The Future of Chatbots – Key Chatbots Statistics and Trends in 2021
In a previous article, Adaptive Learning is the Future of Online Learning, we talked about how most education and learning in 2020 was largely done from home through virtual platforms due to COVID-19.
As we step into the new year with the pandemic still at large, it is expected that there will be another surge in the use of artificial intelligence (AI) and chatbots due to their time- and cost-efficiency. If you’re also planning to use more AI and chatbots, here are the key chatbot statistics and trends for 2021 that you should know.
With 44% of consumers preferring to interact with a chatbot over a human, one thing’s for sure: chatbots are here to stay. But how do you build a chatbot for your business — one that actually helps you solve problems instead of creating more?
In part 3 of my Conversation Design series, I’ll talk you through the process of building a chatbot from scratch, including strategy, scripting, and testing.
Let’s go!
1 — Define the goal
Before you can start building your chatbot, you need to know whyyou are building it. What is the goal here? If you want to automate an existing service, what is the current experience like, and how could a bot help improve it? Take a look at your business goals: if one of your goals is to increase customer satisfaction, you might want to add a chatbot to your customer support team and let it handle the most common FAQs, so your team can focus on the more complicated cases.
2 — Define the use cases
After figuring out the why of your bot, it’s time for the what. What is your bot going to do exactly? What will it do for the user? I cannot stress how important it is to figure this out before you start building your bot, as you need to know exactly what the bot will do and why that is important.
Here are some use case examples:
Book a restaurant
Play a song
Find products
Recommend new products
Get directions
Book a flight
Show local promotions
Process returns
3 — Understand your tech
Know that you know the why and the what of your bot, it’s important to understand the where: where will your bot live? Will it be integrated with WhatsApp? Can customers engage with it via SMS, Facebook Messenger or on the company website? What are the restrictions of each channel? A bot that talks to your users via SMS won’t be able to use as many characters as a bot that only communicates via web. So make sure you understand your tech and its limitations.
In order to design an experience that feels personal, you need to make it personal. Who is this user that will engage with your bot? It’s important to know what they want and how they are feeling during the conversation. What’s their backstory? Their challenges? Their motivations? How familiar are they with your business and using bots in general? I like to create a user ID that I keep close when writing the actual dialogues.
An example user ID for a company that rents out luxury hotel rooms
5 — Craft your bot personality
A chatbot without personality is like a bad Tinder date: they looked great online, but as soon as you start talking to them, you want to end the date as soon as possible. There just wasn’t a connection…
So how can you make sure your users connect with your chatbot and that the conversation is engaging and representative of real human interaction? By giving your chatbot a clearpersonality.
If you can, use your company branding as a starting point and build on it. I’ve also explained in detail how to design your chatbot’s personality in part two of my Conversation Design series.
6 — Script your happy flow
Now that you have a clear picture of who’s communicating (your bot persona and your user ID) and what they’re talking about (your use cases), it’s time to write the dialogues.
A ‘happy flow’ isa dialogue where everything everything runs the way it’s supposed to run. The conversation is natural and smooth, and the user reaches their goal in as little steps as possible. Many conversation designers start with the happy flow because it’s the flow of least resistance. It takes the least amount of effort to script because it doesn’t include many of the inconvenient complexities that can occur.
But they will, and you need to be ready for when they do. More on those edge cases later.
A great way to script natural dialogues is by having a sample dialog. Have two people sit back-to-back and improvise a conversation around a use case, with one person playing the user and the other playing the chatbot. Record their conversation or take notes to see which parts of the dialog need a bit more work.
7 — Script for edge cases
Chatbot technology is not yet capable of understanding every user utterance well enough in order to reply in a correct way — no matter how well the script is written, in tricky situations, it will most likely fail.
So after writing the happy flows, write out the most likely ways a user might go off track and how you’re going to deal with that. The sample dialog should help those pain points, as will user testing.
What if a user asks your bot how it’s doing? What if they tell the bot they don’t like it? What if they want another suggestion? What happens if the user wants to book a table for two, but one person is allergic to gluten and the other one doesn’t eat fish?
If you can, make sure your chatbot’s response strategically guides the user back to an existing flow, like in the example below:
A good answer to an out-of-scope question that puts the user back on track
8 — Create the flowchart
Once you’ve written all possible flows, create a visual flowchart to see how a user would go from start to finish and where they might want to — or have to — dive into other flows. There are a ton of great tools out there to help you visualise these journeys, my two favourites are draw.io and Creately.
Quick tip: The chatlayer.ai platform allows you to visualise your flows straight away whilst scripting your dialogues.👏
A visual overview of the different dialogues using the chatlayer.ai platform
9 — Testing your bot
When you’ve completed your dialogues and created your flowchart, it’s time to take a deep breath. You’re now gonna send your little baby bot to its first test! 🐣
To get some first feedback, you can share your bot with a few friends and colleagues and ask them to complete some specific tasks. For example: “Play a song” or “Order some food” or “Return a pair of jeans”. Make sure to also ask them some detailed questions about the overal experience after:
Personality: does the bot feel consistent? Does it fit the brand?
Onboarding: was it clear what this bot can do for you? Did you miss anything?
Understanding: did the bot understand your questions and answers? Where did you get stuck?
Answering: did the bot give accurate, relevant and clear answers? Did it feel conversational?
Error management: how did the bot handle errors? What went wrong? How did it make you feel?
Overal experience: did you enjoy the conversation?
Some platforms also allow you to do some usability testing, such as chatlayer.ai. There you can see which intents and expressions were recognised (in)correctly, how long each user session was, and when the API plugins and code blocks returned an error.
An overview of the NLP testing area in Chatlayer.ai
10 — Optimising
Internal testing will already give you a lot of insight on how to improve your bot, but its your real users that you want to hear from. So after publishing your bot, make sure to keep monitoring its performance. Monitor the conversations, collect data, create logs, analyse the data, and keep improving the bot for an even better experience.
Phew, you made it till the end! Now it’s time for you to build your own chatbot. 🤖 If you need help getting started on chatlayer.ai, I recommend you watch our platform tutorial series on Youtube:
Hi, I’m Tess, language enthusiast and taco lover. I work as a Conversation Designer at Chatlayer.ai — an intuitive platform where you can build clever AI bots in any language, no coding skills required.
Interested in knowing more about chatbots? Want to share stories or talk conversation design? Let’s chat on Twitter or send me an email.
There was a time when you need to connect offline with businesses for a B2B partnership. However, now you just need to search for them on LinkedIn and click on the connect button.
Every marketer and business knows how much LinkedIn is important for their growth. Alike other social media platforms, LinkedIn is not a general content-sharing social media platform. It is more of a professional platform and was built to bring professionals together for business, career, and company profile building.
The revolutionization of technology and marketing practices has taught marketers how they can get even more from LinkedIn. It is now being used to generate leads too for businesses. Businesses search for prospects on LinkedIn, send them a connection request, and texts them to offer and discuss partnership or work opportunities. As a result of this, LinkedIn marketing practice has been really beneficial for businesses. Marketers claim that they generate 54% of B2B leads from social media, and out of which 80% of them were from LinkedIn.
The question now arises that how you use automation to make this marketing practice easier, more beneficial, and cheaper.
The concept of automation is all about reducing or eliminating human interference from monotonous tasks and allow them to do more important and productive tasks.
Automating LinkedIn marketing can help businesses in making the process more resultant, error-free, and cost-effective as automation minimizes the need to invest many resources. Businesses can use LinkedIn automation tools for the same.
With a LinkedIn automation tool, businesses can do Linkedin prospecting easily. They can use it to search for prospects in LinkedIn and send outreach messages whatever they want. Such automation tools can help in collecting & keeping prospect data saved and managed.
Overall Business Costing Benefits
Check out this example to understand how Linkedin automation affects your Linkedin marketing cost.
For instance (Without Automation)
You are investing two members of your team, the first one to search for prospects and send them outreach messages, and the second one to respond to replies and attend sales meetings. You may have been attending 10–20 sales meetings and also, be doing a lot of hard work on it.
For instance (With Automation)
All that you will now need is to get a Linkedin automation tool and train it to do the work however you want. The major workload for you will be to attend sales meetings (that too, definitely more than you were attending without automation) and monitor stats.
The overall purposes of these examples were to explain how LinkedIn automation can reduce the work stress of your employees and get things done more efficiently. Ultimately, help you generate better ROI.
Linkedin Conversational Ads
Linkedin ads have always been businesses’ first choice for B2B paid ad campaigns. However, to make the conversions from LinkedIn ad campaigns reach their best potential, you can switch to LinkedIn conversational ads.
The Linkedin conversation ads are built on the message ad format, which is delivered through LinkedIn Messenger. They are designed for real-time engagement, as they only send messages when a prospect is active on LinkedIn, which means the possibilities of engagement gets quite increased.
Conversational ads help businesses to make the lead conversion process easier and efficient. With this advertisement practice, businesses would not need to invest much of their other resources like managers’ involvement in responding to the advertisement replies. In simpler words, you can say that conversational ads work like a well-trained chatbot.
Moreover, Linkedin itself offers conversational advertising options. You just need to set your target audience, goals, and design an appropriate conversational flow with good CTA buttons.
(Pro tip: Better your conversational flow, higher the conversion rate)
Get More With Social Media Marketing (SMM) Chatbots
The other way to use conversational ads in your Linkedin marketing is by adding CTA redirection to your conversational landing page in sponsored posts.
When a user clicks on the CTA button of your sponsored post, he/she will land on a conversational landing page, where a chatbot will welcome the user and begin with the conversational flow.
Chatbots can be trained to communicate with leads differently based on the source, location, and many more such factors of the leads redirecting to them. This ensures to give the best engagement experience to the leads, and thus the maximum possibility of conversion.
The time is changing and if businesses don’t upgrade themselves to better technologies for marketing then they may suffer a great loss in the upcoming time. Using Automation & Conversational marketing are two of those things. If you don’t believe it, ask those who are already using them!
Does your chatbot live in Facebook Messenger? If so, it might stop working properly because of Facebook’s most recent API updates. But don’t despair, in this article I’ll explain how to deal with the changes and create a buttonless bot that works, no matter what Facebook does!
To button, or not to button? That is the question.
But not when it comes to Facebook.
On November 30th, Facebook announced they are updating their Messenger API to comply with European privacy laws, which will impact many of the 300.000+ chatbots that currently live on this platform.
If your Facebook page and bot users fit the criteria listed here (such as being located in Europe), users will no longer be able to see bot messages that contain:
Buttons
Carousels (also called generic templates)
Quick replies — update: these might actually stay
Any media other than images — such as video, files, audio
So that’s the bad news. The good news is that you can easily avoid all of the messaging types above by cleverly using intents, context and other conversational design tricks. And if you think about it, we have all these new, smart devices that we can use in almost any way we choose. But out of laziness or lack of innovation, we continue to force users to click a tiny little area just a few pixels wide..
Let’s do something about it. Let’s go buttonless!
1. How to avoid buttons and carousels
You can easily replace buttons by using a combination of match entities and input validation. For example, when the user places an order, the bot can check in a predefined list of available menu items if this order is indeed possible. Sure, you’ll need to create a few extra steps, but it almost makes it more conversational this way.
Using a button to show a URL? Simply turn it into a message containing that same URL. Sure, it’s a few extra steps and it doesn’t look as good, but it makes no difference in the end. In fact, it’s so much more conversational!
Carousels are a way to visualise options by using images and buttons. Each option can have up to three buttons, which are the same as any other buttons.
2. How to avoid quick replies
Instead of showing quick replies, you can add the options to your text message. Create an entity (for example, @shirt.size) and add all possible values (small, medium, large), so your bot is able to detect the correct value.
3. How to send files
Instead of sending the PDF to the user for download, you can host it on any file hosting service like Dropbox, Google Drive, WeTransfer etc. This way, the user can view the PDF in their browser.
ps: Chatlayer.ai recommends a max. size of 5 MB for media files shared on Facebook Messenger, as Facebook seems to struggle with files larger than that.
5. How to use videos and audio
I know, the video thumbnail on the left looks better visually, but when it comes to the actual conversation, nothing really changes. The user still has to click on the bot’s message, it just looks a bit different. And instead of the video being played inside the conversations, users will be redirected to an external website.
Do make sure to create an intent for ‘see product video’ so the bot knows what to do when the user states this type of intent.
To wrap it up
Sure, buttons are handy and fun. I love a good button myself! But if your bot lives on Facebook Messenger, you should skip them all together. Having a buttonless bot protects it from breaking every time Facebook decides to update their API.
Besides, a chatbot’s main purpose is and always will be to imitate a real-life conversation with people. And there’s no buttons in real life anyways. 😉
Hi, I’m Tess, language enthusiast and taco lover. I work as a Conversation Designer at Chatlayer.ai — an intuitive platform where you can build clever AI bots in any language, no coding skills required.
Interested in knowing more about chatbots? Want to share stories or talk conversation design? Let’s chat on Twitter or send me an email.
Try it yourself!
Sign up for our 30-day free trial on chatlayer.ai and follow the easy step-by-step tutorial to build your own chatbot from scratch — no coding skills required.
Don’t forget to give us your 👏 !
How to create a buttonless bot was originally published in Chatbots Life on Medium, where people are continuing the conversation by highlighting and responding to this story.
While RPA robots automates a series of steps which are somewhat predictable in terms of application and data, chatbots automates communication with live chat acting as human-takeover when the chatbot gets questions not trained for.
RPA & Chatbots are natural partners
Symprio is a UiPath Gold partner and delivers automation solutions for customers and also provide chatbot solutions leveraging both our own (BotExpress.ai) as well as other platforms. We see more and more where automation flows are being combined where RPA robots executes processes on top of legacy applications and chatbots and live are being used to automate or simplify communication.
Modern chatbot platforms like our own (BotExpress.ai) detects intents (essence of what you want) as well as entities (data variables) which provide the specifics for a specific request.
More traditional live chat solutions only handle the chat process (and in essence fully unstructured data) but with Artificial Intelligence and richer features in live chat (or chatbot) applications we are able to extract the necessary entities (data variables) and turn this unstructured data in to structured information which are required as inputs RPA robots to executed a process.
Chat sessions can extract in basically two ways:
(1) By leveraging NLP entity extraction process chatbots can find and identify information from unstructured messages to convert for example the phrase “tomorrow” to a specific date.
(2) As part of chat sessions instead of just using text we use something we call SmartForms which increases the reliability of these inputs.
This includes text fields, date selectors, dropdown lists etc. to be able to capture and validate input fields within a chat sessions.
By being able to either select or validate input fields in the chat sessions the entities (or data variables) are more reliable parameters which can be used to trigger RPA automation flows.
Reliability of these data inputs is key for reliable automation flows.
Scenario 1: Triggering RPA Robots to retrieve information or perform transactions from chat
A chat session captures the necessary information (either through live chat or a chatbot) and triggers and an RPA robot to either go to web applications or legacy applications to retrieve information or execute transactions.
A key advantage of RPA robots such as UiPath is that they are not dependent (though also possible) on integrating using API’s and can work on both older type applications (even AS/400 green screens) to modern web applications and even visual changes in the screen. With UiPath’s powerful image recognition engine even Citrix-based environments, which are purely image-based, processes can be executed.
This can be used both with internal and external customers. Internal chatbots can offer employees information or requests (i.e. leave request) about HR, IT, Finance, Supply Chain or Sales & Marketing team which in turn can launch UiPath robots via Orchestrator and once complete return with results/information back to the chat sessions.
Another example are Level 1 Helpdesk chatbot which can capture the necessary information, guide users through troubleshooting for common issues which also can be supported by robots to perform troubleshooting and resolution (below is a scenario on robots doing the IT support)
External bots provide support for customers or suppliers in sales, support, procurement
Scenario 2: For Live Chat agents
The typical Live Agent uses pure chat for the purposes of communication but based on the information captured and (copied/pasted) in to other applications to check on status, retrieve information or to make changes. When you look at the time an agent spends to deal with a customer upwards of 45% of the time is spent on transferring information from chat sessions and in to other applications (this is the time when you sit and wait for a response from the agent) and back.
Leveraging more intelligent data capture mechanisms such as AI entity extraction or embedded forms in the chat sessions which is stored as variables in the session which in turn can be sent to the UiPath Orchestrator to execute processes via robots, saving time from the agent having to transfer data (copy/paste) as well as actually executing the process or retrieval from third-party systems.
Once the information is retrieved this in turn is sent back to the chat sessions and automatically be embedded in response messages from the agent. This effectively lets the agent have robots at their finger tips to execute processes at will.
Even in cases where you are using a simpler live chat platform unable to extract and orchestrate robots, the data can still be captured in to a spreadsheet and agent can launch the robot manually to execute transactions or retrieve information much faster than manually logging in to the applications.
This significantly increases the capacity as well as speed of resolution for agents, while at the same time ready the process for complete automation where a chatbot executes the same process an agent does.
Scenario 3: Leverage chatbots (primarily internally) as a Human-in-loop process
When a robot executes a process it may reach a point where either additional information, a decision or evaluation is necessary. Similar to how UiPath Action Center (as part of Orchestrator) can be used to create tasks which are assigned to people within the organization to act on.
Once the assignee makes a decision or provides the missing information to Action Center the robot takes over and continues the process to the next step.
Similar to Action Center, internal chatbots can also be the medium to engage internal business users for quick decisions/approvals or missing information for bots set up on internal messaging platforms such as Slack, Workplace by Facebook or Microsoft Teams.
Given that all these platforms are mobile enabled the speed of decisions and resolution can often be faster than when relying on emails.
Symprio is a UiPath Gold partner with a strong delivery team located in USA, Singapore, Malaysia & India. Symprio, leveraging our own platform called BotExpress.ai, provide internal and external facing chatbots integrated with UiPath robots to automate end-to-end processes.
