5 common misunderstandings in chatbot design summarized from 100+ customer cases. Don’t make them!

Recently, various AI conversation tools have become popular, including ChatGPT overseas and Wenxinyiyan under Baidu in China. As the trend of artificial intelligence rises, more and more people are beginning to move AI tools into their own products and websites.

SaleSmartly connected to ChatGPT as early as February. In addition, the automated chatbot is also a highlight feature of SaleSmartly .

There is no need to say much about the advantages of chatbots: they save manpower and improve service efficiency.

However, in some settings, we have summarized more than 100 customer cases and found that there are several common misunderstandings. Such operations may not only fail to play the role of chatbots, but also cause trouble to customers! Pay attention!

Therefore, this article will cite five common misunderstandings to help everyone review and understand the solutions to these design errors.

Continue reading below👇


Myth 1: Robot conversations only have questions, no options

Regardless of whether you set up the robot for sales or to answer customer questions, be careful not to put your customers in a conversation situation where they have no choice!

Although robots can save a lot of trouble for real people, scripts that only "ask questions" to customers without providing "reference options" can easily cause fans to interrupt the conversation and make the robot lose its purpose of solving basic problems. As shown in the figure below, such unclear call to action makes it difficult for customers to take the initiative to engage in conversation, and they may even jump out of the conversation window!


Solution: The robot script design must have a “button” to guide customers!

If your robot’s purpose is to sell goods, you can try to classify them by item. As shown in the following figure: While asking customers, you also design buttons for male and female products to help customers find the products they need step by step.


Misconception 2: Lengthy information dialogue without segmentation

The response is long and boring, which makes it hard for the customer to read. When designing dialogue, please pay attention to: do not have too many words!

In order to shorten the conversation process, some users often reply with hundreds of words, which makes it difficult for customers to quickly grasp the key points. Customers have to manually scroll to the beginning of the message to know what the topic is. As shown in the following figure, the user uses a lot of words to answer the question "What is SaleSmartly ?". The overly long message may make it difficult for customers to find the topic and cause visual fatigue.


Solution: Appropriate segmentation, good use of buttons, pictures, and external links

When arranging information, be sure to divide it into sections! If there is too much information, you can use "buttons" to separate the topics, or link the buttons to related web pages, etc. You can also simply simplify the content into pictures to present it.

Looking back at the misunderstanding demonstration in the picture above, if the content is divided into three paragraph buttons, the theme will be clear at a glance and the picture will become refreshing. Subscribers can also choose the paragraph they are interested in and click to read.


Misconception 3: The script design is too confusing and the dialogue becomes increasingly complicated

In addition to the design and arrangement of information content, another very important thing is: don’t make the robot’s conversation structure too confusing!

Many users design chatbots based on what they think of, lacking pre-set scripts, and the button menus are prone to duplicate content, or the robot goes around in circles in an infinite loop. Not only does the user experience lose its smoothness, but it also causes customers to not be able to find the content they really want to see!

For example, the topic of "promotions" keeps appearing in the options. Customers will see it whether they are in the main menu or asking customer service questions about promotions/invitation rebates. This dialogue design is not only confusing, but also makes it easy for customers to forget where they clicked and what they want to do!


Solution: Design the conversation script first, then set up the robot!

It is recommended that you establish a framework before setting up the robot! First, systematically classify all the content to be communicated with customers, and then draft a dialogue script. This way, when actually setting up the chatbot, there will be no mistakes in connection, irrelevant answers, or confusion in the dialogue!

Allow the conversation to develop naturally. Explain what is happening and where the customer is going. Make suggestions for what they might do next and gradually reveal features and elements they can take advantage of.

◎Small reminder

At the end of each branch conversation, remember to let the customer return to the main trunk or the previous level of content, otherwise the conversation will be interrupted and cannot continue!


Myth 4: Let robots pretend to be real people, and customers can’t tell the difference

In addition to the script content and structure, if you want to create a better customer experience, you should pay attention to the following: Be careful when letting robots pretend to be real people to have conversations!

In order to make the chatbot interactive with customers, some beginners may think that the “humanized” tone design is very important! However, if this “real person” tone is often used to respond, it will easily confuse fans, who will not be able to distinguish whether it is a real person or a robot’s response, and may even be dissatisfied with the robot’s response because they mistake it for a real person.


Solution: Avoid over-personification and provide an option to switch to “real customer service”

During a conversation, it is important to let customers clearly distinguish the "object & situation" at the moment! If the problem needs to be solved by contacting a real person, as long as there is a clear message in the conversation that "real customer service" is switched , customers can immediately identify the person they are talking to and wait for the assistance of the customer service staff.


Misconception 5: Expecting chatbots to solve all problems

Chatbots can certainly solve a certain proportion of common and repetitive problems, but everyone should remember that chatbots cannot solve all problems 100% of the time!

Chatbots are not a panacea! If your goal in designing a chatbot is to “replace all customer service staff,” you’ll probably be disappointed. Even if the chatbot’s content is well designed, it still can’t completely replace real people, and there won’t be a website or product that only relies on a robot to operate online.

We must make it clear that the correct way to use chatbots is to design "robot scripts" to sort and filter most of the questions so that customers can get answers from the robot as soon as possible; the remaining questions that cannot be handled will be transferred to real customer service for assistance.

Firstly, customer service staff no longer have to answer repetitive questions, and secondly, operators can reduce customer service manpower and reduce costs!


Five common misunderstandings

  • Don’t just have dialogues, the robot script design must have “buttons” to guide customers!
  • Don’t design reply messages that are too long. Make good use of buttons, pictures, and external links to “segment” the content!
  • Don’t make the conversation more complicated. Design the conversation script first, then set up the robot!
  • Don’t let robots pretend to be real people, let customers know the difference and include a “real customer service” option!
  • Don’t expect that a fan page can be run with just robots; let real people solve difficult problems!


The above are just some tips, but any conversational apps you build need to be based on user research during their initial definition and scope, and be tested, and tested over and over again, to ensure they work in a beneficial way, rather than simply being annoying, repetitive, or a hindrance .

I hope that through the organization of this article, everyone can have a better understanding of content construction, and after referring to the suggestions in this article, you can immediately start building your own chatbot!

If you’re interested in learning more about designing chatbot conversations, sign up for a free trial of SaleSmartly !

Share this Article
Last modified: 2024-05-09Powered by