13 examples of chatbot failures | Detailed countermeasures

Here are some of the top reasons why chatbot deployments fail:

Not selecting the right chatbot for the purpose

There are various types of chatbots, and some tools have specific uses, such as internal use and customer use. If you choose to introduce a chatbot without clarifying the purpose of introducing it, it will lead to failure such as not being able to use it effectively, so be careful.

Required data is not reflected

The chatbot answers questions based on preset FAQs and scenarios. If the data required by the user is not reflected in the chatbot, it will not be able to properly answer the user’s question and the usage rate will decrease. If you say, “I’ve introduced a chatbot, but it’s not being used much,” there may be a lack of data .

Neglecting post-implementation maintenance

Another reason for failure is that chatbots are not regularly analyzed and maintained after deployment . Neglecting analysis and maintenance will lead to problems such as outdated FAQ information and lack of necessary information.

In this chapter, we will introduce common chatbot failure examples and countermeasures. In order to successfully introduce a chatbot, it is important to understand examples of failures and countermeasures.

1. User engagement is low

This is a case where even though a chatbot has been introduced, the user rate is low and the chatbot is not being used effectively. The main reasons are:

• Users are unaware of the existence of chatbots
• Users don’t know how to use chatbots

Since the countermeasures differ depending on the cause, let’s first analyze why chatbots are not being used.

Countermeasure 1: Make the existence of chatbots known

In order to increase the usage rate of chatbots, it is necessary to let users know that they have been introduced. For customers, it is effective to announce the existence of the chatbot through SNS, etc. at the time of release, and through web advertisements and e-mail magazines . If it is for internal use, decide on a method of dissemination, such as a meeting or mass email.

Countermeasure method 2: Create a lead to the chatbot

Install a chatbot in a place where users can immediately use it, such as an EC site or an internal portal site. If you are also using paper advertising, it would be a good idea to attach a QR code to a leaflet or pamphlet to guide you to a chatbot .

Countermeasure 3: Change chatbot design

Change the design so that the chatbot stands out. Aim for visuals that make it easy for your customers to notice your chatbot, such as using a larger banner or stating “contact us here” .

Countermeasure method 4: Inform how to use

In order to make it easier for users to use it, it is a good idea to inform them how to use it when announcing the release of the chatbot. There is also a method of using images and illustrations to convey how to use the chatbot immediately after the user activates the chatbot . It is important to devise ways to make users understand how to use it.

2. The chatbot itself is difficult to use

As an example of chatbot failure, there are two points that are considered difficult to use from the following two points of view.

• Difficult to operate
• Difficult to enter text

If it is not easy for anyone to handle, user satisfaction and utilization will drop.

Countermeasure method: Check whether the UI/UX is easy to use before introduction

Before introducing a chatbot, it is important to check whether the UI and UX are easy to use even for first-time users. For example, the ability to suggest information requested by the user leads to usability. Free trials are available for some chatbots, so it would be a good idea to try them out before introducing them.

3. AI Chatbots Have Low Accuracy

This is an example of a failure in which users cannot obtain satisfactory answers due to the low accuracy of AI chatbots, causing a decline in usage rates and customer satisfaction. AI-based chatbots cannot be used effectively immediately after introduction, and require a learning period to improve the accuracy of conversations and responses .

Countermeasure method: Establish a study period

If you are going to introduce an AI-type chatbot, set a learning period of at least several hours a week to improve accuracy. Even after operation, it is necessary to perform periodic tuning to ensure that questions are being answered correctly.

4. Low Resolution Rate

This is a failure example that the resolution rate is low even if you use a chatbot. The main reasons are as follows.

• User-required FAQs are not accumulated
• Answers are long and difficult to understand

If the chatbot can’t solve the information that the user wants to know, satisfaction will decrease.

Countermeasure method 1: Reflect necessary FAQ for users

As a countermeasure, use the following method to investigate the needs of the target users of chatbots.

• Collect and analyze past inquiries
• Think from the user’s point of view
• conduct a survey

Analyze the survey results and reflect the FAQs that users need in your chatbot.

Countermeasure 2: Register search keywords in FAQ

Most chatbots are designed to respond according to the search keywords included in the user’s question. Registering search keywords in the FAQ will improve search accuracy and improve the resolution rate.

For example, if a user inputs “I want to cancel my membership” in the chatbot, FAQs containing the keyword “withdrawal” will be displayed as answers, but FAQs with the keyword “cancellation” will not be extracted. It is important to register many keywords that are likely to be searched.

Countermeasure 3: Prepare concise answers

Review the FAQ and edit long sentences and difficult-to-understand expressions into concise sentences. It is also effective to guide users to pages with images and illustrations inserted along with the text . If you are an AI type, take measures such as choosing a service with high natural language processing ability and setting a learning period to improve accuracy.

5. Accumulating too much information increased the difficulty of implementation

Accumulating a large amount of information in a chatbot increases the difficulty of implementation. Especially when introducing a chatbot for internal use, be careful as the amount of information tends to be large if the information you want to include is different, such as customer service representatives and management.

Countermeasure: Gradually increase the amount of information

Instead of including a lot of information from the beginning, choose what to register in the chatbot . For example, if it is for customers, you can narrow down the amount of information by limiting it to “Frequently Asked Questions”. It is effective to gradually increase the number of FAQs while performing analysis during operation. It is also important to have a unified understanding within the company.

6. I get asked a lot of difficult questions

Chatbots are not suitable for answering difficult questions or long sentences. Users can be dissatisfied when they don’t get the answers they expect.

Countermeasure method: Decide the business scope of the chatbot

Leave common questions to chatbots and difficult questions to customer support staff. If the user’s question cannot be resolved, it is also necessary to have a flexible response, such as connecting the chatbot to a manned chat or displaying the customer support phone number.

7. Can’t spend time on operations

It is important to improve the accuracy of chatbots through regular analysis and maintenance, rather than just introducing them. If the operation time is insufficient, user needs cannot be grasped, and the utilization rate will decrease.

Countermeasure method 1: Establish an operation system

Decide who will be in charge of analysis and maintenance of the chatbot before operation. Divide the work into multiple teams if necessary.

Countermeasure 2: Request consulting

If the operator does not have the time, it may be a good idea to hire a consultant. There are also services that support from introduction to operation. Although it is expensive to operate, it is useful as a means to prevent the risk of quality deterioration due to lack of maintenance.

8. The chatbot type is not right for your company

It is a case of failure because you choose a chatbot that is not suitable for your service or purpose. There are many chatbots out there, and each type has different functions and features. If you don’t know what you want from your chatbot, you can’t make the right choice.

Countermeasures: Select a chatbot that suits your purpose

Compare and consider various chatbots before introduction according to your company’s service and purpose. There are two main types of chatbots:

• Rule-based type
• AI type

The rule-based type is suitable for fixed questions because it derives answers from questions prepared in advance. On the other hand, the AI ​​type is equipped with artificial intelligence, and is characterized by improving the accuracy of answers by repeatedly learning with a large amount of information and data. Therefore, it is suitable for questions that require conversation. Let’s decide which type to use according to compatibility with your company’s service and purpose.

After deciding on the type of chatbot, consider whether to use a chatbot creation tool or develop it in-house, and how to create it.

9. It didn’t lead to a decrease in the number of inquiries

It is a common failure example that the number of inquiries does not decrease because the chatbot is not maintained and the appropriate answers cannot be given, or the leads to the chatbot are poor.

Countermeasure 1: Review past questions and answers

If there is little maintenance after introduction, we will review it. Let’s make improvements such as making it possible to give appropriate answers to questions that could not be answered among the questions that existed in the past .

Countermeasure 2: Make the presence of chatbots easy to understand

If the chatbot usage rate is low in the first place, it is necessary to take measures to increase the usage rate, such as creating a lead to the chatbot and making the design stand out , as introduced earlier .

10. You didn’t need a chatbot in the first place

Even if you introduce a chatbot, you may not be able to achieve your company’s purpose. For example:

• Hundreds of possible FAQs
• The target user generation is unfamiliar with the Internet

If you have a large number of FAQs, an FAQ system may be a better choice than a chatbot. Also, if your company’s products are targeted at generations that are unfamiliar with the Internet, such as the elderly, they may not be able to use chatbots in the first place.

Countermeasure method: Clarification of purpose/user needs survey

Before introducing a chatbot, clarify the purpose of your company. If the motive is ambiguous, such as “introducing it because it seems convenient,” there is a possibility that the functions used and the operational design will fluctuate, making it impossible to use the chatbot effectively.

In order to determine whether chatbots are really necessary for users, it is also effective to conduct a needs survey through user questionnaires.

11. Not cost effective

This is a case where the initial cost and running cost were high, and the cost-effectiveness was not met. If the chatbot usage rate is low, it will not be cost-effective. In particular, the AI ​​type tends to be expensive, so consider it carefully.

How to fix it: Deploy a chatbot that fits your budget

Decide in advance the budget that you can spend on chatbots at your company. It is a good idea to judge whether the cost-effectiveness can be demonstrated and the compatibility with the price based on the current number of inquiries.

12. Didn’t take time to implement

Insufficient time to prepare for the introduction of chatbots may lead to failure. For example, there are cases where the chatbot is selected incorrectly, or released with low answer accuracy.

Countermeasure method: Securing sufficient preparation period

By scheduling a preparation period with plenty of time, you can release a chatbot that is effective in achieving your goals. Chatbots require a lot of man-hours to introduce, such as clarifying the purpose of introduction, comparing, selecting, implementing, and testing. Although it depends on the creation method, it basically takes several months to prepare.

13. Couldn’t release because it wasn’t finished

This is a failure example of deciding to introduce a chatbot, but not being able to release it because it was not completed. The main reasons are as follows.

• It took a lot of time and effort to prepare, and I was frustrated
• Setting up a chatbot was difficult
• Couldn’t develop a chatbot in-house

How to fix it: Use a chatbot builder

Chatbot creation tools make it easy for even beginners to create chatbots. It is divided into rule-based type and AI type, and each has different functions, features, and services that can be linked, so find a tool that suits your purpose.

The ” chatbot creation tool ” provided by our company HubSpot is completely free from introduction to continuation. You don’t need any specialized knowledge because you can create and customize with no code specifications. If you want to create a chatbot easily, please consider it.

In order not to fail with chatbots, it is important to set goals and analyze them by referring to failure cases. In this chapter, I will introduce the points to avoid failures with chatbots. Please read it before installing.

Clarify your purpose

Clarify your company’s issues and objectives before introducing a chatbot. By clarifying the purpose, you can see the functions necessary for the chatbot and the budget you can spend. Depending on the purpose, a chatbot may not be suitable, such as a FAQ system, so consider carefully.

Select a chatbot that fits your purpose

Chatbot selection criteria include:

• kinds
• function
• UI/UX
• cost
• High maintainability
• Presence or absence of analysis tools
• Availability of support
• Free trial available

It is also a good idea to refer to the chatbot of the service that you feel is convenient after actually using it in a free trial.

Decide on the business scope of the chatbot

Chatbots are not suitable for long sentences or difficult questions. Therefore, we decide the scope of work on the premise that we cannot entrust all FAQs.

For example, “Frequently Asked Questions” can be divided into chatbots, and others can be manned for customer support . By sharing the scope of business of chatbots within the company, it is possible to prevent gaps in internal recognition of chatbots.

Test operation before release

In order to prevent troubles after starting chatbot operation, be sure to test before release. We will check the operability of the tool and the accuracy of the answers, and make improvements as necessary. The method of conducting a questionnaire survey by having users actually use it is also effective in discovering areas for improvement.

put in place an operating system

After introducing a chatbot, regular analysis and maintenance are required to prevent quality deterioration. It is important to set up a chatbot operation system, such as having multiple people in charge.

Set KPIs

In order to analyze whether the purpose of the chatbot is being achieved, KPI (an indicator for measuring the degree of achievement of the purpose) is set. For example, in the case of improving the efficiency of inquiry operations, the following items are KPIs.

• Startup count/startup rate
• Number of responses/response rate
• resolution rate

Analyzing the numbers will help you understand what maintenance is required.

perform regular maintenance

We will regularly maintain the FAQ, such as improving old information and adding FAQs that users need. Failure to obtain the necessary information from the chatbot will lead to a decline in usage, so maintenance is important to prevent quality deterioration.

The main reason why chatbot introduction fails is that the chatbot has not been selected and improved after operation. In order to achieve the purpose of your company, it is important to carefully compare and consider before introducing it and introduce a chatbot that is suitable for your company.

Also, the introduction of chatbots is not the end. Operation and improvement are important points, so let’s prepare an operation system. Use this article as a reference to find a chatbot that will help you solve your company’s or your customers’ problems.