3 Clever Ways to Create a Chatbot for Your Website

Everyone is talking about chatbots. But how do you actually add one to your website?

First off, there is no one-size-fits-all solution to finding the perfect chatbot for your website. But don’t worry, it isn’t rocket science. Here are your options, use cases and examples.

  1. Build a chatbot inside of your website chat software
  2. Use an out-of-the-box chatbot
  3. Implement a chatbot using a chat provider’s API
  4. Use cases for website chatbots
  5. Examples of website chatbots

Build a chatbot inside of your website chat software

Through a survey about chatbot perceptions, we learned that customers greatly prefer an easy escalation route when speaking to a chatbot. The speediness that chatbots can offer is nice, but there are few things more frustrating than being stuck in a loop with a chatbot that cannot get you what you want.

Coupling your bot with your live chat software allows it to forward inquiries to an available agent when the customer has an issue it can’t resolve. It also makes it easier for your team to monitor its performance and make adjustments where necessary.

In our live chat solution Userlike, for instance, you can build your own chatbot with the Userlike Bot Language. With it, you can write bot scripts with simple, minimal syntax to assist customers through your website widget.

There are a variety of ways to use a custom chatbot on your website. To not get bogged down in possibilities, Userlike helps you out with four clear routing modes to choose from:

  • Human. The chatbot operates like any other human service agent. There is no special preference if a human or the bot answers the inquiry.
  • Failover. The chatbot only jumps in if all other operators have their chat slots occupied.
  • Firewall. The chatbot always answers the chat inquiry first. You can set your logic to trigger at which point to forward the chat to a living agent – for example when the chatbot is unable to answer the question.
  • Service Time. This one is active outside of service time. The chatbot answers all inquiries. By your determined logic, it can create a ticket for its human colleagues when a user needs help with something the chatbot can’t assist with.
Chatbot creating tickets out of customer interactions.

When it comes to chatting, we offer a few interaction approaches you can define with our chatbot options:

Text. Both the Logic Bot and the Chatbot API (described below) let your customer send your chatbot regular text messages. With the Logic Bot, this is useful for collecting pertinent information like an email address, order number, etc.

It’s also possible to connect the Logic Bot to an external API to look up the status of a tracking ID entered by the user (“When will my order arrive?”).

Bubbles. These are multiple choice options your chatbot displays for your customer to choose from in a conversation.

When your customer selects a bubble, the chatbot either provides an answer or takes an action, such as forwarding the request to a specific operator group.

Userlike Logic Bot showing text and bubble options.

Carousel. Similar to the bubble input node, the carousel mode lets you show a variety of choices the user can click. But instead of only presenting clickable bubbles, the carousel node allows you to add an image and a headline and description to each option.

It’s an effective way to show the visitor a selection of items, a useful tool in services like online retail.

Userlike Logic Bot showing carousel options.

Use an out-of-the-box chatbot

The quickest and simplest way to add a chatbot to your website is to create a chatbot using a platform like Botsy or PandoraBots. These ready-made solutions help you provide 24/7 chatbot service without the need for a developer.

A pre-made chatbot can take care of simple tasks like answering frequently asked questions, scheduling appointments or welcoming a visitor to your website. Just know that these chatbots often follow a rigid decision tree structure when interacting with a user, leaving little to no room for improving its AI.

On the upside, some platforms offer a chatbot that has the potential to grow its IQ if you add certain plugins like Google search, Bing search or RSS Import. On the downside, many pre-made chatbots are not equipped with a live chat option that allows it to forward the conversation to a human chat agent when it runs out of answers.

Some chatbot platforms, like Landbot.io, offer plugins that allow a chatbot to create a ticket and send it to your customer relationship management system. This lets your agents take on the difficult requests the chatbot wasn’t able to help with.

How you implement the chatbot on your website depends on the platform you choose and your website’s framework. You might have to download a plug-in, like for Wordpress, or you can paste the chatbot's code into your website's backend.

Implement a chatbot using a chat provider’s API

The above two options for adding a chatbot for your website are still rather basic. They are easy to implement and they are fine if you’re looking for simplicity.

For a more sophisticated chatbot setup, however, the best way is to connect your website chat solution with a specialized, advanced AI solution like IBM Watson.

When faced with a customer question, such an advanced chatbot would be able to run through the complete database of previous customer questions and answers given by human support reps. Then it could make a judgment of what is the best answer based on customer ratings and sentiment analysis, pick that answer and send it. And based on the customer's reply, it would be able to learn whether that answer was helpful or not–helping it do better the next time a similar question is asked.

Want to talk to a human about chatbots?

We'd be happy to answer any questions you have about using chatbots in your business and how to get started.

Book an appointment

There are no website chat solutions with such intelligence built in, and no AI solutions with added website chat functionality. For such a setup to be possible, your website chat solution needs a chatbot API.

At Userlike, our customers have used our API for all kinds of cool setups. Once you connect your advanced chatbot to Userlike, you have access to all the routing modes and interaction approaches mentioned above plus much more.

Use cases for website chatbots

Chatbots are not a replacement for human service agents. Although machine learning and natural language processing are evolving, bots still lack the ability to empathize and understand individual scenarios. But many use cases are predestined for the help of our android friends.

  • Collect basic information. Before a service rep can even lift a finger, your chatbot can collect the customer’s details and ask for the issue. Afterwards it can forward the message to a human who will pick up the conversation, already well-informed.
  • Answer common questions. A chatbot, ideally one connected to your knowledge base, can relieve your service team from answering frequently asked questions and guarantees a quick response.
  • Guide your customer to the right operator group. If your service agents have different areas of expertise, your chatbot can ask the customer to pre-select a certain topic or department. That way you spare the customer’s valuable time by forwarding the chat directly to the right person.
  • Create tickets. If all your chat agents are busy, the chatbot can leave a message for its human colleagues. This guarantees a timely follow up and prevents a chat inquiry from staying unanswered.
  • Offer sales support. By creating a multi-option menu within the website’s chat window, you can easily fuel your sales. The chatbot will know about your customer’s interest and offer relevant info.
Car website using chatbot that inquires the customer’s interest.

Examples of website chatbots

Here are some real-life examples of companies that successfully use a chatbot on their website.


LogoEnergie is a major electricity provider in Germany. They first implemented Userlike Live Chat in 2016. Two years later they decided to add the powerful chatbot system IBM Watson – with astounding results. Eighty percent of all customer requests could be answered by their new chatbot Henry.

“What the chatbot can easily solve are, for instance, questions regarding electricity meters and bill copies or if a customer wants to change the monthly part payment. The human staff is needed for individual cases,” says Christina Sologna, project manager at LogoEnergie.

“Of course we were apprehensive that it will lead to frustration if the chatbot can’t answer a question. But within the first four months that Henry was in place, we had no negative feedback – not even from the people who were forwarded to a service agent because Henry couldn’t sufficiently solve the case.”

Screenshot of chatbot connected to Userlike.

LogoEnergie’s chatbot is active 24/7 and only forwards chats to human operators if needed. Outside of service times, customers can ask for a return call or an email response. The LogoEnergie example nicely illustrates that the combination of human chat agents and a skilled chat bot is pretty powerful.


Smart’s strategy is very similar to LogoEnergie’s. On their website, chatbot Sophie is the first “person” in charge to answer customer inquiries.

She is able to help with finding a Smart car dealer in your area, answering questions about the different car models and guiding you to the right page. Sophie doesn’t necessarily answer all questions directly in the chat window, but let’s you know where to find the information.

Sophie also encourages visitors to make use of the Smart configurator. For example, when you ask about the gas or electricity mileage of a certain car model, Sophie doesn’t give you an average number, but sends you to the configurator.

Screenshot of smart chatbot.

Searching for the right car is half the fun, so users can enjoy configuring a car online, taking them one step closer to a potential purchase. It’s a smart move (pun intended) to lead the visitor to the configurator whenever there is a chance.

In case the issue can’t be solved by Sophie, Smart offers the option to fill out a contact form so the customer can be reached by phone.

Mercedes Benz

Although they both belong to Daimler, Mercedes Benz and Smart decided for different chatbot modes tailored to their specific target group. While Smart always tries to solve the issue with chatbot Sophie first, Mercedes Benz puts emphasis on fast, human-based support.

From 8 o’clock in the morning to 9 o’clock at night, live chat with service agents is the main service channel of the luxury car brand. To connect their customers to a skilled operator right away, they use a multi-select query; the visitor can choose a matching category before the chat starts: “consulting and test drive,” “configuration and equipment,” “market entry/delivery times” and “miscellaneous.”

Screenshot of Mercedes-Benz offering live chat support.

During service hours, Mercedes Benz doesn’t use a chatbot whatsoever. Their bot, aptly named Mercedes, only comes into play once the team is offline. After 9 p.m., she is the main customer service channel and operates very similarly to Smart’s chatbot.

TL;DR: You have three options for implementing a chatbot on your website

  1. Build a simple chatbot within a website chat solution like Userlike.
  2. Use an out-of-the-box chatbot that doesn’t require a developer.
  3. Use your website chat solution's API to connect an advanced chatbot.

The best way to add a chatbot to your website?

Although it's tempting to let a chatbot do all the work, we don’t recommend using it as a stand-alone solution for your chat support. Customers prefer a healthy combination of automated service and human live chat support.

Pre-made chatbots like Botsy can be useful for advertising, customer education and entertainment, but you will create the best service experience by working with a professional chatbot created within your live chat solution or one that is API-integrated.

Want to test our Logic Chatbot today? Sign up for a free trial and shoot us a message. Happy chatting!