7 Rules of Chatbot Building

Chatbotize.me Messenger marketing and Chatbot Agency
Share on facebook
Share on twitter
Share on linkedin

Brief History of Messenger chatbots

In 2016, Facebook opened up Messenger’s API and made development possible onto the platform. They actually gave new business opportunities tíwith this o brands and companies. First chatter robots, chatbots had arrived on behalf of Facebook pages, and they started communicating with customers automatically.

These days an average person spends more time chatting than browsing social media, and Messenger is one of the most popular instant messaging platforms of all over the world. We can spend even one hour on Messenger on a daily basis. Continuous online presence, reachability and chat have become part of our everyday life. It’s become natural that we’re able to get any information at any time immediately, especially if we know whom to ask from. Instead of long mailings and phone calls that don’t tolerate multitasking, we started using chat as a better alternative in most cases.

Messenger is a complete new channel for brands

It works with our relatives and friends, and with opening up the Messenger platform, Facebook made possible that it could work with our beloved brands and companies, too. Along with this, a new potential communication channel was born between people and companies which is still free from noise.

The same thing happened when the internet was born, or let’s mention the birth of Facebook pages. We cannot mention a single company today that does not have a web or a Facebook page at least. Ones that noticed it early won the biggest, as they will do so now with Messenger.

How a Messenger Chatbot works and what is it good for?

What is a Messenger chatbot? In practice, it’s the automation of a chat between a Facebook page and a person. As Messenger platform is open, anyone can build or have someone to build a Messenger bot for their Facebook page.

Thanks to the increase of popularity of chat, nowadays we’d more likely to text a business instead of calling or emailing them. According to Facebook, more than 56% of people would rather do that.

Basically, chatbot technology tended to help in customer service with automating repetitive tasks. However, as it revealed in the meantime, Messenger is unbelievably effective with reaching people. Some stats:

  • Average open rate of messages in Messenger is 84%, while it’s 23% in email
  • Average click-through rate is 28% in Messenger, while we’re glad if we have 3% in email

That’s why using a Messenger chatbot for marketing has become really popular. We reached 20-30% increase in conversion with our multiple clients only with introducing a chatbot into their lead generation processes.

The 7 Rules of Building a Messenger Chatbot

Rule #1: choose your tools that match your project, and not vice-versa!

There are basically two ways for creating a Messenger chatbot:

  1. You can develop natively on top of Messenger platform, on its API
  2. There are platforms built for making chatbot creation easier with GUI bot building interfaces

Manner number 1 is advised when:

  • The solution cannot be delivered with a GUI bot builder platform because of technical reasons
  • The policy of the corporation does not allow to have a third-party platform in the architecture or does not allow storing data other than their or Facebook’s servers
  • You need something really complex, something app-like

In all the other cases it’s highly recommended to use one of the GUI bot builder platforms. The 2 most popular platforms are Chatfuel and ManyChat. They run 90% of all existing Messenger bots in the World (approximately 50-50%).

In my opinion, you shouldn’t get trapped into wanting to solve everything with your one beloved platform. At our agency, Chatbotize.me, we have a process to decide which platform is the best for a particular project. We analyze the requirements and the possible further developments and decide which platform will provide the simplest, most stable and most cost-effective solution for the particular project.

Rule #2: decide what your goal is! Is it serving your existing customers better, or more-likely getting more clients?

We use to categorize chatbots into 3 different groups:

  1. Customer service chatbots
  2. Marketing, lead generation chatbots
  3. App-like, more complex bots that give assistance in something

We sort the most complex and unique developments into the third category which are sometimes worth to develop with native coding. However, they represent the smallest category in terms of volume. That’s why I won’t discuss them further in this blog post.

So the two most important categories are lead generation and customer service chatbot solutions. If you already have the idea to create a chatbot, it is likely that your idea will fit in one of these two categories.

At our agency, we even have proven formulas and concepts for both of them, which shows that they are really quite popular kinds of chatbots.

The point is, you need to be able to define what goal your chatbot should serve. You should not get lost in the details.

You will be able to measure the success of your project based on your pre-set goal and the corresponding measure of units. For instance, if you build a marketing chatbot, your biggest KPI will be conversion rate, while in a customer service bot it will be the rate of offloading human work.

Rule #3: create contents that are enjoyable in Messenger!

Maximizing customer experience is a very important thing also in Messenger, it needs special expertise. It’s even got a name: Conversational User Interface.

At first. it seems like only a new buzzword, but when your bot will be used by multiple thousands of people, then all of them will matter that leave your chatbot because of your boring or too-long texts.

You can use a bunch of rules of thumb when designing a conversation on Messenger, but common sense and continuous testing will help you too.

Let me explain a quick example of a rule of thumb we use in our developments. We do not use to send out too long messages, anything longer than 6-8 lines in one text block would be separated into different text blocks. When you separate them, it’s extremely important to send them out in a way that lets the user read the whole text frictionless. It means, we only let the next text block to be sent out when the user can finish reading the current one.

We use typing elements between blocks and set them to the exact timing that the previous block needs to be read by an average person. And how do we know the length of this time? We use a rule of thumb that says 1 line = 1 second. So if we send out a block of text with 4 lines, then we would set a typing after it to 4 seconds, and we would send the next text lock only after this typing element passes.

Rule #4: Call To Actions to everywhere! Do not interrupt a chatflow!

Coherent parts of the planned conversations are called chatflows. We are able to place predefined response options into these chatflows and lead our users with the help of them. It is very important not to leave the users at any point of the conversation without these kinds of options (quick replies, buttons), not even at the end of a pre-designed chatflow.

Why is it so important? Chatting happens in a linear interface which means we only focus on the last thing occurred on the surface, not any more of it. That’s why it is critical to provide Call To Actions at these situations that lead the users to next steps or functions which follow logically from the context.

Rule #5: Keep a good balance of using AI and graphical elements!

In the chatbot scene, people use to call the technology that can recognize the meaning of a typed text AI. What they truly mean under AI is actually NLP, Natural Language Processing.

This is the stuff that makes a chatbot “smart”. This technology is much more sophisticated than a chatbot with only graphical elements. So we have to set realistic goals with our chatbot. We have to balance the rate between the more expensive AI and graphical elements in a way that our problem to be solved with a chatbot needs.

For example, during our marketing-oriented developments, we have a very limited AI capability in the bots we deliver, as our goals do not justify more AI to have. Our customer service projects are completely a different topic. At first, we always do a deep analysis of the business of our customers. We will start balancing the AI, the graphical and human parts of the chatbot only after this analysis while considering always two things: efficiency in operation and return of investment.

Rule #6: Plan beforehand how your users will find and start using your chatbot!

It’s not enough to just design the chatbot, you also have to plan how people will start using it. There are plenty of options for this, I’ll introduce the most important ones of them:

  • One can just easily text the page via Facebook / Messenger. We have to count on the possibility that some users will just text us.
  • A chatbot can be advertised on Facebook, Instagram and Messenger. After a user clicks on the ad, Messenger opens up immediately and communication starts with the bot. This is currently a very cheap way of advertising. The other advantage: we don’t only educate them with the ad, but instantly subscribe them to our Messenger list with that one click-through from the ad.
  • You can embed the Messenger chatbot on your website as a customer chat window. Whoever clicks on it will be available for you even after they close your website. Chat can be continued between the user and the chatbot on Messenger, or you can also chat with them manually from Facebook Page Inbox.
  • You can also embed Messenger onto your website as a button or even as a checkbox. These can be placed onto popups or forms and users that click on them will be your subscribers on your Messenger list.
  • We are able to move people from offline space to online with the help of chatbot shortlinks. You can share the shortlink anywhere, or generate a QR code from it, and place it on flyers or billboards. You can even redirect a company domain of yours to the chatbot shortlink.
  • Facebook posts can be set up with chatbots to catch commenters and lead them into the bot. All commenters can receive an automated Message from the chatbot after they have placed comments, and thus they can subscribe to your chatbot.

So what’s important is you have to consider all possibilities and options and plan how you’d like your users to find and start using your Messenger chatbot.

Rule #7: Do not believe it only takes 5 minutes to build a Messenger chatbot!

Unfortunately landing page of one of the biggest chatbot builder platforms proclaims that “Build a chatbot in only 5 minutes!” which causes many misunderstandings. Of course, they’d like to highlight the usability of their platform, which I have to agree with, but you have to imagine this like WordPress. Creating a web page in WordPress can also be only 5 minutes, but what kind of web page would it be like, what do you think? Do you have to have a web page for your company that had been developed within 5 minutes? Of course WordPress speeds up building a website compared to native HTML coding in a notebook, just like chatbot builders speeds up building a bot. But it doesn’t mean quick and easy wins.

The good news is thanks to platforms like Chatfuel and ManyChat the threshold of entry is really low. Anybody easily can start learning chatbot building. However, building a Messenger bot that really can generate leads or really offloads our customer service it’s necessarily more than just 5 minutes of work.

After a serious strategic and conceptual planning and choosing the appropriate toolset, a precise and consequent development process follows. However, it’s not even the finish line here. In order to maintain or continuously increase the effectiveness of our Messenger chatbot, we have to continuously improve and work on it.


You can see now you have to consider a lot of things when you are building chatbots. If you keep this 7 ultimate rule of chatbot building, I guarantee you that you’ll be successful with your chatbot. I’m sure it will bring multiple returns of investment for your company, and you can build a chatbot that is a match with your goals. I emphasize you now to start or continue building your chatbot, start using Chatfuel or ManyChat, and feel free to tell us your experiences!

Reveal how much money your business could profit

Answer a few questions and find out how much money your business could profit from a chatbot.