Bots for Business
A Comprehensive Guide to Chatbots in Messaging Apps
Which messaging app is the best home for your chatbot? This guide covers the features, benefits, and drawbacks of developing bots for the world’s most popular messaging apps. Explore chatbot use cases and user demographics for every app, while learning which frameworks provide the best support and integration.
April 30, 2019
From websites to mobile apps to social networks, people connect with bots on all sorts of platforms. Messaging applications, in particular, are among the most common places to deploy chatbots—especially when you choose a messaging app that your intended audience already uses.
Every messaging app offers unique advantages and drawbacks. A chatbot that only achieves so-so engagement on one app might enjoy stellar adoption rates and conversation quality on another one.
To ensure your bot is successful, you need to deploy it on the right messaging application. This guide can help you gain an understanding of which messaging app is best suited for you, your bot, and your business model. Here are the apps we cover:
Choosing among chatbot apps
After learning how to make a bot for an app, your next question will be which chatbot app to select. Moreover, there are several benchmarks you need to consider. To break it all down, the following comparison table serves as a cheat sheet for the messaging apps we cover in this guide.
|Primary Uses||Preferred Frameworks||Special Features||Chatbot Examples|
|Viber||Developer tools, including open source APIs|
*Note: All frameworks listed in this table have seamless integration options specific to the app.
As you learn about different messaging apps and their capabilities, keep in mind that you don’t have to commit to a single app when deploying your bot. When your bot lives on its own domain, you can deploy different iterations of your bot across multiple apps.
Tying your bot to a .BOT domain name is like giving it a permanent home. Users will always be able to find your bot there, even as its capabilities change.
As your bot evolves, some apps may not grow with it. Besides, an app’s functionality may change, or its user demographics might shift so that it no longer makes sense to keep your bot there. Giving your bot a permanent home with a .BOT domain name means they can always find you.
Now let’s explore the attributes of several popular messaging apps. We pay particular attention to use cases, framework pairings, and the unique advantages of each platform.
Facebook Messenger chatbot
In terms of monthly active users, Facebook Messenger is the world’s second most widely used messaging app. In the United States and Canada, it holds the number one spot. Needless to say, a messaging app with over 1.3 billion users worldwide is a great place to deploy many kinds of chatbots.
But is Facebook Messenger the right place to deploy your bot? Let’s take a look.
Facebook Messenger features, pros, and cons
Besides the sheer enormity of its user base, Facebook Messenger includes a diverse suite of features for bot developers. For starters, there are built-in payment processing tools for e-commerce (so users can engage a bot to order a pizza or to buy a new watch, for example) and analytics that provide insights about bot performance at a granular level. Online merchants can take advantage of ads that click to Messenger, which allow your target audience to start chatting with your bot right away.
Facebook Messenger also includes a few one-of-a-kind advantages, like short URLs that are easy for users to remember and codes that users can scan to go directly to your bot.
For advertisers, the Facebook Messenger limitations on push messaging might be a disadvantage but probably not a deal breaker. You can only push a message within 24 hours following any user’s most recent message, unless you pay Facebook for the privilege. Sponsored Messaging is a paid feature in Facebook Messenger that lets you send promotional content to users who engaged with your bot past the 24-hour window.
Best frameworks to use
There are all sorts of ways to develop Facebook Messenger bots. If you’re new to building bots and don’t want to—or can’t—deal with code, Chatfuel is an excellent tool to check out. It has everything you need to build a basic bot for Facebook Messenger and to launch it right away.
Chatfuel, of course, is a graphical tool. It’s arguably not a “true” chatbot framework. Wit.ai is Facebook’s proprietary framework, and it comes loaded with natural language processing (NLP) and machine learning capabilities. It’s also free, so you can build your bot and deploy it to Facebook Messenger without paying a cent.
Other frameworks with ready-to-go Facebook Messenger deployment tools include Amazon Lex, Microsoft Bot Framework, Dialogflow, and Botkit.
Facebook Messenger chatbot examples
Organizations large and small leverage Facebook Messenger for human-bot interactions. Consider TechCrunch, which has a popular Facebook Messenger chatbot that can field queries like, “What’s the word on ride-sharing apps this week?” Following a question like that one, the bot displays recent TechCrunch news on that topic.
For an example of an e-commerce chatbot on Facebook Messenger, look no further than 1-800-Flowers. The 1-800-Flowers chatbot is, to put it mildly, capable. It can request delivery addresses, show users specific flower arrangements, and take delivery orders—right from the chat screen.
Need to send or transfer money to a business or an individual outside your home country? The RemitRadar Messenger bot provides instant quotes on transfer fees (depending on the location you’re sending to and the currency you’re using) and helps you initiate the transaction.
If you’re looking for ways to get involved with your local politics, consider Resistbot. This simple messaging bot will keep you up to date during elections and also gives you the opportunity to write letters to your elected officials.
WhatsApp is the most widely used messaging application in the world. It’s also the most popular app of its kind in Central and South America, as well as parts of Africa, Asia, and Europe. If your bot caters to users in these areas, WhatsApp might be your platform of choice.
WhatsApp features, pros, and cons
First things first, WhatsApp probably won’t be your go-to platform for deploying sophisticated bots capable of performing a variety of functions—yet. That’s because WhatsApp is a relative newcomer to the bot world, and they’re taking things slowly.
The WhatsApp Business API, which you use to get your bot onto WhatsApp, became available long after Messenger, Kik, Slack, and others mainstreamed chatbots on their platforms. What’s more, only large and medium-sized businesses can use the API. Everyone has to apply to host a bot on WhatsApp, and you won’t get approved unless WhatsApp decides you’re a good fit.
After receiving approval, you get unfettered access to a suite of developer tools, which include message templates and integration with your APIs for purchases, bookings, and other transactions. However, aside from these fairly straightforward activities, WhatsApp doesn’t support the multifaceted conversational experiences of Facebook Messenger and others.
If you’re a small business, you’re limited to the WhatsApp Business app. Strictly speaking, this isn’t a chatbot platform. Aside from automated away messages and instant greetings, it’s human chat facilitated by some simple automation.
Best frameworks to use
Gupshup is the only framework that has a dedicated integration for WhatsApp for Business. Its handy Gupshup scripting tool allows you to get your bot into the platform with ease.
Of course, if you prefer to go an alternate route, there are workarounds. Developers at redBus, a bus booking service, actually used the Google Dialogflow framework to build a bot that they then integrated with WhatsApp, utilizing a variety of tools. Another developer built a WhatsApp bot in 30 minutes using the Twilio API for WhatsApp and the DuckDuckGo Instant Answer API. The result was a simple bot that regurgitates information from Wikipedia in response to a user inquiry. For organizations already cleared for the WhatsApp Business API, Botsociety offers a toolkit that helps you build a chatbot for WhatsApp.
WhatsApp chatbot examples
If you’re flying with KLM, you can use the KLM WhatsApp bot to receive flight confirmations, check-in notifications, and flight status updates. You can also get your boarding pass sent straight to WhatsApp. In the event you need to speak with a human agent, you can use the WhatsApp thread to request assistance.
Indonesian bank, Bank BRI, uses WhatsApp to provide customers with a personal assistant experience. The bank’s bot, Sabrina, shows customers how to find a nearby branch or ATM and can even field complaints. Previously available on Facebook Messenger and Telegram, Bank BRI is leveraging the WhatsApp Business API to meet customers on a platform where they’re already active.
If you use Slack at work, you know that conversing with colleagues and team members is pretty much the point. With its chat-centric interface and focus on workplace activities, Slack is a great channel for deploying chatbots that enhance productivity.
Slack features, pros, and cons
The Slack API enables anyone to add custom features to their workspaces. It also lets you deploy bots designed to help teams and individuals alike. Indeed, Slack offers a repository of bots that you can add to your channels to get more work done, faster.
The Slack API documentation is very bot friendly and includes an entire section on deploying your bot on Slack. After creating a Slack app, to deploy your bot, you step through the following process:
- Create a bot user. This gives your bot a proper display name and profile.
- Set up the Slack Events API. You do this so that your bot can respond to particular events, such as user messages and activities.
- Install your bot. That is, get your bot from wherever you built it (a framework, for instance) onto Slack.
- Work with event handling. This includes receiving and responding to all sorts of user activity.
Of course, you’re free to integrate your Slack bot with any other relevant data sets or tools. That way, the bot can field an information request, grab the data elsewhere, and send it back to users within the chat screen. That’s what Coinbot does. You can ask about current prices of Bitcoin or Ethereum and receive real-time updates without leaving Slack.
As for potential disadvantages, it’s not so much what Slack can’t do, it’s more like what it probably shouldn’t be used for. Remember, Slack is a tool for small to medium-sized teams. Its users are mostly people in professional environments collaborating to get work done. Slack isn’t Facebook Messenger. It’s not a place where insurance agents process quotes or where e-commerce organizations display the price of tennis shoes.
If your bot does something related to productivity (think team collaboration, human resources, workflows, or calendar management, to name just a few), Slack is an excellent platform for that. However, if you’re launching a bot for e-commerce, customer support, or sales and marketing, Slack is probably not the right place
Best frameworks to use
Amazon Lex, Microsoft Bot Framework, Botkit, Botpress, and Rasa are popular frameworks that integrate with Slack.
To get a sense of how the integration works, consider how Slack works with Amazon Lex. After building your bot and creating a Slack app, you enter data into the Amazon Lex console, head over to the Slack API Console, and enter more data there. Voila! Your bot is launched.
Slack chatbot examples
The bot listing on Slack’s website is a great place to get a sense of what people are building for the platform. There’s MailClark, which lets you send and receive emails directly from Slack, and Workstreams, which adds a Kanban-style board to Slack to assist with task management.
Well-known tech brands have bots on Slack, too. Using the SurveyMonkey bot, you can automate office polls and receive notifications about the results.
WeChat dominates China. Although the Chinese government has banned or censored WhatsApp, Facebook Messenger, LINE, Telegram, and others, WeChat is different. Unlike platforms developed elsewhere, the Chinese government subsidizes the app and can monitor and control users’ communications.
For many users, WeChat is more than just another messaging app. As Tech Analyst Ben Thompson notes, “[Every] aspect of a typical Chinese person’s life, not just online but also off is conducted through [WeChat] (and, to the extent other apps are used, they are often games promoted through WeChat).”
In China, WeChat pretty much is the internet. If you’re building a bot for the Chinese market, it’s almost a given that your bot will live on WeChat.
WeChat features, pros, and cons
The WeChat Open Platform offers extensive English language documentation for developers. There are SDKs for mobile apps, websites, hardware, and WeChat Pay, the platform’s very own payment processing tool.
Most chatbots on WeChat are, for better or worse, very basic. They’re only capable of responding to keywords that appear in lines of text. So, if you engage a grocery shopping bot by saying, “Show me prices for cereals,” the bot might pick out prices and cereals to work out what you’re asking. For simple customer service bots or for bots charged with finding and delivering tidbits of information (such as a weather bot or a bot that displays headlines), this kind of functionality is probably sufficient.
This doesn’t mean that WeChat can’t support a more sophisticated chatbot. The WeChat Mobile App SDK documentation lists several machine intelligence APIs, including speech synthesis, image recognition, and NLP (Chinese only).
Best frameworks to use
None of the major frameworks offer any integration capabilities for WeChat. That being said, you can theoretically build a bot using whatever tools you want and deploy it on WeChat. This “Building Chatbots for WeChat” tutorial shows you how to do so using Falcon (a Python-only framework) and the WeChat sandbox.
WeChat chatbot examples
One bot that’s representative of what’s available to WeChat users is the Meici bot. In China, Meici is a popular flash sale website. Users can engage with Meici to find the best deals on everything from handbags and shoes to complete outfits.
Microsoft, however, has managed to build a much more sophisticated bot for WeChat. Xiaolce is a social chatbot that users can engage with for simple (and, for the most part, superficial) conversations. She can write poems, provide “opinions” about various topics, and identify famous people in photos, assuming that she has previously learned about them.
Viber, owned by Tokyo-based Rakuten, has a global presence, and a lot of big brands have started using it to connect with customers. Although Viber claims to have around one billion unique users worldwide, some third-party research says 260 million is more likely.
Regardless, Viber is a widely used messaging app. It’s especially prevalent in Central and Eastern Europe, Russia, Southeast Asia, the Middle East, and Oceania.
Viber features, pros, and cons
If you’re targeting markets where WhatsApp is dominant, but you aren’t enthusiastic about its limited options for chatbots, consider deploying your bot on Viber.
For one thing, you don’t have to be a large or medium-sized business or go through an application process. You can sign up and start building your bot right away. However, there are a few bot-related guidelines from Viber that you need to plan for, including:
- The ability to respond to all types of Viber messages, including text, stickers, and images.
- Fast response time (under five seconds) or a message explaining why the bot is taking so long.
- Greeting all users who engage the bot.
- Ensuring your bot functions the way you say it’s going to function (so no surprises for users or for Viber).
As long as you’re able to follow Viber’s guidelines, you should be able to deploy your bot on this app with ease.
The biggest pros for Viber? Extensive developer tools, including Viber API Community open source contributions, that can help you contact users in the same markets where WhatsApp won’t let you reach them. The documentation also includes reference material for the Viber Node.js Bot API, Python Bot API, and REST API, all of which may accelerate the deployment process for you. The cons? Viber isn’t Messenger or WhatsApp, and it doesn’t have as many users to connect with—at least not yet.
Best frameworks to use
Dialogflow is the only major framework with documentation that walks you through Viber integration. That doesn’t mean you can’t use a different framework. It just means Dialogflow makes it a little easier, right out of the box.
Getting your Dialogflow-built bot onto Viber is simple. All you have to do is navigate to the Integrations area of Dialogflow, click Viber, and enter data into a few fields. Then you’re done!
The critical thing to remember about framework integration is that you don’t always have to use a framework that was built with the app in mind. For example, Amazon Lex can integrate with any messaging service. You might have to use a few ancillary tools and APIs to make it happen. Other developers have leveraged the Microsoft Bot Framework and Microsoft Azure services to build bots that were ultimately deployed to Viber.
Viber chatbot examples
Viber bots perform various functions, and the audience skews very consumer/social (think Messenger) versus professional/productive (think Slack). For example, you can engage the Tech Talk bot to find articles about given tech topics. Just key in mobile banking, and the bot shows you the latest stories related to that topic. Mica, the Hipster Cat Bot, is another informational Viber chatbot. If you’re in a particular city looking for, say, a sandwich shop, Mica points you in the right direction.
To see how big brands are using Viber, look no further than Glamour magazine. The Glamour Viber bot shares new content with its Russian audience, including pieces the bot determines might be attractive to specific users. Coca-Cola's Viber bot took things even further by inviting users to join its “under-the-cap” sweepstakes and by encouraging them to also invite their friends. According to Viber, the campaign resulted in a 30 percent conversion rate.
Based in Tokyo, LINE is the most popular messaging app in Japan, Taiwan, and Thailand. With a heavy emphasis on social games and its expansive sticker market, LINE may be a great place to deploy your consumer-oriented, social, or entertainment chatbot.
LINE features, pros, and cons
Outside of a few markets in Asia, LINE isn’t particularly popular worldwide. However, in places where it’s popular, LINE is very popular. For most LINE users, the app offers a messaging app/social network hybrid experience where playing games and accumulating stickers (i.e., large emojis, many of which are animated) are all part of the fun. The LINE audience skews younger, but not by much.
If you want your bot to live on LINE, head over to the LINE developer tools. There, you gain access to the LINE Messaging API and LINE Bot Designer tools, which allow you to prototype bots for LINE in an accelerated fashion. Bot Designer is software that you run locally to design your bot, and it’s available for free. If you’re creating a bot for LINE (that is, not for multiple channels simultaneously), Bot Designer might be the most efficient way to get your bot onto the platform.
Additional features that are unique to LINE include LINE Pay, which lets you add e-commerce functionality to your bot, and Beacons, which trigger your bot whenever a user is within physical proximity of an actual Bluetooth beacon. For the moment, Beacons are only available in Japan, Taiwan, and Thailand.
Best frameworks to use
Major frameworks like Amazon Lex, Dialogflow, and IBM Watson Assistant don’t include straightforward integration with LINE. However, this LINE Bot development tutorial can give you a sense of what it’s like to create a LINE bot using C# and Microsoft Azure, and the documentation shows you how to build a bot using the Messaging API.
And, of course, you can always use the LINE Bot Designer application to create a simple bot and deploy it to the platform.
LINE chatbot examples
LINE is chock full of gaming and entertainment bots. One of them is Erwin, a chatbot who peppers users with “knotty” riddles. Erwin might be a nice cure for boredom, but it’s hard to say that he’s particularly useful. Fufilo, on the other hand, is quite useful. If you’re in Asia and want to buy products from the United States, just tell Fufilo what you want. She lets you know whether the item is available and how much it would cost to have it shipped directly to your door.
On its FAQ page, Telegram includes this question: “How is Telegram Different from WhatsApp?” The answer is a long one, but part of it involves the app’s Bot API. The Telegram API is open to everyone; the WhatsApp API isn’t.
Telegram features, pros, and cons
Initially launched as a secure alternative to WhatsApp (Telegram offers end-to-end encryption), the app quickly became popular in Russia, which subsequently tried to ban the app but to little avail. And while some still question whether Telegram’s data is truly secure, users around the world rely on the app not for messaging but for sharing potentially sensitive files.
For bot developers, the perception of Telegram as a secure messaging app means it could be a good platform for bots that process banking info, integrate with cryptocurrency wallets, or send money. Indeed, Telegram bot literature stresses the straightforward integration of its payments platform, as well as support for inline queries, HTML5 gaming bots, and custom keyboards for simple replies.
As with other less widely used platforms, like Viber and Kik, the only actual con for developing a Telegram bot is its adoption. Facebook Messenger and WhatsApp have many more potential users than Telegram. Two hundred million users certainly aren’t nothing, but other platforms boast one billion or more.
Best frameworks to use
Among the major frameworks, only Dialogflow offers straightforward integration with Telegram as part of its core features. Of course, you can use any framework you want to create a bot and deploy it to Telegram. For example, the botpress-telegram tool makes it easy to build a bot on the open source Botpress framework and to launch the bot on Telegram right away.
Telegram chatbot examples
There are all sorts of bots on Telegram, but those involving payments, cryptocurrencies, and secure communications seem to be more prevalent compared to other platforms. Take XirkleBot, for example, which is both a cryptocurrency wallet and bot that lets you send and receive Bitcoin, Ethereum, and other digital currencies. There’s also the Means bot, which facilitates collective shopping for events by showing balances owed for each participating user.
If you’re wondering whether larger companies are building bots for Telegram, the answer is yes. BBVA now lets customers in Spain send money via the bank’s Telegram chatbot.
In Kik’s very own words, its “bots are built for being social. You can chat with them, do quizzes, get fashion tips, news, advice and more.” The Kik Bot Shop should give you a pretty good idea of what’s possible on the platform.
Kik features, pros, and cons
According to the Kik Brands page, 70 percent of its users are between the ages of 13 and 24. North America is Kik’s primary market, and a lot of American brands (including Yahoo, CNN, and The Weather Channel, among others) are leveraging the platform to connect with teens.
If you’re building an English-language bot with a teenage audience, Kik is definitely a platform to consider. But if you’re building a bot that helps with more grown-up activities (like shopping for a new car or checking stock prices), you might not get as much traction.
On the development side, there’s a lot to like. The Kik Bot Support Center offers several how-tos and tutorials that can accelerate the creation and deployment of your bot. You can also refer to Kik technical documentation for building bots, which includes an API reference and developer guidelines for building a functional bot that meets the platform’s specifications.
Additional Kik features include Kik codes (a QR-like 2-D bar code) and custom keyboards with text and picture responses. Kik also has its own SDKs for Python and Node.js.
Best frameworks to use
You’ve got lots of choices when it comes to frameworks. Three of the most powerful bot-building frameworks offer Kik integration: Amazon Lex, Microsoft Bot Framework, and Dialogflow. Kik even has blog articles all about using Dialogflow and Amazon Lex to build a Kik bot.
Kik chatbot examples
For an example of a just-for-fun Kik bot that epitomizes teenage tomfoolery, consider Zo. Per the bot description, Zo “roasts your friends, plays games, and tbh…has some serious chat skills.” The same goes for the Truth or Dare bot, which leads users and their friends in a game of—no surprises here—Truth or Dare.
Among the larger brands using Kik is J-14, which actually has two bots on the platform—one for news and another that shares articles and images. J-14 is using the data it collects from the bots to develop editorial strategies for its readership.
To help teens find emotional support in times of need, the Kik bot Koko helps users seek anonymous counsel and/or provide help to others who need it. Koko is part crowd-sourced therapy and part crisis- management tool. Kik can even direct possibly at-risk users to the bot and invite them to converse with it.
Know your chatbot’s audience
Know your audience, and deploy your bot where your users already are. Those are the most important things to remember when you choose a messaging app for your bot.
The apps discussed in this guide have a lot in common with one another. They’re places where people go to communicate with friends and relatives. Most of the apps have robust developer tools, framework integration options, and helpful features. There’s much that unites them and little that divides them, except for user demographics.
When choosing an app for your bot, keep in mind your users’ physical locations, languages, and age ranges. And remember, you can always build a new bot for a different platform or deploy the same bot to multiple platforms to reach as many users as possible.