ChatGPT and the hype around it caused us to turn our heads to chatbots world again. However, you don’t need to be an AI enthusiast or super tech savvy to understand chatbots: product people have been using chatbots since the 1970s to streamline customer interactions.
Let’s dig into the chatbots together and discover how you can benefit them!
“Chatbot” is the general name we use for software that understands user questions and generates answers to them within a conversation. It’s not a new technology; the first chatbot, called ELIZA, was developed by Joseph Weizenbaum at MIT in 1966, a chatbot for simulating a psychotherapist conversation.
After the 1970s, we started to see chatbots in commercial applications too, especially in customer service and support. Now, chatbots find themselves in new industries and new use cases everywhere.
However, we started to talk about them more with the rise of social media in the 2010s. Communicating through a digital medium gave us the option to communicate with chatbots. Nowadays, we have gone a little bit crazy about chatbots again with Open AI’s Chat-GPT.
There are several ways to categorize chatbots. For example, chatbots could be text-based or voice-based.
A majority of the time, we think of text-based chatbots. However, Siri, Alexa, and Google Assistant are also part of the chatbot revolution. Now, we are able to learn the weather by asking, “Hey, Siri! Is it going to rain?” This conversation aligns with the definition of chatbots.
But the main difference is coming from another categorization: technology. Based on the technology used to create chatbots, we can have two types:
You might guess what the difference is from their names. Rule-based chatbots facilitate conversations based on the specific keywords or phrases. Therefore, they cannot go beyond their predefined responses. For this reason, they are better suited for scenarios where there is a pattern and a predictable conversation.
For example, assume you are working at an online travel agency (OTA). If your users are contacting you about how to complete their check-in, you may create a rule-based chatbot to collect required information and handle their customer service enquiries. Rule-based chatbots are recommended when customers keep asking common questions that can follow a pattern.
On the other hand, AI-based chatbots can learn from user interactions and improve their responses over time. Their technology enables them to understand natural language and provide more personalized responses.
If you are dealing with a more complex case, the capabilities of a rule-based chatbot might not be enough for you. Let’s assume that you are working at an OTA again. This time your goal is to create a travel assistant for your users. It’s hard to create rules on this transaction because the user needs more personalized recommendations and expects you to come up with flights, accommodation, and activities according to their preferences. However, an AI-based chatbot can handle this level of complexity.
Like all product developments, there is a trade-off between these two types. It’s easier to build rule-based chatbots, but they may not cover complex scenarios. It could take longer to develop an AI-based chatbot, but it’ll enable you to cover more complex cases. The choice depends on your resources.
Like all technologies, chatbots have advantages and disadvantages. Let’s start with the positives first:
You might like what you read. I suggest taking a look at the disadvantages before making your decision. Here are the negative side of chatbots:
Regardless of the advantages and disadvantages, chatbots are exciting, and AI-based chatbots are even more exciting. So, how do UX teams benefit from them?
As product people, we are used to starting with “Why?” This is the case with designing a chatbot as well:
For example: Do you want to decrease the length of your customer support time? Do you want to decrease the amount of simple tasks that your customer support is handling? Do you want your brand to be more approachable?
Assume you want to decrease the customer support time and you saw that your customer support has an information collection pattern. If you need certain information before providing services to your users, a chatbot can handle this information gathering for you. Like we mentioned earlier about the travel industry, KLM is collecting required information to support their customers on Facebook Messenger via a chatbot.
In a different industry, Limbic is collecting information about your mental wellbeing via the chatbot. The chatbot is embedded in the core value of the product this time. The product provides personalized paths according to the user’s answers.
In a similar manner, the chatbots can start the initial conversation for leads coming to your website. Why not start the conversation yourself instead of waiting for the user to come to you? A chatbot can do the icebreaking and start the conversation.
Chatbots can keep your users engaged by sending messages to them and asking whether they need any help from you.
Your “why” is going to help you to decide what to build or buy. I suggest you consider ease of use, scalability, and integrability to your systems while making this decision.
Customer support platforms naturally provide chatbot as a feature, such as HubSpot, Intercom, Zendesk. These platforms make the connection between chatbot, ticketing system, and knowledge base, enabling a comprehensive solution. These chatbots are able to be proactive or reactive according to your customer support strategy.
For sales purposes, you can look at chatbots from Drift, Tidio, Qualified. These chatbots are typıcally proactive: that means the chatbot starts the conversation to navigate the visitors in the sales process. They are integrated with common sales tools like CRMs, marketing communications, and appointment schedulers.
As you can see, there is a specific chatbot for every case. You might compare and filter out your options from the G2’s chatbot list as well.
I bet your business has several rules that can be automated. However, you don’t necessarily need to bring all of them to chatbots. Keep your scope simple with specific tasks and focus on designing to handle them efficiently at first. Monitor how users are interacting with your first chatbot and you may learn new things about your business too.
We mentioned that there are exceptions, and sometimes people want to interact with a real person. Don’t force them to use the chatbot and give them options to talk to someone when needed.
For example, customers may enjoy bots for basic support issues, but they’d like to speak with a human for technical support. That’s why starting simple can help you to decide which kinds of conversations a chatbot can handle for you.
For customer support chatbots, you may start with a review of the closed tickets of your agents. The ones that close quickly may be easy to automate.
People get disappointed when they realize they’re speaking to a bot, assuming they’d been conversing with a human. Letting users know that they are interacting with a chatbot is going to set their expectations right, but they will still expect to have a conversation.
If you are preparing defined responses, keep your answers conversational. Just because you are creating a bot doesn’t mean you need to be robotic while creating those scripts.
If you have user-specific information, use that information to personalize the experience. For example, you have a website analytics tool and you are using chatbots for lead generation. Based on the lead type, you either let the user self-serve or introduce a salesperson to help the user.
Customer: I would like to learn more about your website analytics products and pricing. Can you help me with that?
Chatbot: Absolutely! To get started, can you tell me how many monthly visitors you have on your website?
Customer: 5K visitors per month.
Chatbot: Great, thank you for that information! Are there any specific goals or challenges related to website analytics that you are trying to address?
Customer: Yes, I’m trying to improve my website’s conversion rate.
Chatbot: Thank you for sharing that! Based on your needs, I’d recommend checking out our “Small” package, which includes features like event dashboards and real-time tracking to help you understand how your visitors interact with your website. However, we also have other packages available that may be a better fit depending on your specific needs. Would you like me to send you more information about the available packages?
Customer: Yes, please.
Chatbot: Great! Here’s a link to view all of our website analytics packages. Let me know if you have any questions or if there’s anything else I can help you with.
Customer: I would like to learn more about your website analytics products and pricing. Can you help me with that?
Chatbot: Absolutely! To get started, can you tell me how many monthly visitors you have on your website?
Customer: 2M visitors per month.
Chatbot: Great, thank you for that information! Before I connect you with our sales team, can you please confirm your phone number and preferred time for contact?
Customer: Sure, my phone number is XXXXXX, and I’d prefer to talk today in the afternoon.
Chatbot: Thank you! I’ve noted your phone number XXXXXX and preferred contact time. Our team will reach out to you today in the afternoon. Is there anything else I can help you with?
In the big fish case, you may also see that chatbot provides clear feedback about the information it gathered. Using the information you collected in chatbot responses lets users know that their input has been received correctly. This simple addition helps users to learn what to expect next.
ChatGPT didn’t become ChatGPT overnight: we had GPT-3, GPT-3.5, and GPT-4 while I was writing this article. Even though they may have developed the most mind-blowing version of a generative AI chatbot, they need to learn and iterate.
It’s similar for you, too. Maybe you’ll realize that there is a medium fish case between small and big fish cases and you need to introduce medium fish to your sales team, too. Maybe you’ll learn that there are some loops in your conversation flows and users can’t reach out to a human when they need to.
Chatbots are not something you build and forget about. You should consider asking your users to rate their experience and give feedback, and check how many times your chatbot fails to give a helpful response. If not, you will be far off from the initial goal you set in step one.
Header image source: IconScout
LogRocket's Galileo AI watches sessions and understands user feedback for you, automating the most time-intensive parts of your job and giving you more time to focus on great design.
See how design choices, interactions, and issues affect your users — get a demo of LogRocket today.
A/B testing is great for comparing two versions of a design, but multivariate testing helps teams evaluate multiple design element combinations at once. Here’s how both methods work, how they differ, and when UX teams should use each one.
Design engineering has always lived between design and code. But with AI tools turning prompts into interfaces and code into editable canvases, that bridge is becoming a new way of building.
A three-week mobile banking project taught me that the “proper” UX process is not always realistic. Sometimes, the better approach is to work with what you know, identify what you still need to learn, and make the strongest decision possible under real constraints.
A/B testing compares two versions of a design to see which performs better with real users. Here’s how UX teams can use it to test hypotheses, measure outcomes, and make smarter product decisions.
Learn more →
