Chatbot Development

Building Bots with Chatfuel

Chatfuel is a graphical platform for building Facebook Messenger chatbots. If you’re not a developer but still want to build a chatbot, Chatfuel’s easy-to-use bot-building tools might be the answer. With basic NLP capabilities and a variety of useful plugins, Chatfuel is a feature-rich option for many organizations.

If you’re looking for a fast, foolproof way to build a chatbot for one of the world’s most popular messaging apps, Chatfuel might be just the tool. Created for the express purpose of building Facebook Messenger chatbots, Chatfuel caters to professionals and hobbyists who want to build a chatbot but don’t know how to code.

Given its target audience of non-developers, you might think Chatfuel is light on features. It’s not. According to the company’s homepage, almost half of all bots on Facebook Messenger were built with Chatfuel. Big brands, like T-Mobile and Adidas, use the platform. Chatfuel tools, it seems, are powerful enough to handle communications for companies with reputations that depend on high-quality customer interactions.

Let’s take a closer look at what Chatfuel has to offer. We’ll go over some basic pros and cons and discuss a few key features. In addition, we’ll consider how “smart” Chatfuel is by examining how (or really, whether) it processes users’ natural language and delivers an appropriate response.

How to use Chatfuel

Practically anybody can build a bot with Chatfuel. Just sign up for an account, and start building a bot with the tool’s straightforward GUI. That being said, there are a few prerequisites. Before diving in, it’s a good idea to get familiar with the fundamentals. These are examined in a useful Chatfuel Getting Started tutorial, which covers:

• Key concepts for building bots with the Chatfuel GUI. The interface provides a block-like structure for creating conversation flows.
• Structuring your bot to effectively automate different kinds of conversations.
• Pausing your bot, when needed, so that a human can take over.
• Directing traffic to your bot so that people actually get to use it.
• Ensuring that lead and customer data makes it into your CRM.

All of these tactics relate to how you might use Chatfuel in a very hands-on sense, and that’s helpful. But what about the platform’s overall purpose? What kinds of chatbots is it most capable of helping you build—and why does that matter?

Terminology like customer conversations and CRM suggest that Chatfuel bots (in particular, those used by businesses) are best suited to marketing, sales, and brand building. And, when you look at the ways different brands are using Chatfuel, it becomes clear that those activities are indeed its most compelling use cases.

Chatfuel chatbot examples

The T-Mobile Chatfuel-built bot helps new and existing T-Mobile customers shop for phones, learn about deals, and get directions to nearby retail locations.

You can also verify coverage, look at plan benefits, and request pricing details for different plans. Like most Chatfuel-powered bots, the T-Mobile bot relies on the Facebook Messenger quick replies (the in-chat buttons you use instead of typing a response). This architecture makes it possible for every user input to trigger a preconfigured response on the T-Mobile end. In other words, the bot doesn’t have to deal with unique and unusual user inputs and attempt to decipher them. It’s a relatively simple bot.

Another bot built using Chatfuel is the Golden State Warriors chatbot, and it is intended to increase sales. After activating the chatbot, users can receive game information, scores, and videos. They can even view ticket offers and shop for gear.

For Golden State Warriors fans, chatting with this bot could be an excellent way to interact with the brand—and maybe even score some tickets to a game.

The LEGO chatbot is a fun one. LEGO used Chatfuel to create a bot that helps you shop for LEGO sets. Just tell the bot how old the “builder” is, and provide your shopping budget.

To further refine your selections, the bot also asks what the builder is interested in (transport, superheroes, or animals, among other options). Ultimately, you’re presented with LEGO sets you can purchase online.

Pros and cons of Chatfuel

Chatfuel pros

For many organizations, the biggest advantages to Chatfuel are the most obvious ones. It allows you to build and deploy a chatbot to Facebook Messenger, a channel that millions of users and potential customers engage with every day. The graphical interface also makes it accessible to business users. Marketing professionals can build and maintain chatbots without involving IT—a big win for efficiency at many organizations.

Since Chatfuel offers a free tier, there really aren’t any barriers to entry. Anyone can build with it, and anyone can afford to start using it. Even if you ultimately plan to build a more complex bot using more sophisticated tools, you can start with Chatfuel to prototype a bot that you can later build on elsewhere.

If your primary use case for a bot is to automate repetitive sales, marketing, or support requests, the ROI from Chatfuel might be immediate. As more users activate your bot to solve simple problems, you and your team will have more time to focus on valuable tasks.

Chatfuel cons

The biggest plus of Chatfuel might also be its greatest limitation. The platform lets you build bots for Facebook Messenger—but that’s really it. If you need a bot for other channels or even for your own website, you’ll need to look beyond Chatfuel.

After all, it makes sense to deploy a bot to all the platforms where your customers are. Facebook Messenger might be among those platforms, but it’s unlikely to be the only one. What’s more, you’ll almost certainly want to deploy your bot to a dedicated domain. A .BOT domain is perfectly suited to this purpose, and virtually every chatbot framework gives you a way to deploy your bot there. That’s not the case with Chatfuel.

This impediment underscores another potential problem with Chatfuel. It lacks the features and capabilities found in a comprehensive chatbot framework. To be clear, Chatfuel isn’t trying to be a chatbot framework. It’s a channel-specific toolkit, and that’s fine. However, some may find the Chatfuel GUI and limited AI capabilities to be overly restrictive.

Chatfuel features and integrations

The Chatfuel GUI is its most compelling feature. An extremely easy-to-use block-based builder lets you add content to your bot, assign specific response options, and add extra content to each response.

Needless to say, the seamless integration with Facebook Messenger is another plus. Other non-platform–specific tools are a little clunkier when it comes to deploying bots to messaging channels. With Chatfuel, you’re signed in to Facebook Messenger while you create your bot, and Chatfuel associates your bot with the company’s Facebook page before you even start building.

To jazz up your bot, Chatfuel also comes with plugins. There are options for video and audio messages, as well as a CRM-like plugin that lets you document user attributes, track user behavior through your conversation flows, and subscribe users to follow-up messages. A Google Sheets plugin also makes it easy to export user data to a spreadsheet. Other plugins deliver efficiencies, like Google or Bing searches or an RSS import for displaying content from a feed.

There are also plugins for live chat and conversation handover, so humans can take over when your bot’s predefined messages aren’t enough to solve a user’s problem. This is a good feature, but it also underscores some of the platform’s development limitations.

Chatfuel pricing model

Chatfuel has three price tiers:

• Free. Get up to 1,000 subscribers and basic conversation automation.
• $15/month. With this plan, you get unlimited subscribers, Facebook ads integration, and the ability to remove Chatfuel branding.
• $199/month. Aimed at larger businesses with multiple bots, this plan includes a dedicated account manager and VIP support. You also gain the ability to push updates to multiple bots at the same time.

How does Chatfuel learn?

Since Chatfuel isn’t exactly a chatbot framework, it doesn’t include the robust natural language processing (NLP) engines and machine learning modules available in other vendors’ toolkits. However, it does include modest NLP capabilities that are reminiscent of the intent recognition and entity extraction fundamentals that the “real” frameworks are known for.

Via the Chatfuel bot-building interface, you can define some conversation rules for your bot to follow. Essentially, you’re telling your bot, “When a user inputs hi, hello, or hey there, respond with, Hey yourself! How’s it going?”

Luckily, your bot is smart enough to interpret some variations of a greeting as sufficiently similar to whatever you taught it when you defined your conversation rules. So, in this case, the bot might recognize, hey bot as a phrase that’s similar to something you taught it to recognize. When it does—and when it processes the user’s input accurately—it will issue the appropriate response.

The result? Your Chatfuel bot only knows what you teach it, so you’re not getting a true machine learning model that improves over time. Regardless, this is pretty impressive when you consider that Chatfuel is a strictly GUI-based tool for non-developers. You get some of the functionality that’s central to NLP-based tools in a basic interface that’s suited to business users.

How to get started with Chatfuel

Although you could just select the Get Started link on the Chatfuel homepage, it’s a good idea to first read through the Getting Started guide. For that matter, you might want to read through the Chatfuel documentation, as well. There’s a lot of information there, complete with screenshots, and it provides a good overview of what to expect when building bots with this tool.

Given the relatively risk-free nature of Chatfuel, you may as well give it a try! If it’s your first foray into the world of chatbots, a simple bot builder like Chatfuel can be a great way to get your feet wet and create a prototype that you enhance when the time is right.