How to harness the power of technology to automate your bran
Chatbots can help you save time, money, and precious staff resources, and can be deployed to assist with nearly everything, from online commerce to customer service. You might worry that directing your customers to a bot instead of a live person may put them off, but bots can actually be a great way of keeping users on your site and retaining their interest.
Research has discovered that 69% of customers favor talking to a chatbot over a real person, primarily because of speed and efficiency. If you want to use bots to your best advantage though, you first need to make a structured plan which addresses key elements you must assess before you can create a bot user will engage with. Then, you need to build and create a bot suitable for deployment on your chosen platform, such as Facebook, your website, or Skype, using quality tools that create an engaging and helpful user experience.
Finally, you’ll deploy your bot on the specific platform of your choice and test it on a small section of target users to see how you can improve it before you make it public. In this guide we’ll be talking through the main points you must consider if you want to create an effective bot that can smartly automate part of your business.
ChatBot Canvas – Start here to create a high-functioning chatbot
Chatbot Canvas is a superb place to begin if you want to build an effective chatbot that will help automate your marketing or give customers useful info. It presents you with a colour-coded grid that lays out your chatbot design and prompts you to answer helpful questions that determine the specifics of your bot.
ChatBot Canvas can be used as a prep step before you create your bot, or sit down with a development team, and is an invaluable tool as it helps you make a clear and structured plan and allows you to get granular on every element of your bot.
Let’s look at some of the key elements covered by ChatBot Canvas.
Before you start creating your bot you need to understand why you are building it. First define the main reason your chatbot will exist.
- Do you want it to assist customers with their online shopping or give users updates on the stock exchange?
- Does it need to take bookings, or answer customer questions about your products?
Once you know what your bot’s primary function will be, it’s time to delve deeper into its specifics.
- What problem are you trying to solve?
- How will your chatbot help your users or customers?
- What kind of features does it need to save your business time or money?
- How can it help get customers the information they need faster?
Answering these questions will help you form the basis of your value proposition and will aid you in structuring and configuring your bot going forward.
After you’ve decided on your value proposition, you need to consider your users in more detail.
- What type of people will be using your bot?
- Are they business users or consumers?
- What demographic do they come from?
- Where are they located?
- What are their interests or core values?
The answers to these questions will help you create a useful bot that is best suited to your target audience.
It’s vital to consider what language your users speak, as chatbots are conversation based. You’ll want to build a bot that uses language naturally, like your users would. It’s not as simple as merely selecting a language or regional dialect though, you also need to ensure that the programming for your bot accurately reflects how your customer’s converse.
You’ll likely want to give your chatbot a personality to make it seem more natural and create a better experience for your users. To decide on an appropriate personality, assess your target user, to ascertain what type of persona and conversation style would hold most appeal. Consider naming your bot or giving it a visual avatar or face where appropriate. Think about whether a formal conversation style or a more casual one would be better suited.
To give your bot a sense of personality and keep users engaged, you can make it do unexpected things. Would your users enjoy jokes, or funny anecdotes, or would they better appreciate helpful facts or trivia about your brand?
As well as your users, you also need to think about your bot’s owners and administrators, as they will need to access your chatbot to configure or change settings, as well as check on and monitor performance. Status reports will be crucial to understanding how to tweak your bot’s performance so it’s more user friendly, so think about how you can make administrator access as easy to navigate as possible.
You need to understand how your users will access your bot, the types of devices they own, and their likes, dislikes, and demographic info.
- Will they be mostly accessing your chatbot from a mobile device like a smartphone or tablet?
- Are they business users who will be using it from office desktops?
- Perhaps they are work from home entrepreneurs, who are likely to connect from a laptop.
Remember you can also make use of voice bots for landline users, so even if your demographic isn’t online, you can still take advantage of automated assistants.
New devices like Google Home and Amazon Echo allow for even more possibilities, enabling you to create bots that help users in their homes without them having to go through a smartphone, tablet, or laptop.
Once you know which devices your customers will be connecting to your bot on, you need to look at the channels they use most, so you can pick the best one to deploy your bot on. The channel or channels you choose will vary depending on the nature of your bot, what it’s intended to do and who it’s created to help.
Many businesses create bots on platforms like Facebook Messenger, Skype, or Slack, but these are not your only options. You can also create bots on platforms like WhatsApp, or Viber, or integrate them with AI using Google Assistant, and you can even deploy bots on Telegram or Twitter to respond to DMs on behalf of your brand.
Don’t ignore SMS, as you can successfully use bots here for marketing, taking bookings, or answering common queries. You can build voice bots for mobile and landline phones, or create email chatbots, which can automate part of your marketing or customer service.
Traditional platforms like these can be an excellent option if your primary target users are not tech-savvy but they are also good for locations where internet or computer access may be limited or less integrated.
Conversational Flow Design
One of the key requirements of an effective chatbot is its ability to hold a natural conversation. You don’t need to code this from scratch, as there are plenty of third-party platforms and tools you can use instead.
From open-source NLP (Natural Language Processing) platforms to FeedYou Designer and Mobile Monkey, there are a plethora of services available to construct a chatbot that can process and output authentic speech or text, and we’ll be taking a look at some of these services in more depth later.
If you are creating a voice bot, you can use services such as Amazon LEX or Google Speech to construct a natural-sounding bot that interacts well with your users.
Once you have chosen your platform of choice, there are some key areas you must address to perfect your chatbot and tweak its responses to best suit your target market.
After you’ve chosen your build-a-bot platform you need to create a conversational model that assigns tasks and responses to your chatbot. This will allocate words and phrases to certain tasks, that activate when the user enters them by text, or by voice if you are building a voice bot.
To create this model, you need to think about the kind of tasks you want your chatbot to perform.
A good tip is to organize your value statement into conversation topics and then segment these topics into tasks you want your bot to perform, which you assign trigger words and phrases too. For example, your value statement could be:
You want your chatbot to answer questions about products
Your topics could be:
- Product Specifics
And your tasks within specific topics such as Deliveries would be the specific questions such as:
- Do you deliver to the UK?
- How long does delivery take to the UK?
- How is my package tracked?
Tasks can be identified easily when you do it this way as you can see what you will need your chatbot to do. For example, to answer a question about shipping rates, your chatbot needs to first respond to user input i.e. “How much is Shipping to the UK?” and then it needs to calculate the answer by accessing shipping rates information.
Third-party services such as FeedYou Designer will help you construct tasks by linking your chatbot to phrases and common words you input, and with the information services it will need to access to perform these tasks.
When creating your chatbot you also need to consider its modalities, or ways that users will interact with the bot. Basis bots primarily make use of button-based interaction, where users select from a choice of different buttons that activate certain responses.
If you want a chatbot that incorporates more natural language interaction, you can construct a bot that responds to keywords of phrases.
For bots that utilize speech-based input, you’ll first need to use a platform such as Google Speech, that supports this type of bot creation. Then you’ll use that platform’s software to link user speech input to tasks you want the bot to perform, such as accessing and delivering price information.
You might also want your bot to respond and understand image-based input, such as emoticons, and if this is the case you’ll want to select a platform that will allow you to configure this too.
As well as responding directly to user input, your bot needs to be able to simultaneously hold a conversation while performing other essential steps required to perform a related task. This may include multiple steps.
To deliver accurate seating information for example, your bot may need to first check the seating plan, then the booking information, and finally the price info, to let your customer know which seats are available in their price range and where they are located.
Other bots, like shopping bots, may require many more steps, as they may have to search for a product, find product information and availability, create a cart, and take payment. Lead generation bots will need to liaise with CRM platforms like HubSpot to create lead enquiries and send email’s, while talking to the user at the same time to keep them engaged.
Simpler bots, like weather bots, may only have one or two tasks to perform while chatting, such as accessing the latest weather updates in a specific region.
Whatever the number of tasks your bot needs to perform, you should always keep user engagement at the forefront of your mind, Along every step of the way construct your bot so it can carry out tasks while chatting with the user to keep them updated and engaged.
KEY TAKEAWAY: Create a list of back-end steps your bot needs to perform to carry out a task. Then connect those steps to the task and link that task to key words or phrases that will be inputted by the end user.
Tooling – Third Party Services to help build your bot
FeedYou /designer is an excellent place to start building your bot as it doesn’t require any coding experience. This cloud-based chatbot creation program uses an AI-powered composer to auto-create natural style conversation based on specifics you input.
You can use FeedYou Designer’s inbuilt customization model to create NLP models tailored to your unique company requirements and can tweak and adjust them over time to suit your needs. You can also make use of the multiple pre-created NLP templates that have been developed for use within industry sectors, such as business, shopping etc…
Mobile Monkey is a free tool that uses a visual bot builder to construct useful bots for Facebook Messenger. It’s ideal for non-technical users, as like FeedYouDesigner, no coding is required, and it offers a range of customization abilities, including Chatbot Templates, Custom Attributes, Q&A Triggers, and Live Chat Takeover options. Mobile Monkey offers you the option to send out messages to multiple users at once, via its ChatBlast feature, and gives you the ability to upgrade to access even more features, like scheduling, analytics, and the ability to set up drip style campaigns.
Other visual bot builder platforms
Mobile Monkey is one of several visual-style bot creation platforms, available, and is specific to Facebook Messenger. There are other visual style bot builders you could use to quickly construct bots for alternative platforms, such as Chatfuel, Motion.ai, and FlowXO. All these platforms require zero coding knowledge and are a great way to quickly build simpler bots that don’t need to depend heavily on processing natural language.
Many of these bot-building platforms now allow users to connect to live customer service agents to their bot or configure it to be able to access other information services. They are also typically easy to integrate with the deployment platforms you will be rolling your bot out on, such as WhatsApp, Facebook, and Slack.
NLP based platforms
NLP-based platforms allow you to make full use of natural language, enabling you to create a more complex bot, that offers your users an immersive and authentic experience. Tools such as Google DialogFlow, Haptik, IntelliTicks, and Wit.AI make use of Artificial Intelligence that learns from user input. They offer a wide array of features, such as the ability to create text and voice bots, and can be integrated with mobile wearables as well as Google Assistant, Cortana, and Facebook Messenger.
Some of these platforms are free, like Wit.Ai, others offer both a free option and the ability to upgrade while some may require a paid, upfront subscription.
Facebook Tips for Chatbot Creation
Define your main goal
First, you need to prioritize your Messenger bot’s main aim. No one Messenger bot can be programmed to carry out every task, so you need to decide what is important. Choose one main goal for the bot to handle, then break it down into sub-tasks to plan out and create your bot’s architecture. For Facebook Messenger, you might want a bot that provides 24/7 customer service to users, or you might need a bot to identify quality leads or promote your products when your live agents aren’t present.
Let your customers know what to expect
Facebook insists on transparency, so tell your customers what to expect from your Messenger bot so they don’t get frustrated or give up. Your bot won’t be perfect and there will be certain things it just won’t be able to do, so make clear how it can help and where it might fall short. If you’ve built a bot to check for seating, for example, let your customers know it can help with booking in this way, but inform them that if they have very specific or unusual requests, they may have to talk to a customer service agent.
You can program your chatbot to keep customers informed as it guides them through every interaction. Bot-building platforms for Messenger like Mobile Monkey allow you to connect your bot to a live agent or help page, if there are certain tasks it can’t perform, or if your customers need to talk to someone about special requirements.
Get to know your customers by inviting them to narrow down their preferences so your Messenger bot can deliver more accurate information. This is key to creating a bot users enjoy interacting with and will really help to drive engagement. If you are using your bot to qualify Facebook leads for example, ask your customers questions to assess their level of interest and help your bot connect them to the information they need.
Save users’ time
Does your bot make user’s life easier by helping them get information faster?
Can it cut out the middleman by booking their ticket directly or automating the online shopping process?
You shouldn’t just build a bot simply because you can but instead should create one that improves user’s lives. Saving customers time or hassle is a great reason for building a bot and should be at the forefront of your mind when you are constructing the tasks your bot will perform. Studies have demonstrated that users drop off when your bot is slow to respond or inefficient at retrieving the information they need. Bear this in mind when you structure your bot’s responses and tasks to make the user experience as helpful as possible.
Offer the option to connect to live help
Sometimes there will be occasions where users will need to speak to a live person, and you should always give them this option, or risk losing their interest. Your Messenger bot can offer up this ability by linking them to live customer chat, or by directing them to customer service lines automatically. Alternatively, you can program your bot to take help requests and send an email to one of your agents, who can then follow up on the request to answer your customer’s queries. Hootsuite allows you to integrate your direct messages in one place, so you can view and reply to all communication, including communication sent from Messenger chatbots.
WhatsApp Best practices for ChatBots
Sign up to the WhatsApp Business API
If you want to create a chatbot for WhatsApp, you first need to sign up to the WhatsApp Business API. This is a beta program with a current focus on SME’s and the sign-up process allows you to select whether you are an end user or a provider. To enrol, you’ll need to provide your company name and domain, as well as details of the company representative who will be using the platform, then submit your application to be reviewed. Once you have been approved you can begin creating your bot using the WhatsApp API.
To build an engaging experience for your users, you first need to think about the type of questions your customers will ask and the kind of information they will need. As well, you should consider the way your target audience talks and understands things so you can create a chatbot that appears as natural and seamless in conversation as possible. Don’t just think about the questions your chatbot is prepped to answer, think about how it will respond when it can’t help directly too.
Can you point your users to helpful resources that can assist them in finding the information they require, or connect them with a member of your team who will be able to answer their queries?
Save time with bot-building platforms
You don’t need to reinvent the wheel, when there are so many platforms out there that can help you create a natural, full featured WhatsApp chatbot. Applications such as Botsociety will aid you in creating a great WhatsApp bot and are especially suited since they can be hosted on a database. It’s a WhatsApp requirement that their Business API is hosted in this way, to guarantee secure encryption.
Test, Test, Test
Use your bot builder to create prototypes of your chatbot and then test them to see what tweaks can be made to improve. Ask your bot the same questions your customers will ask to assess how it responds, then makes any necessary adjustments to improve functionality. Don’t skip the testing phase, it’s vital if you want to build a high-quality bot that offers users a seamless experience.
Viber’s Best Chatbot Practices
To create a bot on Viber, there are some simple steps you must follow, to ensure you get a smooth-running experience. To start, you’ll need to make a free smartloop.ai account and then click on the Create New Bot Option. The platform will create a container linked to your subscription where your bot will deploy from.
Once you’ve created a new bot account, you’ll need to configure your bot. You can do this by starting with the basic building blocks, such as start, about, and default. Fill these in to get an idea of the structure, then configure the AI. Once you’ve completed these initial steps, return to the build tab, and select a plug in that you will build your bot on, such as Viber’s JSON API. This plug in lets you import and embed externally hosted data, saving you a lot of time.
Configure the JSON API to build and perfect your bots’ architecture, including setting custom headers, POST body content, user variables, and query string parameters.
The final step is to test your bot to see how it runs and responds, then hit the publish icon on the navigation bar to make your bot live!
Apple Business Chat / RCM Top Tips for ChatBot Creation
Creating a business bot on Apple can help your company connect with its user base round the clock, by taking bookings, completing transactions, and answering customer queries. Because of Apple’s reach and multi-device ability, you can dramatically boost your engagement, as you’ll have access to your audience 24/7 through a variety of devices, including Apple Watch’s, iPads, and iPhones.
To get started, register for an Apple Business Account, fill in your business details, and allocate roles, including your technical and administrative points of contact. You can choose to register either as a business or customer service provider, but must submit your application and wait for it to be approved before you can go ahead and start creating your Apple bot.
Select your platform
If you want to use Apple Business to create a bot, you’ll also be required to sign up with one of their chosen, integrated bot-building platforms. There are currently 18 to select from, including SalesForce, ZenDesk, and Quip, so pick the one best suited to your industry, audience, and requirements. When choosing your creation platform, consider the learning curve needed and the knowledge of bot building you or your developers already have.
Research and assess
To construct an effective Apple Chatbot, you need to consider your bots main purpose and how it can best help your users. Carefully consider conversation and flow, using insights gained from your market research, so you create a bot that your customers feel seems natural.
PRO TIP: Plan out a typical conversation that your ideal customer might have with your bot. As you go through, it will become clear what tasks need to be performed, and how your bot might respond to appear natural and keep your users engaged. You can then use this as a template for all questions your customers might ask, by making some simple tweaks and adjustments each time.
Make use of protype tools
Before going live, test out your bot using integrated prototype software, so you can identify any errors or weaknesses. BotSociety allows you to build a test chatbot for Apple that you can test on a select audience before you go public with your launch.
Microsoft Teams Best practices for Chatbot Creators
Microsoft Teams incorporates the Microsoft Bot Framework, giving you the ability to design bots of your own, or use bot building software if you don’t know how to code. It offers a wide range of options and allows developers to construct complex or simple bots, depending on their target user’s requirements. Teams Bots tend to be aimed at business users rather than consumers, with the goal of encouraging and automating collaboration in the Microsoft Teams environment. They can help you communicate with third party services through chat, without having to exit the Teams app, helping to keep you and your employee’s workplace activities all in one place.
What do Teams Bots do?
With a Teams bot, instead of having to switch apps, you can take part in an AI conversation instead, and have your request or command to a third-party app carried out for you by the bot. You can schedule meetings, send messages to personnel, and give briefings and updates using Teams Bots, making them an excellent addition to your Teams workspace.
Your users will interact with the bot you build through text, cards, and task modules, with exchanges ranging from simple search queries right through to complex, AI driven conversations.
Most Teams bots are geared to boost employee productivity and communication, by cutting out the need for app-switching and automating specific tasks.
Integrate with third party apps
Integrate and benefit from third party app functionality such as Zoom, or Creative Cloud and you can receive and send information without the need to switch from Teams. Creating a bot that liaises between Zoom for example, allows you to schedule, set up, or attend meetings direct from Microsoft Teams, by assigning simple bot commands to actions.
Initiate workflows from conversations
You can kickstart tasks and get information through Teams bots that you build and can customise them to suit your users. Try automating some of your low-level business processes using Teams Bots like calculating holiday time left, so your personnel staff can focus on more complex tasks. That way, instead of having to communicate with a time-pressed employee, your staff can consult the bot, who can perform a series of pre-programmed steps to calculate how much vacation time they have available.
Provide 24/7 assistance to staff without tying up IT
Help bots can be a superb way to get the most out of Teams, particularly if they save your technical teams time and resources. A bot that deals with IT requests for example, could cut down on the amount of time your IT staff spend by organising requests by priority, and could complete certain types of basic requests all on its own. You can program your bot to provide troubleshooting guides, make it fix common errors, or perform simple checks.
As well, your help bot can be configured to provide relevant answers to likely questions employees may ask and can point them in the direction of further assistance, such as support pages and websites.
PRO TIP: If you’re low on coding experience but want to build a bot using the Teams platform, you can use compatible software like AtBot that will simplify the experience. Platforms like this help you build smart workflows quickly, using an editor instead of requiring you to custom code from scratch, and connect easily to Microsoft Teams so you can deploy your bot when you’re ready to go live.
Slack – Best Chatbot Creation practices
Slack is heavily dependant on apps, bots, add ons and integrations, as these help companies and users personalise their space in a way that’s suited to their workflow and organisation.
To create and successfully deploy a bot on Slack, you first need a dedicated workspace, then you need a Slack application that’s assigned to the bot. You can access this option through the bot app dashboard, where you will be able to select the bot option and start creating a new bot driven application.
Bots on Slack are primarily geared towards companies and their employees rather than consumers and are generally created with one of three main aims:
To assess and manage activity on specific Slack channels
Bots can act as 24/7 channel monitors for channels they are invited to and can also perform administrative and other tasks, such as collecting messages that are sent and forwarding them to a specific address or user.
To act as a reminder or update service by posting messages and announcement in Slack channels
As well as collating messages, bots can post to any channel they have been invited to, sending users updates, timed briefings, as well as new information. They can also respond to individual users with relevant, informative information.
To add interactive capabilities to channel messages
When a bot is associated with a Slack app, it can enhance messages with interactive elements like buttons, to give users a way to trigger certain actions on the company’s server. This means staff can remotely operate systems, perform tasks, or access company information, providing they have the right authorisation.
The Events API is a useful tool which allows your Slack bot to access data and respond to user messages, channel changes, and other events that occur within Slack. You can configure your bot through the Events API so that when certain events occur, data will be sent to your bot that will trigger a relevant response. This response could be to provide a user information they have requested, or it could be to perform a certain task.
To give your bot access to the Events API, you need to add a bot user to your Slack workspace, and you can do this by installing the app your bot has been assigned to. When you’ve done this, you’ll received a token that contains the bot scope, and you can then configure the data that will be associated with each event your bot will respond to.
Your Slack bot needs to be able to interpret the data sent to it and assign it to an appropriate response that you pre-configure.
If you wanted to build a basic conversational bot for example, you would configure it so that when your bot is mentioned it triggers the bot to respond with a couple of follow up messages.
When you’re building a Slack bot there are almost unlimited possibilities and with all these options, it can feel overwhelming at first. An excellent place for Slack bot creators to start is Botkit Community, where you will find a plethora of resources that will help you start building a functioning Slack Bot.
Siri / Alexa / Cortana / Google Assistant Best Chatbot Creation practices
Siri, Alexa, Cortana, and Google Assistant are complex bots that offer a full feature of options to the end user. To create a bot as in-depth as this you would need a full development team with specialist skills, but that doesn’t mean you can’t offer a similar type of experience to your users.
Siri, Cortana, and Alexa offer users a rich experience, that goes beyond internet search allowing them to control elements in their home, shop on the web, book holidays, and find out specific information. They offer an element of personalisation and relatability, which is key to building successful user engagement, by answering fun non-task related questions, telling jokes, and even singing songs.
Platforms like Oswald and Rasa allow you to build and integrate your bot with services like Alexa and Google Assistant, via plug ins and add ons. These can be set up to activate or communicate with Alexa, or Google Assistant when certain commands, or words are spoken or typed. In this way you can configure your own bot to present to your user base, which also can make use of the fuller range of features Google Assistant and Alexa offer.
You can also make use of Alexa’s Skill’s developers kit, or Google’s Actions package. These are spaces where you can construct your integrated bot using the platform, your favourite build-a-bot software, and Google or Amazon’s text to speech and speech to text engine.
If you just want to create a similar style bot to Siri or Cortana and you want to do it fast with no prior experience with bot creation whatsoever, you can use an easy to build site like Rebot. Rebot helps absolute beginners instantly create a Siri like chatbot, albeit with markedly fewer features. This can be the easiest way to quickly create a simple but high quality chatbot that can be immediately embedded on your website. You could use a bot like this to promote your products or give visitors information.
If you are slightly more ambitious, have more bot creation experience, or are prepared to put in a bit more effort, software like AimyBox and OpenSource Rassa can help you build more complex mobile voice assistants for Android with a much fuller range of features.
PRO TIP: Be creative and inject some personality into all your bot’s responses. Don’t just have your bot respond with a generic answer when it doesn’t know something, have it say something funny, or set it up to give different, and unexpected answers every time.
Tips for creating Voicebots (IP Phones)
Platforms like Skype are now making use of AI powered bots your customers, leads, and clients can chat with over VOIP. Users call your number and engage in conversation with a voice bot that can be created to do anything from answering questions to taking payments.
The Microsoft Bot Framework can help you build and deploy a Voicebot into Skype, which can be configured to perform many actions. As well as answering calls, your Azure based bot could activate Group video calls, message groups, answer FAQs, and could also be embedded into your website.
BeyondVoice. Ai is another platform that offers voice bot creation with full-featured integration into many channels, including Facebook Messenger, Telegram, Slack, and Skype. The platform uses inbuilt NLU to process Natural Language and give a relatable sounding response and allows you to personalise your bot with custom logos, backgrounds, and fonts.
A voicebot allows you to be present on a variety of platforms, in a way that gives your customers a more in-depth personalised experience than a text-based bot. They have multiple uses, and are a great way to qualify leads, collect applications or queries, and provide around the clock customer service.
The inability to staff service lines at all times is a problem in our interconnected, always online world, particularly for brands that are just starting out. A voicebot can help you fill the staffing gap and maintain a presence for your customers without you having to incur the expense of employing staff on the payroll.
Think about why you need a voicebot, and what main function you want it to perform.
- Do you want an FAQ bot that can cut down on your team’s time by answering basic questions?
- Would you like to use your bot to pre-screen interested customers, by asking them a series of questions, so you can gauge whether they are worth your sales team’s investment?
- Perhaps you need a bot on hand to take automated payments when your staff are sleeping?
- Or maybe you’d like provide voice updates on your service so customers can feel connected to your brand if they have an issue and there isn’t a live agent to handle their problem?
PRO TIP: After you decide on the function of your voice focus your attention on speech as this will be one of the most important factors in defining the quality of the user experience. Make sure to use bot creation software that uses Natural Language Processing and configure your bot so it responds as naturally as possible, to maximise user retention.