Creating a Chatbot for Your Business: Why and How
For Devvela team, 2017 will be remembered as a year when we became deeply engaged in chatbot development. We finished works MessagePro, a chatbot for Slack that we’ve been building from the ground up, talked with many small and midsize business owners, and even developed our own chatbot builder that joined a startup accelerator program the same year.
The major conclusion we made from researching the subject from the inside? Put it in a nutshell, the future of customer service belongs to bots, and now to the good news: this future hasn’t yet come. Business owners still have a chance to learn about how this technology can help them solve their business goals, discover new ways it could be applied to their specific case and take steps today to make the most of it tomorrow.
While talking with business owners, we had a chance to meet the lucky ones who claimed they were doing well without all the tech stuff like chatbots, social features, their own mobile apps or whatnot. Well, their situation is enviable. If you haven’t yet found your own way to one-up the competition, we suggest taking a closer look into the chatbot technology.
Today’s notable chatbots
Today, chatbots are introduced in popular social platforms and messaging applications, various mobile apps, mail clients and enterprise software. They adopt the role of online consultants on many websites.
The chatbot boom first started in 2016, when the technology got noticed by industry giants. Some of the most notable chatbot examples include:
- Tacobot that helps TacoBell customers order their meal
- The Guardian bot delivering latest news
- Judy, a bot by Disney that you can chat and play games with
- HealthTap bot you consult with about health issues
- Amy and Andrew Ingram bots that help you schedule meetings and organize communication with humans
How chatbots are changing the game for businesses
Chatbots aren’t just a fancy feature. Businesses make chatbots part of their processes to achieve very specific goals, and already have a proven success record. Let’s see some of the most prominent examples of how the technology has helped companies improve their figures.
- Sephora Reservation Assistant: Sephora is known as a global prestige beauty retailer. They decided to take their social media marketing to the next level by introducing the Sephora Assistant, a Facebook Messenger bot. It made the process of booking makeovers at standalone locations across the US easy and fun, which resulted in an 11% increase in their booking rate. Sephora also claims to have seen an average incremental spend of over $50 from clients who have booked an in-store service.
- CNN chatbot on LINE: The media giant experimented a lot with chatbots on several social platforms. Alex Wellen, CNN's senior vice president and chief product officer, claims that the company have seen “meteoric growth” on the LINE messaging app. “We are now the No. 1 news publisher on that platform, surpassing longstanding brands like BBC, the Wall Street Journal, BuzzFeed, and The Economist”, Wellen said in April 2016.
- Aviasales bot: Russia’s top travel planning company, Aviasales introduced a chatbot across several messaging apps in 2016. The bot helps users find and book cheap flights in all directions. At the moment, the company claims to have 34,000 bot users and 9,663 sold flights. This demonstrates a conversion rate close to 30%.
Should you create your own chatbot?
Enterprises, as taken as an example above, have lots of resources and face stiff competition. This is why it’s no surprise that they are the first to tap into the new technology, test it and reap the rewards. If your business has a success story, a huge audience, and is struggling to hit the next level of client engagement then introducing a chatbot could be a solution.
Experiments with new technology always implicate extra expenses, which isn’t such a problem for enterprises, as opposed to small and midsize businesses that often operate on a tight budget. If the latter sounds relevant to you, we recommend analyzing your communication with leads and customers before you make a chatbot part of your sales process. Sometimes, introducing a chatbot could be a more efficient way to close gaps in this process, rather than using human resources.
Three types of chatbots
You must have already had come across a chatbot, haven’t you? Chatbots were invented to imitate humans on the Internet in what recently only humans could do: text communication. In other words, chatbots handle human requests for information, give feedback and, in more advanced cases, process the text information received from humans to take actions accordingly.
The two main functions of chatbots are identifying user intent expressed in a message and triggering processes meant to satisfy it. Each of the functions represents a certain challenge to developers, and can be achieved in multiple ways. Based on technologies applied, chatbots can be broken down into the following three categories:
Simple chatbots identify user intent by offering a user with a detailed guide that comprises a command, a word or word combination for each possible intent. This means that it depends greatly on the user whether their intent is identified correctly or not. Such a chatbot selects its responses according to the commands entered by user and hard-coded rules.
Pre-set communicator bots
Advanced chatbots communicate with users very much like humans. They search for key-words (not to be confused with keywords used in SEO) in a message to identify user intent. The intent identification accuracy depends on how rich and up-to-date the pre-set internal synonym base is. Finally, such a chatbot asks user to confirm whether it got their request correctly.
Automated communication is a domain where artificial intelligence excels. AI-powered chatbots are sometimes very hard to distinguish from human assistants, because they learn constantly using the technology known as machine learning. However, this is a complicated technology, and pitfalls lie in wait for unwary developers. Remember the story of Tay, the Microsoft chatbot on Twitter which has quickly absorbed the darkest part of human nature? You may want to make your own bot intelligent while keeping it manageable.
How to build a chatbot
If you have come to a conclusion that a chatbot could help you achieve your business goals, you must already be asking yourself how to build your own bot like this. We suggest the following four options.
Option 1: Building from the ground up
The simplest, command-operated chatbot can be created by a developer using the open APIs of social platform and messaging apps. The development process of such a bot comprises the following four steps:
- Setting up a list of keywords that will be identified as commands
- Setting up a list of responses
- Setting up associations between commands and responses
- Integrating the chatbot with preferred messaging app or social platform
Creating a chatbot from scratch is a viable option for you if:
- Your goals can be achieved with the limited functionality of a command-operated chatbot
- You can hire a back-end developer for this task
- The messaging app or social platform you are targeting has an open AP
Option 2: Using visual chatbot builders
The growing popularity of the technology prompted developers to design user-friendly tools to create a chatbot without coding. These applications are managed through a visual interface, which makes the technology available to users who are not tech-savvy.
Many of the chatbot builders utilize machine learning and allow you to train your virtual assistant using a simple question-answer system. In addition, they store the interaction history which can be analyzed to further improve the performance of your chatbot.
The only drawback of chatbot builders is that the range of platforms and messaging apps with which they can be integrated is often quite limited. For example, if you want to create a Messenger bot, you will have to take time to find a builder that supports Facebook Messenger.
Using a bot builder platform is your choice if:
- you can’t or don’t want to hire a developer
- you have time or human resources to learn about creating a chatbot with a bot builder
- there is at least one builder that supports the social platform or messaging app you are targeting
Below are some of the visual builders that allow you to create your own chatbot without coding:
Option 3: Using chatbot platforms provided by messaging apps and social media
Most of the popular social networks and messaging apps provide their own development platforms to help you create your own bot. The main problem here is that you will want to hire a developer to write the logic of your chatbot from scratch and integrate it with the social platform or messaging app using its AI. In exchange, you will get a plethora of useful features.
Chatbot platforms are a good fit for you if:
- you can hire a chatbot developer
- you want a wide range of messaging features which include file exchange, photo exchange, emoticons, customizable layout, and more
- you may want some specific additional features provided by the social platforms, such as standard webview, allowing users to load webpages inside a message platform
- you want to enable purchases with your chatbot
- you want to use the chatbot not only on the social platform or messaging app, but also on your website, as well as to utilize it in your ad campaigns
The following social media and apps provide a platform to create your own bot:
Option 4: Using AI-powered platforms for chatbot development
IT behemoths providing AI as a service (AIaaS) are keeping up with the trend, knowing how to make a chatbot more intelligent. IBM, Microsoft and Pandorabots provide tools to create your own chatbot for each and every popular messaging app and social platform, such as Facebook, Slack, Skype, Twitter, and Telegram, and even for your own app, using open APIs.
What makes these platforms unique is additional libraries, algorithms and functions developed by the companies’ programmers. This enlarges the range of chatbot features with natural language processing (NLP), language understanding and interpretation, such as provided by Microsoft’s LUIS and Pandorabots’ AIML, speech recognition (see Cortana by Microsoft), automated translation and online search, such as powered by Microsoft’s Bing API. Another advantage of the AI-powered platforms is that the IT enterprises that have built them can afford to create software development kits for multiple programming languages thus sparing you the problem of finding developers specifically for this or that chatbot framework.
Creating a chatbot on an AI-powered platform is a relevant option if:
- you can hire a developer to build your own bot
- you want to build an AI-based conversationalist bot capable of modeling large and complex flows with the mechanism of intents and contexts
- you want some special features, such as speech recognition or online search
- your specific needs require training your bot by example with machine learning
- you are targeting the widest range of social platforms and messaging apps possible
Below are some of the AI-powered platforms for chatbot development:
There is one last thing to say about introducing (or not introducing) a chatbot. It is clear that the technology is meant to provide a competitive advantage for your business, but there is something else. Before you come up with the final decision, whether you choose to tap into the technology or not, you will have to take a closer look at the communication processes between your business and your clients. This may help you discover growth opportunities that you haven’t yet been able to see.