Your website is the first impression that many of your customers will have of your business. And while there are plenty of ways to optimize your site for maximum engagement, one of the most effective is to add a chatbot. But if you’re not a natural wordsmith, the thought of writing smooth responses can be daunting (and time-consuming). So, how can you create chatbot conversation flows with responses that are engaging, informative, and effective?

When you know a lot about your customers but are short on time, the simplest solution might be to use one of the best AI tools for marketing writing—like ChatGPT—to whip up fitting responses. If you’ve never used a virtual writing assistant before, figuring out how to use this type of resource might feel intimidating. But guess what? Automated, smart writing tools are here to help, not hinder.

Some of the topics covered in this article include:

  • Three popular types of chatbots
  • What is a chatbot conversation flow 
  • Benefits of using AI tools to write chatbot responses
  • Prompting tips for using ChatGPT to write great, on-brand chatbot responses
  • How to write chatbot conversation flows (with examples)
  • Mistakes to watch out for when using AI tools for creating chatbot responses

Keep reading to learn how to talk to ChatGPT’s bot so that the conversational scripts you prepare for your website chatbot enhance the customer journey. If you follow our tips, you’ll save time and just might end up looking like the new office hero for figuring out the nuances that go into successful chatbot conversation marketing. Let’s begin.

Popular types of chatbots

Before you start writing responses for your website’s chatbot, it’s important to understand the differences between the three common options available, which include: 

  • Simple chatbots
  • Smart chatbots
  • Hybrid chatbots

While the names might seem self-explanatory, the differences between the three can have a big impact on the overall user experience. Knowing which type of chatbot is best suited for your website’s needs can help you create conversation flows that are both effective and engaging. 

So, let’s take a closer look at each type and what sets them apart.

Simple chatbots

A simple chatbot, also called a rules-based chatbot, follows a pre-determined decision tree or flowchart to respond to user inputs. 

Simple chatbots are typically rule-based and rely on pre-set answers to user inquiries, making them relatively straightforward to build and implement. They are best suited for handling basic queries and frequently asked questions, such as providing business hours, directing users to a website page, or providing simple information about a product or service. 

Due to their limited functionality, they are generally less expensive and easier to develop than more advanced chatbots. However, they may lack the ability to understand natural language and may not be able to provide more complex or personalized responses to users.

Examples of questions that simple chatbots can answer

This type of simple chatbot is great for handling basic inquiries, so your team is free to focus on more complex customer support issues. 

These chatbots work by following pre-defined rules and responding to specific keywords or phrases, making them an efficient and cost-effective addition to your team. 

Stats on chatbots and customer service

According to researcher hi-tech communications expert Nick Maynard, more and more businesses are implementing chatbots that work 24/7 to drive cost-savings: “By embracing automated customer service with chatbots, retailers can act in a more flexible and efficient way.”

Here are more fast facts concerning how website users feel about using chatbots:

  • 69% of customers in 2022 said they would be willing to interact with a website chatbot for simple issues, a 23% increase from 2021. (Zendesk 2022 growth areas report)
  • 54% of customers say that their biggest frustration with chatbots is the number of questions they must answer before being transferred to a human agent. (Zendesk 2022 opportunity report)
  • A majority of millennials prefer live chat for customer service over every other communication channel. (Comm100)
  • The healthcare chatbot market is predicted to reach $944.65 million by 2032 from $230.28 million in 2023. (Precedence Research)
  • 90% of Americans use customer service as a factor in deciding whether or not to do business with a company. (Microsoft)
  • 89% of companies with “significantly above average”customer experiences perform better financially than their competitors. (Qualtrics XM Institute)
  • 65% of Americans would pay 5% more for a product or service if they knew they would receive outstanding customer service. (Emplifi)

They provide you with an ideal way to answer routine questions through your website concerning a wide range of topics, such as:

  • Hours of operation
  • Product or services
  • Billing, payment methods, and refund policies
  • Troubleshooting technical issues
  • FAQs
  • etc.

For instance, a customer support chatbot for an online store might be set up to answer questions about shipping, returns, and products for online shoppers who need quick, accurate information.

Smart chatbots

A smart chatbot, also called an AI-powered chatbot, is an advanced type of chatbot that uses machine learning and artificial intelligence to understand and interpret natural language input from users. It can learn from past interactions and adapt its responses on the fly accordingly.

Think of it like having a super-smart salesperson available 24/7 to assist your customers.

Smart chatbots use advanced algorithms to analyze data and provide personalized responses based on the user’s intent and context. They are able to understand and handle more complex queries and can provide more intelligent and tailored responses than simple chatbots. 

They can also be integrated with other systems and applications, such as CRM software or e-commerce platforms, to provide a seamless user experience.

Hybrid chatbots

What if you could combine the efficiency of rule-based chatbots with the intelligence of AI-powered chatbots? Hybrid chatbots do just that, providing quick answers to routine questions while also offering personalized product recommendations and support. It’s like having the best of both worlds at your fingertips.

What is a chatbot conversation flow?

A chatbot conversation flow is the backbone of your chatbot’s interactions with users. It’s the step-by-step structure that guides users through a conversation, making sure they get the information they need while feeling heard and understood. 

Without smooth chatbot conversation flows, users might get lost, frustrated, or worse—leave your website altogether.

Simple chatbot conversation flow example

Benefits of using AI tools to write chatbot responses

Think of an AI marketing bot as an eager intern who is ready to lend a helping “hand” with your chatbot conversation design project. It requires clear instructions from you (and it may not get it right on the first try), but the more you put in, the better results you’ll get that you can incorporate into your chatbot conversation flows.

ChatGPT isn’t sophisticated enough to do the entire job for you (yet), but its savvy helper bot can get the job started with your expert guidance.

Some other specific benefits of using ChatGPT for writing chatbot responses include:

  • Time-savings: Instead of manually crafting every response, ChatGPT can quickly generate numerous high-quality options, allowing you to focus on refining and perfecting the conversation flow for your chatbot.
  • Adaptability: ChatGPT’s ability to understand context enables it to generate relevant responses for a wide range of conversation scenarios, making your chatbot more versatile and useful.
  • Consistency: By providing context and guidelines to ChatGPT, you can ensure that the generated responses adhere to your brand voice, creating a cohesive experience for users.

8 steps for writing a smooth chatbot conversation flow

Whether you’re working on answers for a simple chatbot or hoping to write responses to use with a more complex hybrid chatbot, the key to a great user experience is to craft engaging, informative, and efficient responses that make your website users feel like they’re having a seamless conversation with your customer service team.

With the following eight-step guide, you’ll learn how to plan, write, and refine responses for your chatbot conversation flows.

1

Define your chatbot’s purpose, goals, and role

Before diving into the nitty-gritty, ask yourself: What do I want my chatbot to achieve? Is it here to answer questions, generate leads, or provide support? By defining its purpose, goals, and role, you’ll have a clear roadmap to follow as you create your chatbot conversation flow.

2

Identify your target audience’s needs and preferences

To create a chatbot that truly resonates with your users, you need to know who they are and what they want. What are their pain points? What language do they use? 

When you take the time to specify what are your audience’s needs and preferences, you’re armed with the information you need to craft responses for chatbot conversation flows that feel natural and engaging.

3

Create a compelling chatbot persona

Your chatbot’s persona is its voice, tone, and personality. It’s what makes your chatbot unique and helps users connect with it on a deeper level.

Think about your brand and audience: What kind of persona would resonate with them? A friendly advisor? A witty sidekick? By crafting a compelling persona, you’ll end up with a chatbot that’s more memorable and enjoyable for users.

4

Use ChatGPT to draft potential responses

Follow the guidelines below for step-by-step tips on how to talk to ChatGPT to write potential responses you can refine and add to your chatbot scripts.

5

Refine ChatGPT’s responses

ChatGPT is good, but it’s not perfect. Tweak the responses its AI bot provides to ensure every answer is effective and efficient by following these guidelines for writing great responses for your chatbot conversation flows:

  • Keep dialogue short and sweet
  • Mind the depth of your conversation
  • Don’t ask to answer irrelevant questions
  • Ditch robotic wording
  • Be adaptable to unexpected user inputs
  • Collect feedback about users’ experience

Tips (with examples) of how you can tweak your ChatGPT chatbot answers to make them sound even better

Keep dialogue short and sweet

In most cases, every user is busy. The reason they’re talking to your chatbot is because they need a piece of information quickly. This is why users universally prefer concise, easy-to-understand responses over those that are complex and meandering. Serve your users better by simplifying ChatGPT’s answers to maximize user engagement.

Example website chatbot user input—
“How long does standard shipping take?”

🤖 Initial poor ChatGPT response: “Thank you for reaching out to our customer support service. Your query is important to us. We would like to inform you that our standard shipping time is approximately 7 to 10 business days. Please take note of this information.”

😟 Why the original ChatGPT output is bad: This response is too long and overly formal. Yuck!

🤖 + ✏️ New response (edited based on ChatGPT’s suggestion): “Hi! Our standard shipping takes 7–10 business days.”

😃 Why the revised version is better: The improved example is concise, easy to understand, and uses a friendly tone.

Mind the depth of your conversation

Avoid overwhelming users with too many conversation branches. Stick to a few relevant topics and guide users through the conversation smoothly.

Example website chatbot user input—
“How much do your cosmetic skincare treatments cost?”

🤖 Initial poor ChatGPT response: “Which treatment are you interested in? We have services for treating age spots, wrinkles, sagging skin, acne, dull skin, congested pores, uneven skin tone, spider veins, unwanted hair, and more.”

Website chatbot user response: “Wrinkles”

Follow-up ChatGPT response: “Would you like to know about surgical or minimally invasive treatments for wrinkles?”

Website chatbot user response: “Minimally-invasive”

Follow-up ChatGPT response: “We have various minimally invasive treatments for wrinkles. Which one do you prefer?”

😟 Why the original ChatGPT output is bad: Oof! The conversation has too many branches, which can be annoying and time-consuming for your website chatbot user.

🤖 + ✏️ New response (edited based on ChatGPT’s suggestion): “Can you please specify which treatment you’re interested in?”

😃 Why the revised version is better: The improved example streamlines the conversation by asking for the specific product directly.

Don’t ask to answer irrelevant questions

Bombarding users with unnecessary questions isn’t going to be well-received. Limit the number of questions you ask and focus on gathering essential information.

Example website chatbot user input—
“I want to subscribe to your email newsletter.”

🤖 Initial poor ChatGPT response: “I can help you with that. First, I need your name, home phone, work phone, mailing address, and email. Also, please tell me how you found out about our email newsletter and why you want to subscribe.”

😟 Why the original ChatGPT output is bad: I get it. It would be really interesting to get all of this information, but you don’t really need it to help your website chatbot user complete the task at hand. They may get annoyed, or—in the worst scenario—they may get impatient and leave your site and go elsewhere. 

🤖 + ✏️ New response (edited based on ChatGPT’s suggestion): “Please provide your email address so I can add you to our list.”

😃 Why the revised version is better: The improved example is concise, easy to understand, and uses a friendly tone.

Ditch robotic wording

Replace impersonal phrasing with conversational, human-like responses. This can help your website’s chatbot users feel more comfortable and fosters a better connection.

Example website chatbot user input—
“Where can I find your return policy?”

🤖 Initial poor ChatGPT response: “Your query has been received. Kindly wait for a response.”

😟 Why the original ChatGPT output is bad: The response is cold and impersonal, making the user feel less connected to the chatbot.

🤖 + ✏️ New response (edited based on ChatGPT’s suggestion): “Got it! Give me a sec. I’ll find the info for you.”

😃 Why the revised version is better: The improved example uses a conversational tone, making the user feel more comfortable and fostering a better connection.

Be adaptable to unexpected user inputs

A good chatbot user experience is one that keeps the conversation progressing without hiccups. As much as possible, try to anticipate and prepare for unpredictable user responses. This forethought can help your chatbot assist the greatest number of users and handle many different situations.

Example website chatbot user input—
“I’m looking for blue shoes.”

🤖 Initial poor ChatGPT response: “Invalid input. Please rephrase your query.”

😟 Why the original ChatGPT output is bad: If your chatbot provided this response to a potential shopper then they may get frustrated that their simple question wasn’t understood.

🤖 + ✏️ New response (edited based on ChatGPT’s suggestion): “Sure! Here are some blue shoes we have available: [Product links]”

😃 Why the revised version is better: The refined answer anticipates and adapts to the website chatbot user’s input, providing a relevant response and keeping the conversation flowing smoothly.

Collect feedback about users’ experience

Encourage users to provide feedback on their experience with your chatbot. This invaluable information can help you identify areas for improvement and make necessary adjustments.

Example website chatbot user input—
“Thanks for your help.”

🤖 Initial poor ChatGPT response: “Please rate your experience with this chatbot on a scale of 1 to 5.”

😟 Why the original ChatGPT output is bad: The request for feedback is too impersonal and might not encourage the user to respond.

🤖 + ✏️ New response (edited based on ChatGPT’s suggestion): “Did I help you find what you were looking for? Your feedback helps me improve.”

😃 Why the revised version is better: The revised example uses a friendlier, more conversational tone, making the user more likely to provide actionable feedback.

6

Integrate user-friendly conversation branches

No two users on your website are the same, so your site’s chatbot should include responses appropriate for various conversation paths. 

When you integrate user-friendly conversation branches into your chatbot conversation flows, you give users the freedom to explore different topics and find the information they need.

7

Develop a plan for handling chatbot errors

Even the best chatbots make mistakes. To ensure a smooth user experience, develop a strategy for handling errors and misunderstandings. This might include providing alternative responses, asking clarifying questions, or offering to connect users with a human support agent.

And, in those cases when your website users need a human touch, consider how you will set up your chatbot to transfer the conversation to a customer service agent so that they experience as little frustration or delay as possible.

8

Regularly optimize your chatbot conversation flows

Your chatbot’s conversation flow is never truly “finished.” Continuously test, analyze, and maximize its performance to ensure it’s providing the best experience for your website users.

Examples of basic customer service info that website chatbots can provide

What do people really want from their chatbot experiences?

A good chatbot experience is one that feels like a seamless conversation with a helpful and knowledgeable friend. Here are some key factors that contribute to a great chatbot experience.

Personalization

Users expect chatbots to deliver personalized experiences tailored to their individual needs and preferences. In fact, according to a customer engagement survey by Salesforce, 88% of Americans say that the experience a business provides is as important as its services and products.

Responsiveness

Chatbots must be responsive and able to provide helpful information and guidance quickly. According to a survey by Chatbot.com, 69% of consumers prefer to use chatbots because they provide instant responses. And, according to Hubspot’s research in artificial intelligence, 40% of users don’t care if they get help from a chatbot or a human agent as long as they get the customer support services they need.

Easy-to-use interface

A survey by Ubisend found that 51% of consumers say that a chatbot that is difficult to use would discourage them from interacting with it, proving that its words aren’t the only thing that matter. 

Your chatbot’s functionality and design also “speak” to your users. Ensure its interface is easy to use and navigate, with clear prompts and helpful suggestions. 

Ultimately, a good chatbot experience is one that meets the user’s needs and preferences while providing a seamless, natural, and responsive conversation that leaves the user feeling satisfied and informed.

How to create smooth chatbot conversation flows using ChatGPT

To create a chatbot conversation flow with an AI tool acting as your virtual writing assistant, follow these steps:

1

Create a list of possible website chatbot user queries

Start by putting together a list of all of the customer service questions that you want your chatbot to answer. Some examples of this type of basic questions might include:

  • What are your shipping options and costs?
  • Do you ship outside the US?
  • What is your return policy?
  • How long does it take for my order to arrive?
  • Do you offer exchanges?
  • Where can I find more information about a specific product?
2

Provide ChatGPT with context (and answers) for each query

To ensure that ChatGPT’s virtual writing assistant helps you come up with good answers, give its bot information about your brand voice and information that they can use to answer your question. This step helps get its AI writing tool up to speed (and prevents it from making up answers that include inaccuracies).

How to train ChatGPT on your brand voice, etc.

It might seem tedious at first, but taking the time to do this initial step is what will ensure ChatGPT has everything it needs to help you and create the best possible responses for each query you ask it to write responses for. Here’s a specific example to show you the level of detail needed to get great answers.

What you might type into ChatGPT—

You are a customer support chatbot for an online store called Banana. We offer affordable, longer-lasting, high-quality alternatives to Apple products.

Our target audience includes people in lower-income brackets who want top-of-the-line, modern devices that don’t need to be replaced frequently and environmentally conscientious consumers.

Our focus is on quality, affordability, and sustainability. Our products are known for their sleek, minimalist design, durability, and user-friendly interface. We actively work to reduce our climate impact by using clean power sources and driving energy efficiency in our products and operations.

Our brand voice is modern, approachable, and innovative. 

When writing responses for our chatbot, follow these guidelines to align your responses with our brand voice and values:

  • Write short, simple sentences to convey information
  • Avoid technical jargon and complex language
  • Use positive words to convey our brand’s optimistic outlook
  • Emphasize Banana’s commitment to user experience and innovation
  • Choose friendly, personable language to connect with customers
  • Use Banana-specific terminology, such as “bPhone,” “bPad,” “BananaTV,” and “BananaBook Pro”
  • Use inclusive language that reflects our commitment to diversity and inclusivity

Our company offers standard shipping (3–5 days) for $5.99,  express shipping (1–2 business days) for $14.99, and free standard shipping for all orders over $100.

Please respond that you have received this information then wait for further instructions.

How ChatGPT will likely respond—

I have received your information. I will wait for further instructions.

The above is just one example of what you might tell ChatGPT to train it. Every company is different. The more clear the details are that you can give ChatGPT, the better your results will be.

Some “writing with ChatGPT” guides online recommend shorthand prompts, like “write in the voice of Seth Godin” (or another famous thought leader with a distinctive writing style). While this easy solution can be good in a pinch, it lacks essential specificity. Your results will need much less editing if you can give the bot more specific details that relate to what makes your brand voice distinctive.

3

Ask ChatGPT to generate response options for your chatbot

Now that you have provided ChatGPT brand voice guidelines, info about your target audience, and context for answering your first query, its bot has the source details that it needs to draft potential responses for this question about shipping costs and turnaround times for your chatbot conversation flow.

Instructions you can give ChatGPT to help you write responses for different queries your website chatbot users might ask

Here’s an example of how to talk to your virtual writing assistant to get its bot to provide you with some sample responses for a common customer service query using your brand voice guidelines (established in step two).

What you might type into ChatGPT—

Using the provided information about our company, its target customers, and our online store’s shipping options and costs, please provide multiple response options for the question that match our brand voice:

“What are your shipping options and costs?”

How ChatGPT will likely respond—

Option 1: Hi there! We offer standard shipping for $5.99, which takes 3–5 days, and express shipping for $14.99, which takes 1–2 business days. For orders over $100, we offer free standard shipping.

Option 2: Great question! Our shipping options include standard shipping (3–5 days) for $5.99 and express shipping (1–2 business days) for $14.99. If your order is over $100, you’ll get free standard shipping.

Option 3: Hello! We have a few options for shipping. Standard shipping (3–5 days) costs $5.99, and express shipping (1–2 business days) is available for $14.99. Orders over $100 are eligible for free standard shipping. Hope that helps.

Option 4: Hi! We’ve got a few different shipping options available. Standard shipping takes 3–5 days and costs $5.99, while express shipping takes 1–2 business days and costs $14.99. If your order is over $100, you can take advantage of free standard shipping.

4

Select the most appropriate answer to refine

Now you have four perfectly workable answers to edit to create a final answer for this question in your chatbot conversation flow. Refine the response you like best and tweak it to ensure its information is accurate and the response matches your chatbot persona and brand voice.

Been skimming this article? Want to learn from specific examples? Go back up for actionable tips on how to turn ChatGPT outputs into even smoother responses for your chatbot.

5

Repeat steps two through four for each query to create responses to use as part of your chatbot conversation flow

When you provide ChatGPT with the right context and input prompts, you can generate realistic and engaging responses for your chatbot. Also, by iterating through these steps for each user query, you’ll create a comprehensive conversation flow that addresses the needs of your target audience—fast!

Top 6 mistakes to watch out for when using AI tools to help you write chatbot responses

AI tools can make your job writing chatbot responses easier, but make sure your tool’s recommendations will win customers over (instead of turning them off) before you integrate them into your chatbot conversation flows. 

Crafting a smooth, engaging conversation flow that aligns with your brand voice is no easy feat. However, avoiding common mistakes can go a long way toward making sure that your chatbot conversations are productive, informative, and, ultimately, profitable. It’s especially important to pay attention to these potential pitfalls when using AI tools like ChatGPT to help you draft responses. 

Even the most advanced AI can’t always tell the difference between good and bad conversational marketing. That’s where you come in. 

By learning from the mistakes of others, you can create a chatbot conversation flow that exceeds customer expectations and helps you achieve your marketing goals. Here’s a rundown of six common mistakes (plus advice on how to avoid them).

1

Overcomplicating your chatbot conversation flow

Keep your chatbot’s conversation flow simple and user-friendly to avoid confusion and frustration.

2

Neglecting your chatbot’s persona and your company’s brand voice

A well-developed persona makes your chatbot more relatable and engaging. Take the time to create a consistent, unique personality for your chatbot that aligns with your brand voice and values.

3

Failing to provide a clear call-to-action

Users need guidance on what to do next. Include clear, concise calls-to-action throughout your chatbot conversation flow.

4

Ignoring feedback from your chatbot users

Pay attention to what your customers say and how they act when using your chatbot to evaluate its performance. Make appropriate adjustments based on these insights.

5

Relying solely on keywords for understanding chatbot user intent

Avoid only using keywords to understand what users mean when they ask your chatbot a question. Keywords can be helpful to remember when crafting your responses, but they don’t always give the full picture. 

Instead, think through each question variation and write targeted answers that show you truly understand the nuance behind what your users are trying to say. Taking this approach will help you avoid misunderstandings and provide more accurate responses.

6

Overusing industry jargon and overly technical terms

Keep your chatbot’s answers simple and accessible. Review outputs from ChatGPT and refine them to remove any unnecessary jargon or technical terms that could confuse or alienate your website’s chatbot users.

Additional chatbot script issues to avoid

When you’re revising your ChatGPT responses, here are other issues to avoid:

  • Over-personalization: While personalization can create a more engaging experience, overdoing it may come across as intrusive. Be cautious when using personal information and ensure it’s relevant and appropriate for the conversation.
  • Time-sensitive references: Including mentions of specific times of day, days of the week, or seasons can make the conversation feel more natural, but using outdated or incorrect information can confuse or frustrate users. Make sure your chatbot is updated regularly with accurate time-sensitive information.
  • Behavior-based customization: While customizing greetings based on user behavior can be helpful, it might also feel invasive if the user perceives their actions are being closely monitored. Strike a balance between being helpful and respecting user privacy.
  • Promotion mentions: Mentioning current promotions can be useful, but overemphasizing sales or discounts may annoy users who are seeking your help. Ensure that promotional information doesn’t overshadow the main purpose of the conversation.
  • Tone mismatch: Although a warm and empathetic tone can be comforting, it may not be suitable for all situations or industries. Ensure your chatbot persona’s tone of voice aligns with your brand (and the context of the conversation).

Regularly monitoring user feedback and making necessary adjustments can help you optimize your chatbot’s conversation flow for a more satisfying user experience.

AI tools aren’t the only way to take the stress out of creating chatbot conversation flows

Still feeling overwhelmed by the prospect of setting up a chatbot and writing responses that will engage your customers and meet their needs? Worried that even with help from an AI resource (like ChatGPT) that the task is too big?

Even as experienced digital marketing professionals, all of us here at TBH Creative understand that the process of creating a chatbot that helps you reach your business goals by truly connecting with (and helping!) your target audience can seem like a daunting task. Our experienced team has helped dozens of clients deploy chatbots on their websites that provide support to their users.

We can help you develop a chatbot strategy, craft a compelling chabot persona, and even write responses that align with your brand voice and values. Let’s talk about how we can help you create a chatbot experience your website users will love.