There’s no question that chatbots are big business. Facebook Messenger hit 100,000 bots this April (a 3x increase since the previous September), and research from Digitas LBi found that one in three Americans would be willing to make a purchase from a chatbot.
As if that wasn’t enough to convince you, 51% of people say that a business needs to be available at all times, 45.8% would rather contact a business through messaging than via an email and 49.4% would rather use messaging than a phone call.
One of the big reasons for the success of chatbots is that they allow businesses to create an automated branded experience that adds real value to the end user. When deployed correctly, chatbots aren’t just some marketing campaign – they’re an extension of the business that adds value to everyone, and to the end users of the bots in particular.
But despite this, there are still plenty of myths and misunderstandings about chatbots that stem from a lack of knowledge. That’s why we wrote this article. We want to dispel those myths and to open your eyes to the true potential of the chatbot revolution. Prepare yourself.
What are chatbots?
Chatbots are a service application that uses a conversational interface like Facebook Messenger or Slack. They’re often designed to automate simple tasks and to allow real people to get real information through the form of a conversation, although some chatbots are designed to be more conversational and entertainment-based.
As artificial intelligence continues to grow (it’s set to double annual economic growth rates by 2035), chatbots have started to surface as a dominant tool for businesses. Chatbots are AI services that interact with people through text-based conversations. As the conversations develop, a chatbot can offer everything from information on current weather, traffic, and sports scores to much more in-depth problem-solving solutions.
Building a voice for your brand has never been more important than right now. With the use of bots, you can develop a voice that matches your brand while providing genuine value to the end user. You can even allow your bot to use emotions to provide a more realistic experience. Better still, as the field of AI continues to advance in line with Moore’s Law, the ability of bots to experience emotions and to hold conversations will only improve.
Why chatbots are such a big opportunity
Global mobile web browsing has overtaken desktop browsing for the first time in history, but despite this ongoing shift towards a mobile-first society, many companies have witnessed a decline in mobile app usage. At the same time, people have started to turn to Facebook Messenger and other chat-based interfaces to report customer service issues or to receive information about products and resources.
By now, we’ve all heard stories about people turning to social networking sites to report problems. We’ve also heard our fair share of horror stories about companies that failed to address issues and ended up with a backlash on their hands. After all, 70% of consumers have used social media to report customer service issues and 60% of people expect a response within an hour.
Chatbots, then, come in useful for a wide range of use cases, from solving basic customer service issues to acting as an interactive marketing campaign that sends special offers and personalized content based upon the people that they’re talking to. They allow businesses to go where their customers are, which is why many companies are abandoning their mobile apps and seeking alternatives.
This is unsurprising, because the majority of smartphone users download zero apps per month, and even when they are downloaded, almost a quarter of people only use an app once before deleting it. Six million companies now have an app, which might sound like a lot, but over sixty million are currently active on Facebook Messenger alone.
Put simply, app usage is on the decline and Messenger usage is on the increase. In fact, 11% of the world’s population use Facebook Messenger at least once a month, and it recently hit 1.2 billion active users.
But don’t make the mistake of thinking that Messenger is the be all and end all, even if it does have four times as many active users as Twitter. There are some fantastic use cases for bots on all different sorts of platforms, and Facebook Messenger is just one of them. On top of that, the amount of value that developers can both offer and receive isn’t necessarily related to the number of users that a platform has.
How chatbots are used
By now, you should have already seen some of the different ways that chatbots can be used, and with 30,000+ Facebook Messenger bots and 100,000 Facebook Messenger chatbot developers, you can bet that there’s a huge variety of functionality already out there.
Loosely speaking, though, there are four main types of chatbot. Let’s take a look at them.
As the name suggests, utility chatbots are designed as a tool to facilitate the sharing of information. They often await input from the user and then respond with the answer to a query, a little bit like a text-based version of Siri or Alexa. They’ll process the query that you give to them and return a result, whether they’re sharing sports scores and live news updates or whether they’re performing calculations or running search queries.
Example: Poncho is a weather application with a difference. Available on a range of platforms including Messenger, Slack, Kik, and Viber, it sends a personalized weather forecast every morning while throwing in the odd joke and offering tongue-in-cheek dating advice.
Marketing chatbots are generally designed to provide a branded experience for consumers while either giving away freebies and special offers or while providing some sort of entertainment that keeps people coming back for more. In the last couple of years, the chatbot landscape has been increasingly dominated by marketers with big budgets who have seen the potential of the technology to sell a product.
Example: Pizza Hut now accepts orders through Facebook Messenger and Twitter chatbots that are designed to make it as easy as possible for people to order food. The Messenger bot even allows people to connect the app with their Pizza Hut account so that they have access to their order history and can receive personalized offers and recommendations.
Communication chatbots can be further divided into two subcategories: public and private. Private communication bots typically run on internal social networks and other closed-loop systems with a limited number of users, while public ones are available to anyone. Either way, though, these bots are used to facilitate communication, whether that’s by sharing automatic alerts or whether that’s by streamlining customer service interactions by providing templated responses and troubleshooting information.
Example: Slack’s inbuilt Slackbot is a great example of a communication chatbot, which is unsurprising considering it’s the flagship bot of a company that’s entire business model is based around communication. Users can set reminders for themselves and for their co-workers or take advantage of other tools like automatic responses.
Chatbots don’t necessarily need to provide people with information or enhance communication. Sometimes they’re just there to get a giggle or to help people to while away the hours. Entertainment-based chatbots are often created in the form of a game, such as a trivia bot, and they’re also often created by individual companies for marketing purposes. They differ from traditional marketing bots, though, because their overall goal is to generate engagement as opposed to pushing for a sale.
Example: A great example of an entertainment chatbot is, Narcos. The Narcos Messenger chatbot is a games simulator that plays off of the Netflix original series.
How to build your own chatbot
If you think that the ability to build a chatbot is exclusive to high-end developers and tech geeks then think again. The good news is that building a chatbot isn’t too difficult, as long as you take some time to familiarise yourself with the basic concepts. Once you’ve done that, you can start to use one of the bot building services that are out there on the market.
Here are a few of our favorites.
Chatfuel is all about helping you to build a Facebook bot without needing to code, and they promise to allow you to launch a full-featured chatbot in just seven minutes. You can get started on the service for free and you need zero development experience to understand their intuitive user interface. If coding frightens you then this is the service for you, although you make some sacrifices when it comes to functionality.
ManyChat is Chatfuel’s biggest competitor, and the two of them are similar in many ways. They’re both dedicated to making the bot creation process as easy as possible, but ManyChat is more service-driven with a goal of helping you to drum up sales and to offer support. Most people get started with either Chatfuel or ManyChat before moving on to other platforms.
This is another tool for building an interactive Facebook app, and the developers promise an average clickthrough rate of 40% with no coding necessary. What’s particularly interesting is that while Octane.ai does cater to companies, it’s often used by musical artists, reality TV shows and other high-profile celebrities and influencers, so it may be the best choice if you’re promoting a company figurehead or focusing on personal branding.
Botkit.ai touts itself as “the leading developer tool for building chatbots, apps and custom integrations for major messaging platforms”, and it’s mainly focused on developers and enterprise bots on internal communication tools like Slack. The solution is essentially a downloadable toolkit with different code samples that can be used to develop a custom solution, but it’ll take a lot more programming than other options. On the plus side, it’s almost infinitely more customisable.
Don’t worry if none of these options take your fancy, though. There are plenty of other ways to develop a bot, and a great place to get started is by browsing through the options at BotList. You can look through the different categories to find the right tool to support you, whether you’re building a trivia bot or developing a custom solution for customer service.
Best practices for building your own chatbot
When you’re building your own chatbot, you’re given a certain amount of freedom. After all, while each individual platform has its own unique quirks and constraints, you’re pretty much limited to your own imagination.
That said, there are still some best practices that you’ll want to consider, no matter what type of bot that you’re building. It all comes back to user experience and creating a bot that can hold its own against any human. The more three-dimensional and believable the bot is, the better the experience.
Here are just a few of the things that can help to make a successful bot.
While it’s certainly true that most bots have an ultimate purpose – whether that’s to provide information or to make a sale – that doesn’t mean that you can’t give them some personality to make the journey from start to finish more fun. We’re not saying that you need to turn them into a cartoon character, though – just that it helps to write the script in such a way that it comes alive to users and feels more like they’re talking to a human than to a robot.
There are several ways for you to communicate your bot’s personality, but one of the easiest ways is to look at their language and the way they talk. At its basic level, does it say “hi” or “hello”? And at a more advanced level, how does it string sentences together and how does it ask for information? Does it tell jokes? Are there any easter eggs (i.e. secret responses to a specific input)? Consider all of this at an early stage before you start to write the script because it’s easier to get it right early than to go back and to have to rewrite it all.
There are always going to be occasions in which your bot is unable to understand a user’s input, and that’s okay. You can plan ahead by providing your bot with a range of responses — such as “I don’t understand” or “please can you rephrase that?” — that are called into play when it’s struggling to follow the conversation.
You also need to consider every possible path that a conversation could take — so if your bot asks someone to accept terms and conditions, you need to plan what happens if they say no, as well as when they say yes. You can’t just design one conversation and hope that users always follow the path that you want them to follow.
Have multiple variations
If your bot always uses the same phrases to respond to users then it’s quickly going to lose its charm, especially if you’re hoping for people to come back often and to use your bot again and again.
You can combat this user fatigue by creating multiple variants of the same sentiment and serving them up randomly or at certain times. For example, instead of the bot always saying “hello”, throw in the odd “hey”, “good morning” and “hi, how are you?” It might seem like a minor tweak, but it has a huge impact in the minds of users and can make your bot seem much more sophisticated than it actually is.
If you really want to build a brilliant bot then you’ll need to use analytical software to monitor how people are using it. With this data in hand, you can make changes to the bot over time to further improve it. If you’re building a marketing bot then you can even deploy A/B testing, calls-to-action and other marketing techniques to make your bot more profitable over time.
Analytical tools can also help us to identify bottlenecks or other problems with the bot that are stopping it from living up to its full potential. It also doesn’t have to be too hard to get started, so read on to discover some of our favorite tools for monitoring and managing chatbots.
Analytics tools for your chatbot
Without installing analytical software, you’ll never be able to gauge the ultimate effect that your chatbot has had or to identify exactly how people are using it so you can further improve it in the future. If you’re using your bot for marketing purposes then analytical tools are even more important because you’ll want to test and refine the bot and its calls to action over time to improve conversion rates.
There are two main tools that we recommend to get started and each of them has its pros and cons. Let’s take a look.
Dashbot.io’s analytical suite claims to help users to “increase user engagement, acquisition and monetization through actionable bot analytics”, and the real-time counter on their homepage shows that they’ve processed over 11 billion messages. It supports Amazon Alexa and Google Assistant as well as Facebook Messenger, Slack and other platforms, and the useful charts that it creates are easy to take in at a glance.
Perhaps the best thing about Dashbot.io is that it can handle unstructured data and can process images and audio, but it can also use conversational analytics and other bot-specific metrics to really take you inside the hood of your chatbot. It’s also easy to use, which makes it a pretty good choice for first-time developers.
Botanalytics is like a more accessible version of Dashbot.io, supporting some additional platforms and using a slightly different approach to combine real-time and historical data to answer the questions that are really bugging you.
When it comes to further developing and improving your bot, Botanalytics won’t let you down. One of its most useful features is the way it allows you to identify bottlenecks and to find steps in your bot’s development that cause users to abandon it. That will allow you to make some changes and to track the effect that they have in real-time. You can even manually jump into the conversation with a human rep, opening up the ability for a hybrid bot that passes the baton if it’s struggling to complete a task.
Other advanced features for your chatbot
As with most technological fields, the pace of innovation in the chatbot industry is immense. People are coming up with new uses for chatbots at the same time as technological developments allow them to scale faster or to make smarter use of new and existing data.
It’s a fool’s game to try to predict exactly how chatbots will develop in the coming years, but there are at least two trends that we can point to that give us an idea both of what’s capable and of what’s potentially inevitable.
Natural Language Processing
Natural language processing refers to the ongoing attempt to create software that can understand language in a uniquely human way. When we talk, there are subtle nuances that change the meaning slightly and other indicators of not just what’s being said but what the speaker’s actual intent is.
On top of that, language has little quirks like homonyms (multiple words that share the same spelling or pronunciation) which can be hard for traditional bots to understand. For example, the word ‘sole’ can mean both ‘only’ and ‘the bottom of the foot’, and when it’s picked up using voice recognition it can also mean the spiritual, immortal part of a human being or an animal.
All of this makes it difficult for bots to understand people, which is why more and more people are trying to find new ways to process natural language and for bots to listen and speak to users more like a person and less like a robot. The end result is a bot that’s so convincing that people might not even realize they’re not talking to a human.
Artificial intelligence is essentially the process of creating a piece of software that’s so human-like that it can be said to be artificially intelligent. Many people cite the Turing test, which was developed by genius mathematician Alan Turing, as the definitive way to determine whether a machine has true artificial intelligence — and it largely involves judging the natural language conversation between a human and a piece of artificial intelligence that’s designed to respond with human-like responses.
The interesting thing about AI is that it has implications across all sorts of industries, because bots could be created to crawl through huge amounts of data and to arrive at conclusions that no human could ever hope to achieve. They can carry out calculations and analysis much more quickly than a human can, effectively turning ‘big data’ into just plain old data.
The future of artificial intelligence is likely to be machine learning, which is effectively a branch of AI research which involves allowing the software to effectively ‘teach’ itself by arriving at and storing its own conclusions about information that isn’t even mentioned in the source code. It’s pretty cool stuff if you’re a researcher or an innovator, but it could be considered overkill to play around with AI and machine learning if you’re just using your bot to sell products or to file support tickets.
Marketing Your Chatbot
It’s important to have an idea about your audience before you begin building your chatbot. You should understand the problem your chatbot is solving and who your ideal user is going to be. One of the best strategies used by a lot of people is “scratching your own itch” and creating a chatbot that solves an issue you personally have or deal with on a daily basis.
Slack, for example, has a chatbot that helps new users understand how to use the platform. Thanks to Slackbot their customer support probably have fewer people contacting them, thus giving them the freedom to focus on more important tasks.
If you can understand the exact problem you are trying to solve with your chatbot everything else in marketing becomes a lot easier.
Create a great Chatbot
Just having a great product won’t guarantee you millions of users but that’s a good starting point as a great product is always easier to market and will help you stand out from the crowd.
A/B test your chatbot, share it fellow bot creators and potential end users to find any potential features or issues that exist that you may have missed. Having a community behind your chatbot telling your their pain points and suggestions will help you improve your chatbot.
Don’t be afraid to share an imperfect product, perfectionism is usually the death of most creators. Just ship! The community will respond and voice their opinions and you will learn a lot in that process.
When BotList started out, we weren’t perfect (we still aren’t as we are always learning and growing). When we were hunted on Product Hunt we were the #1 product on the day and the community gave us feedback which we applied to get better and we later launched BotList version 2.0.
Being the best isn’t always about have the most features, factors like ease of use also matter to the end user.
Market Your chatbot
Advertise on BotList.co
“Advertising on BotList directly impacted the growth of our chatbot, Obie. Our listing drove traffic, installs and brand awareness that is critical at our stage in development. The team at BotList was super friendly, accommodating and open to feedback. My experience has been extremely positive and I’d recommend other bot makers to showcase their solutions on BotList.”
– Chris Buttenham, Co-Founder of Obie.ai
We have recently launched Bot Makers on BotList, a platform for bot makers to showcase their work and also put themselves up for hire!
Adding your bot to Facebook’s Messenger Discovery section will widen the reach of your chatbot. Facebook is also featuring Bots with high user retention and increasing daily active users to its messenger users within the app.
Use social media and leverage your audience on Facebook, Instagram, and Twitter to drive traffic and users towards your bot. Facebook groups related to Bots will provide you with a lot of beta testers, to begin with.
Groups related to your niche will be great for growing your users over the longer term. Eg if you bot is related to marketing, get in Facebook groups around digital marketing and form relationships. Don’t just spam, instead let people know what you are trying to do and the problem you are trying to solve.
Offer a freebie as an incentive to get people to sign up and try using your chatbot, this could be in the form of an ebook or a course or even a checklist.
For example, we at BotList use the GrowthBot by Hubspot in our slack group and we love the features it offers. To get new users, Hubspot could offer a free ebook on what they learned marketing over the past 5 years and it’ll incentivize the type of people who would use GrowthBot to sign up in order to get the ebook and give GrowthBot a whirl.
Don’t neglect SEO when it comes to your chatbot, create quality content in your niche to drive targeted traffic to your site.
When starting out, don’t compete for the highly competitive keywords, for example instead of trying to become the first result for “marketing chatbot” your resources will be better spent optimizing to come up for “analytics chatbot” or more specific keywords like “marketing chatbot for e-commerce”.
There are 200+ factors when it comes to SEO, however, are 3 factors that you shouldn’t ignore in your marketing strategy.
- Content: The longer the better as long as you are not giving are on quality. Don’t write longer content just because of SEO. Write longer content because you have something to say.
- Backlinks: Links still matter in SEO and getting your bot featured on many websites in your niche is sure to improve its ranking on Google’s Search Engine results.
- Value: There are so many other factors like Rankbrain (a machine learning AI technology) ….. to the title and heading tags etc. However focus on first bringing value to your readers with the content, solve problems and later dive into the more technical factors like heading and title tags.
SEO is a long game however it will bring in recurring targeted traffic and potential revenue to your business.
What’s next for Chatbots?
By now, you should be convinced of the potential that chatbots have for disrupting the way we do business, whether we’re marketing new products to customers or whether we’re using chatbots to solve their problems and to respond to customer service issues.
It’s also clear that bot usage is likely to continue to grow, especially while Facebook Messenger and other chat and instant messaging platforms continue to grow in popularity. With many people in developing countries accessing the internet for the first time through their mobile devices, it also offers an opportunity for companies to tap into emerging markets.
Meanwhile, bots are continuing to get smarter, faster and ultimately more useful, and venture capitalists and other investors are pumping more and more money into the field. New advances like natural language processing and AI will continue to push bot usage even further and only time will truly tell the ultimate uses that society is able to find for them.
One thing is for sure, though. The world will never be the same again, and the people who are the first to dismiss bots as just as a fad are the same people whose companies will eventually fade away and be forgotten.
Make sure that you’re not among them.
- Building a brand voice is more important than ever
- Bots are getting smarter thanks to advances in AI
- Chatbots allow brands to speed up customer service
- App usage is on the decline, but Facebook Messenger usage is on the rise
- Chatbots have four major uses: utility, marketing, communication and entertainment
- You can build your own bot using Chatfuel, Manychat, Octane.ai, Botkit.ai and other tools
- Give your bot personality, use fallbacks, have many variations and use analytical data to further improve
- Dashbot.io and Botanalytics offer reasonably good analytics suites
- Natural language processing and artificial intelligence will push the industry even further
- Chatbots are getting smarter, faster and more useful
- Don’t get left behind!
Thanks to our SEO consultant Jeremy Noronha for contributing to this article.