An Introduction to Chatbots
Follow along on NapkinLab.com
I’m a couple weeks behind on the challenge due to some needed attention in my marketing business, but now i’m back with a quick post on a topic that I see as being a game changer for many businesses in the near future:
Chatbots! I’ve been hearing the buzz for a few months now — especially from Andrew Warner on Mixergy — and finally decided to look into them a little more.
I’d previously thought of them as just another not-very-intelligent chat tool that would clutter my phone screen with notifications. I barely read the messages that I get from people I actually know, why would that be any different with an artificial entity? But when I looked a bit deeper I found something pretty incredible. It took all of about five minutes to really start to see that this is going to be a massive opportunity for anyone who is willing to take a little risk and learn the rules of the game now, before it really blows up.
What Are Chatbots?
To really explain the opportunity, let’s take a step back and start with just what exactly they are before going into what all they can do. The very simple explanation is that:
A chatbot is software that lets you communicate with your customers via messaging apps.
This is a very simplistic version, but it pretty much nails the core of what it is and why it is important. Don’t think communicating with your customers is important? Let’s take a look at your bottom line and see how you are really doing in business.
The longer version is that a chatbot is an artificial intelligence that lives within a messaging system of some kind for the purpose of assisting your customers or communicating between customers and your business. This could be anything from a very simple auto-reply when certain words are used up to a reasoning and semi-understanding to diagnose problems or offer specialized assistance. A chatbot could even deliver content for a course or offer regular updates.
Right now some notable examples of bots currently in use include a healthcare startup using bots to converse between doctors and patients, the Rio Olympics using chatbots to keep subscribers up to date on their team’s standings in real time, a workout tracking chatbot, food tracking for weight loss, and a Q&A platform that answers health questions. There is even a chatbot that delivers games in-chat to over 4 MILLION monthly active users!
How Do Chatbots Work?
Right now there are two main types of chatbots: Rule-based chatbots, and Machine Learning/AI chatbots.
Rule-based chatbots are kind of like playing a choose your own adventure game: The chatbot will ask you a question and you will answer. Based on the decision tree that it’s creator wrote, it will then either perform an action or ask you another question. A slightly more advanced version of this can include rules that are executed based on keywords or phrases.
For example, if the user says the word ‘discount’, then the chatbot will send a message with a discount code. This can be extrapolated to just about anything that you could imagine from creating a simple dictionary bot to delivering dripped content for a course. Just because it is ‘on rails’, so to speak, doesn’t mean that it is severely limited in it’s functionality. It does mean, though, that it is only as smart as the author makes it.
The second kind of chatbot is more advanced. They are built using machine learning algorithms and artificial intelligence and can both learn from it’s interactions as well as make decisions about a conversation on the fly that the author may not have anticipated. This kind of bot allows a lot more flexibility since it can understand language and the intent of a question without relying on specific keywords to trigger a response. This kind of bot will get smarter and learn how to better interact with it’s conversation partners the more interactions it has.
Of course, the dark side of this is that the bot may not always respond how it was intended. A now-famous example of a bot gone wrong was the Microsoft bot, Tay. She was an experimental bot that lived on Twitter, and boy did Twitter users have a field day with her… In just minutes, Twitter users turned her into a racist- prompting a fast shutdown by Microsoft before she could be corrupted any further. But for every example of the internet using chatbots for other-than-intended uses, there are thousands of examples of incredible AI and machine learning-based bots that are solving new problems and developing new solutions to the problems of modern life.
How Long Have Chatbots Been Around?
This is where things start to really get interesting. In some form or another, we have been playing with AI and Machine Learning (the backbone of chatbots) since Alan Turing started his milestone work 70 years ago!
Oddly, since that time we have progressed very little in the way of artificial intelligence research. Perhaps it is because we are afraid of what it might become, perhaps it just fell out of style for a while. But either way, it’s back and we generally have an idea of what is capable when we invest in developing it further.
One of the big milestones in recent years that has helped spur this recent boom in chatbots is the widespread availability of platforms that have opened their APIs for the average developer to start building these kinds of tools. Recently, Google also open sourced it’s AI-engine, TensorFlow. While not really applicable to the average Joe or Jane, this kind of sharing and availability of tools of this caliber open the doors for people much smarter than I to start building tools that anyone can implement.
In this case, that means that we have very user-friendly and simple to use tools for building chatbots without coding that can be deployed to platforms like Facebook Messenger, Slack, or Telegram.
Why is it Important That We Can Use Messaging Platforms?
Why not deploy them on our own website and own the data rather than getting even more in bed with the big platforms?
Because that’s where the people are. For the first time in history, messaging app usage has now surpassed the entire total usage of the top social networks. I don’t know about you, but if i’m are going fishing it would be wise of me to go where the fish are.
Now obviously there are plenty of great reasons to keep the chatbot on your own website- maybe you are in a space with privacy issues or you will be addressing sensitive information. Maybe you already have a massive brand built up with millions of web visitors every day. You will have to analyze your own business use case and determine what is right for your business, but for most businesses going to the platforms where people already know and are comfortable with will be your best choice of all.
“People are now spending more time in messaging apps than in social media and that is a huge turning point. Messaging apps are the platforms of the future and bots will be how their users access all sorts of services.” — Peter Rojas, Entrepreneur in Residence at Betaworks
The Killer App That Won Me Over
That’s great and all, but really who cares? …is what I was thinking to some extent up until this point. We’ve still just got some cool toys that talk to you when you message them. Where’s the magic that shows why this could be a big deal?
Conversational Commerce. If you sell anything at all, either for yourself or for someone else, how often do you need to talk to someone? How often, when you do talk to someone, does the experience for the customer improve and end up with a sale?
In recent years there has been an obsession by the internet in taking humans out of the sales equation anywhere possible in an attempt to achieve the elusive hands-free sales pipeline that requires no work. Well as it turns out, in doing so we have also lowered the customer experience in many cases and even lost sales due to removing the ability to talk to a knowledgable sales person.
What if you could put that sales person back in the sales and marketing mix — but not ever need to add an actual person? That’s where chatbots can help. They can be programmed to provide customer assistance in answering questions, looking up information such as shipping information, or even complete sales right inside the messaging app!
Don’t believe me? Try messaging Nordstrom on Facebook right now.
Go ahead i’ll wait.
You will be talking to a chatbot who can help you shop right from the messenger window and go from initial discussion all the way through to a completed sale.
“Messaging is where we spend a ton of our time and expect to communicate. It is ridiculous we still have to call most businesses.” — Josh Elman, Partner at Greylock
Here Is Where The Magic Really Happens
What if you could also include that chatbot in your marketing funnel before someone even knew who you were? Or if they could drive a potential customer further into your funnel?
By using the Facebook Messenger platform to house your chatbot, you can also integrate this chatbot with your Facebook Ads.
That’s right, you can deploy a Facebook ad campaign which when clicked, will send a viewer to a messenger chatbot that can start interacting with your potential customer and do anything from answer questions to registering them for and delivering a webinar. Ezra Firestone uses this technique for his retargeting ads in his company Boom by Cindy Joseph.
Another example of using chatbots in a Facebook ad strategy is by Rick Mulready of The Art of Paid Traffic Podcast, who uses a messenger destination to retarget people who registered for his webinar but did not buy his course and asks them if they have any questions he can answer (See image above).
Now in his case he only used the bot to initiate the conversation and chose to handle all incoming messages in person, but he could have just as easily built a Q&A-style chatbot to handle that from an automated angle and then taken over manual control if needed.
Now just like on email you can build drip campaigns, send blasts (broadcasts in chatbot speak) to all contacts who have messaged you, and even collect data on the people who are interacting with it. Basically, it’s like having all the powers of a full email automation campaign with EVEN MORE data, on a platform that regularly gets better engagement than email (20% open rates for email compared to 98% open rates on messaging according to HBR), and has an audience of over one BILLION users on Facebook Messenger ALONE. And unlike email automation, this platform is completely free. Facebook has said that it will begin charging for this in the future, but for now it’s like having the keys to the candy shop as a five year old.
As an experiment, Mixergy.com tried switching to using Facebook Messenger and achieved an average 80% open rates and 50% click rates!
So why wouldn’t you want to get in early while the trend is still building?
How to Build a Chatbot Without Coding
Generally, the next question that people are likely to ask is something around the fact that they can’t code, or don’t know AI, so how can they build a chatbot without hiring a developer to build the thing. Fortunately for us non-devs, there are several great tools out there that are particularly user friendly and cheap, or even free!
As it turns out, the issue isn’t actually with building the bots themselves. It’s designing a user experience that will stay out of the way and feel natural. For now, most businesses don’t actually need heavy AI to run their bots. Most can get by with a rules-based bot that will deliver nearly all the functionality that they require. But to do that, they first need to determine what their end goal is for the user. Are we trying to drive sales? Are we delivering content? And where will the user enter this conversation along their customer journey?
Asking these questions up front will save you lots of headaches and time later on, as well as speed up the over all time to deployment.
Three specific tools that I want to mention are: ManyChat.com, ChatFuel.com, and Motion.ai. Each offers a simple visual interface for designing the conversation of your bots and each offers a free plan that lets you try out the system before really committing and investing money in a paid solution once you start getting lots of users. Each has also had some battle testing at this point from entrepreneurs around the world and proven to be able to hold up to the kinds of traffic needed to use a tool like this at scale.
Hopefully this post has illuminated some of the opportunity that is out there with chatbots and maybe provided some ideas or tools that you can use in your own business.
Follow along on NapkinLab.com