What are the types of chatbots?

Types of chatbots can be roughly divided into scenario type (rule-based type) and AI type. Let’s take a look at each feature.

Scenario-type (rule-based) chatbot

Chatbots without AI are called scenario-based or rule-based. Scenario type is a method to follow the scenario prepared in advance according to the choice selected by the user and return the answer.

The advantage is that users can easily obtain answers by simply answering the questions displayed on the screen. In addition, since the initial cost and monthly cost can be kept lower than the AI-equipped type that will be introduced later , it is also advantageous for companies that have introduced chatbots.

However, since it is necessary to create a scenario from questions to answers, it may take time to introduce if there are many expected questions and answers.

AI type

The AI ​​type is a mechanism in which AI analyzes what the user wrote in the chat and returns an answer. Unlike the scenario type, the advantage is that it does not take time to introduce because it is sufficient to incorporate AI into an existing site.

Since it can respond flexibly to various questions, it is useful in cases where a wide range of inquiries are expected . However, the disadvantage is that the cost is higher than the scenario type, and the initial introduction cost is also high.

In addition, AI type can be broadly divided into “machine learning type” and “original AI type”.

• Machine learning type: AI learns rules and patterns from large amounts of data. Accuracy is improved by increasing the number of data to be registered and the number of trials of conversation with users.
• Original AI type: AI performs machine learning by itself. Management is easy because there is no need to provide data to AI or modify FAQs.

In the machine learning type, it is necessary for the administrator to prepare a lot of data and let the AI ​​learn in order to operate it accurately.

On the other hand, in the case of the original AI type, it can be said that it is easy to manage because the AI ​​itself makes corrections and improvements based on the data it acquires. It costs more to introduce than the machine learning type, but if you want to reduce the operational burden, we recommend introducing the original AI type.

If you want to reduce the introduction cost, one way is to develop a chatbot in-house. Please refer to this article for details on how to create a chatbot.

The types of chatbots can be divided not only by the presence or absence of AI, but also by purpose. Here, we will introduce four types of chatbots for different purposes.

1. FAQ type chatbot

An FAQ chatbot is a tool that combines an FAQ system and a chatbot. It is possible to automatically answer user questions by combining a chatbot with the inquiry function of an EC site.

Since the chatbot answers questions from users instead, the cost of responding to inquiries on the operation side can be reduced . In addition, from the user’s point of view, it is easy to find out questions about products and services, which is advantageous for both the user and the operation side.

Both scenario type and AI type are used for FAQ type chatbot. It is common to use the scenario type when there are few types of inquiries, and the AI ​​type when there are many types of inquiries.

2. Substitute processing chatbot

Substitute processing type chatbot is a tool that performs subsequent processing on behalf of the user according to the content entered by the user. Chatbots will take over complex tasks and tasks that are prone to human error.

For example, on the site of a travel agency, users can follow the instructions of a chatbot and simply enter the date, time and place they want to visit, and the function will act as an agent for them to book accommodation plans.

In addition, it is a chatbot that can be used in scenes that require a lot of processing, such as restaurant reservations and mail redelivery, with inquiries from various users . Scenario-type chatbots are used because chatbots take over the processes that can be automated.

3. Distributed chatbot

A delivery-type chatbot is a scenario-type chatbot that automatically transmits information at a set time instead of interacting with users. Distribute using SNS and chat tools to deliver product information to users and guide services .

It can be used for a lot of sales, and there is no need for people to directly explain products, so if you introduce it, you will be able to reduce the cost of sales.

4. Conversational Chatbot

A small talk chatbot is a chatbot that keeps the conversation going, not for the purpose of the user. Keeping a casual conversation going with your users can help improve your branding and customer loyalty .

If you can introduce your company’s products skillfully through small talk, you will be able to create opportunities for users to naturally learn about your products.

A chatbot is a tool for users to enjoy conversations, so it is necessary to adopt an AI type.

Chatbots can also be classified according to how they operate. Here, we will introduce three types of chatbots by mechanism.

1. Choice type

A choice type is a mechanism in which the chatbot asks the user what they want to know and returns an answer according to the answer. The user selects a choice displayed on the screen and returns the correct answer. It is often embedded in scenario-type chatbots, and it is characterized by low introduction costs because it does not require complicated processing .

However, it may not be suitable for resolving different types of inquiries, as it requires creating a pattern of responses to selections. Since it is necessary to prepare an answer for each expected inquiry, it will take time to introduce it if there are many questions.

2. Dictionary type

The dictionary type is a mechanism in which [keywords] and corresponding [answers] are registered in advance, and responses are returned according to the keywords entered by the user. For example, if the user enters the keyword [Login], answer [How to log in] or [How to change login information].

It is mainly built into AI type, and AI understands the input words and returns answers that match the keywords in them. It is often used for systems on sites with a wide variety of inquiries. However, it takes time to introduce it because it is necessary to prepare the expected keywords and answers in advance .

There is also a [choice & dictionary type] that presents choices at first and asks questions if the answer is not found. [Choice & Dictionary type] can respond flexibly to various questions, but on the other hand, you have to create both keywords and answers for the choices and questions.

3. Log type

In the log type, when a user asks a question to the chatbot, AI analyzes the accumulated interactions with other users and returns an answer. It is a function that AI learns the intention of the question from the interaction with the user so far and generates the answer like talking to a person.

It is a mechanism that is mainly adopted in the original AI type, and the more it is operated, the more data is accumulated, so the accuracy of the answer will increase .

However, the disadvantage is that the introduction cost is high and the quality is greatly affected by the training data that is input at the time of introduction. Teaching data is data that serves as a basis for making decisions when AI is activated.

If the quality of this training data is low, the accuracy of responses will decrease. You can read more about how chatbots work in this article.

When introducing a chatbot, you may be wondering which type to choose. It is recommended to choose the type of chatbot to be introduced according to the purpose of introduction and the number of inquiries.

Here, we will explain in detail how to choose the type of chatbot to introduce. Also, the differences between chatbot products are explained in detail in another article, so please refer to that as well.

Select functions according to the purpose of introduction

When choosing the type of chatbot, it is important to first choose the functions that match the purpose of the introduction. For example, let’s choose:

• I want to make inquiries more efficient → FAQ type
• We want to reduce human errors in complex processing → processing substitution type
• I want various people to know the product widely → distribution type
• I want to have close contact with customers → delivery type

By clarifying the purpose of introducing the chatbot, it is possible to determine the functions to be adopted .

Choose with or without AI according to the number of inquiries

After choosing the function, decide whether to use AI. It is a good idea to judge whether AI is necessary based on the number of inquiries. Specifically, we recommend the following criteria.

• 50 or more inquiries: AI type
• Less than 50 inquiries: Scenario type

Scenario creation becomes complicated when the number of inquiries exceeds 50, so consider introducing an AI type .

Choose according to initial cost and running cost

Chatbots are expensive to introduce and operate for each function, so it is also important to choose based on how much you can spend. Especially in the case of AI type, not only is the product itself expensive, but it also needs to be trained regularly, so the operation cost will increase .

Please refer to the following for the cost of the scenario type and AI type.

[Scenario type]

• Initial introduction cost: 50,000 yen to 500,000 yen
• Operation cost: 1,500 yen to 30,000 yen

[AI type]

• Initial introduction cost: 200,000 yen to 1,000,000 yen
• Operating cost: 300,000 yen to 1 million yen

If there are many types of questions, but it is difficult to introduce AI type due to budget, there is also a method of responding to frequently asked questions in scenario type. Let’s make good use of chatbots according to expenses and budgets.

Chatbots have strengths and weaknesses depending on the type, so it is important to choose the one that suits your purpose when introducing it. If you can choose a chatbot that matches your purpose, you will be able to realize the benefits of streamlining your work and creating more points of contact with your customers .