a non-conversational CRM bot, with a commercial targeting objective.
If it is effective it has a major drawback, visitors do not like to answer 10 questions in a row without being able to ask questions or search for information. The number of leads collected is 2 to 4 times lower because of visitors abandoning conversations.
If you really want to use your chatbot for sales I advise you to read
on this subject.
3) Use your documents in one click with the opentext chatbot
The improved version is the chatbot with a semantic search. It searches beyond the indicated keywords, it analyzes the context of the query, the keywords and the order in which they were entered in order to capture the user’s intention, which leads to more relevant results.
Whatever the type of search, opentext or semantic, the problems remain the same, the information that is documented represents only 20% of the knowledge and these documents are rarely updated. Nevertheless you exploit your documentation!
4) The FAQbot or dynamic FAQ: more flexible and more efficient
Finally, the last type of chatbot is the FAQ bot based on a dynamic FAQ and which, like the chatbot with semantic search, captures intentions. It is the equivalent of a 4X4, since it can handle any type of road. When the user’s question is in the database it will provide the answer directly (right route). If it is not there, he goes to an expert to get the answer (winding road). The answer is sent by e-mail to the user who asked it and is recorded in the software.
Therefore, as time goes by, the number of questions capitalized increases and the less your teams need to intervene; at TeamBrain the average rate of automation of answers to questions is 95%.
Like the Conversational Chatbot, the FAQ bot can be interactive, and thus exchange a few questions with the user to better understand his intention and give him the most appropriate answer possible.
And that’s not all, the setup is fast compared to conversational chatbots; in 1 hour at TeamBrain your FAQ bot can be operational. We start with a database of questions collected by your support team and it will naturally be enriched by new questions from users. For more information on the technology behind the FAQ bot I let you consult
our page on this subject
Comparison of the different types of chatbots
|Automatisation (réponses aux questions)||40-60%||100%||20-60%||95%|
|Usage par les visiteurs||Moyen||Faible||Elevé||Elevé|
In the end, each of the chatbots presented in this article can be used in specific contexts, you just have to define which objective you want to achieve. Now you are pros of chatbots, here is an article to know all
the different types of dynamic FAQs