Forget silly chatbots and everything you heard before that links “bad” to the word “Chatbots”. Forget about it. Because there’s now a true opportunity to scale a business, if you actually know what you’re doing. We have several valuable chatbot resources and we have A.I. now. Besides, the advancements A.I. is bringing to the A.I. field are really impressive, displaying some new and surprising possibilities, with A.I. Agents now heading towards working together as one single unit, just like humans do when we need to socialize and communicate with each other to solve a specific problem. Powerful. No wonder industry experts have said things like “B2B companies will soon have bot-to-bot conversations” not too long ago. The “future” is impressive, but it is also very scary and extremely competitive as you’re about to see next.
Now, if you’re a CEO, you’ll understand THE NECESSITY behind having a Chatbot after reading this quick summary of the marketplace that I’m giving you. So let’s do a quick reality check:
- MASSIVE LAYOFFS — A HORDE of companies are doing massive layoffs worldwide. That’s not news anymore. If you investigate this subject with care for a few days, you’ll understand where the global marketplace is really going. Here’s a hint: Chinese factory replaces 90% of human workers with robots. But this is just part of the puzzle.
- THE NEXT ECONOMIC CRISIS — Here is where people are not paying much attention. For instance, aside massive layoffs, have you heard what FIVE HIGHLY RESPECTED financial experts are warning us about, regarding a serious Market Crash that is about to happen? No? Maybe you heard then, about the one-of-a-kind move of a Hedge Fund that returned HUNDREDS of millions to clients citting imminent “calamity”? No? Nothing? Then perhaps you heard a bit about Italy’s 2.4 trillion debt which will unleash “financial chaos” upon Europe, with New York Times best-selling author Doug Casey calling it “THE MOST VIOLENT economic shock in history, dwarfing the Lehman Brothers bankruptcy in 2008 and the 1929 Wall Street crash.”. And I’m not even going deep here. Again, this is only part of what is actually happening.
- THE RISE OF A.I. — Many entrepreneurs and business owners are also not paying attention to this, but Google, Twitter, Intel and Apple are in an absurd race to buy as many A.I. startups as possible. You think that’s all for nothing? Negative. These companies are well aware about the 7 startups that are giving A.I., EMOTIONS. They also know about the 7 AI Chatbot Startups that are giving Technology a Voice. Yes, there’s a lot going on here.
- NEWBIE ENTREPRENEURS — Many people who got fired in massive layoffs, are starting their own online businesses because they need to provide for their families, or because they’re thirsty for Justice (they hate the corporate world and want a better life).
- SMART ENTREPRENEURS — Many people are leaving their jobs to start their own online businesses because they saw a true opportunity with the advancements that Tech brought to us all. As a result, many smart people, who were once employees are exploring advanced online resources to build a powerful and profitable business.
- FAKE ENTREPRENEURS — Thanks to Youtubers and entrepreneurs/celebrities like Gary Vaynerchuck, Entrepreneurship is now a trend. Something sexy. The path of the Heroes. Therefore, as a result of this social trend, many fake entrepreneurs are arising, producing garbage-like content and increasing noise on the web. They are posting pictures of themselves posing next to expensive cars that aren’t theirs on Instagram, posting pictures of famous actors with inspiring quotes and more with just one goal: to make money so they can call themselves a success. They don’t care about people at all and don’t have the spirit of business nor they want to help others. And lying to another person is a viable option to generate income for them.
- THE ULTIMATE “ATTENTION WAR” — As a result of the points above, the whole Internet (social networks, websites, blogs, forums) is becoming a place where you work for DAYS to produce some content that fights for SECONDS of people’s attention. You can produce great content, you can make smart moves, you can use Neuromarketing secrets for business, you can even merge A.I. with your business, but despite all that, companies are still living daily in a brutal war for people’s attention.
Welcome to the game.
I know, it’s a pretty hardcore game.
BUT WHAT DOES IT ALL HAVE TO DO WITH CHATBOTS?”
QUICK ANSWER: Because you’ll need to scale your business really fast now. Really really fast to keep playing in this Next Era of business. And putting money and resources into building a chatbot is a very good investment (if you know what you’re doing).
LONG ANSWER: So yes, I have good news.
There’s one way out to win this highly competitive business war: Chatbots.
But not just any chatbot. No. A true clone of YOURSELF in the form of a Chatbot for your business. I’m talking about replicating yourself at scale in a way that you can talk to other CEOs, investors, fans and customers. Someone who can address whatever you can talk about business, like this for example:
Or maybe something more personal, which journalists would like to know:
That’s what I’m talking about. Not just talking about another “customer service chatbot” or some basic stuff. I’m talking about a true Machine Learning-oriented chatbot that is basically a replica of you. A well-built, powerful, and robust clone of a CEO, Entrepreneur and Marketer. A chatbot that can give customers advice, handle sales objections, engage in normal conversations with journalists, and share branded, helpful, funny content (and much more).
But I know what you might be thinking:
“I don’t know. I don’t believe a Chatbot can connect with other people”.
All right. Understandable. So let me address this question with external facts, and then with real-life logic:
As I studied how people interacted with the tens of thousands of agents built on our platform, it became clear that humans are far more willing than most people realize to form a relationship with AI software.
I always assumed we would want to keep some distance between ourselves and AI, but I found the opposite to be true. People are willing to form relationships with artificial agents, provided they are a sophisticated build, capable of complex personalization. We humans seem to want to maintain the illusion that the AI truly cares about us.
We Need to talk about the Power of A.I. to Manipulate Humans
by MIT Technology Review
NOW HERE’S A QUICK QUESTION FOR YOU
Please be honest: Would you like to chat with a bot that is 100% a clone of Tony Robbins and powered by Tony himself, with all answers provided by Tony so the chatbot can be a serious digital copy of him?
Think about it. Pretty interesting for Marketing and PR if you know how to assemble such robot. Right? Everybody would like to know the answer for a specific question from Tony Robbins “The Legend”, and if the bot can properly answer all questions that most business people (and fans/customers) ask, then it’s the same as if it were Tony himself answering your question because, well, it is. And the question asked can be about business or about personal life.
But let me open your mind. What would you like to know from Tony Robbins THE MAN behind the brand? What would you like to know about Warren Buffet? Bill Gates? Elon Musk? What would you ask them yourself? That level of information that usually only big media companies can extract and spread, can be accessible to all via a true A.I. Chatbot. You can increase the number of your company’s business oportunities and create more one-on-one relationships with fans and customers via a Chatbot.
Or do you think these two facts below are not possible?
- AUTOMATED INTERVIEWS (with journalists) — The benefits of having a business chatbot to answer journalists? Very simple: you can get easier backlinks, mentions in media and a lot more media coverage. Because if the journalist knows you have a robust chatbot, that can give solid answers to complex questions that you would give in person about your brand/business/market/industry/personal life, then where do you think he’ll go to in order to get those answers? HARO will be too slow. ResponseSource will also be too slow. Because if your brand has a chatbot with very good accuracy, then you can automate the process of answering questions journalists like to ask business owners, and you automate the process of getting backlink and PR at the same time as well. Now, THAT is powerful.
- AUTOMATED RELATIONSHIP BUILDING (fans/customers) — The moment you give any random person some nuggets about your personal life, your personality, you personal tastes and other types of personal info, the likelihood of connecting with that person is definitely higher. Why? Because people connect when they find points in commong. We all know that. For example, if I like watching The Matrix movies and you too, we’ll probably like each other more. Instantly. Find more “connection points” and we’ll probably want to meet each other or do business together. Right? That’s because there’s a mutual preference that makes it easier for us to become friends. Why your heroes become your heroes? Because they went through some hardships that maybe is exactly what you’re going through, or maybe they have similar personalities. Therefore, the more you show about you — the person behind the brand — the better.
But don’t take my words for granted.
I’ll show you next why this mindset above for a business chatbot has ground.
One last necessary word before we talk about building a powerful CEO-like chatbot.
CHATBOT: THE POWER BEHIND GETTING PERSONAL
You probably heard of HARO, right? It basically connects professional journalists (New York Times, Forbes, The Wall Street Journal, Mashable etc) with business owners. OK, good. Then here’s a simple HARO email that I want you to pay attention to:
Now THINK: Where would journalists go if you had these questions ready to be answered in your company’s chatbot? What would happen if most journalists were aware of this too? Don’t you think they’ll take the fastest route? Definitely. Therefore, how much data and how many connections in media outlets can you get by having answers to such questions? The answer is simple: Many connections and lots of data. Sure, some journalists may use fake profiles just to get what they want from you, but many won’t. And at least you get to make your bot more popular and smarter.
But this was all just an introduction.
The real article starts now.
CREATING A LEGENDARY CLONE OF A CEO USING A CHATBOT POWERED BY MACHINE LEARNING
This is where the real work begins. Here, I must warn you that this will take A LOT of work if you’re seriously thinking in building a digital clone of yourself. OK, once that’s clear in your mind, from now on I’ll display which great marketing and business moves any company can make using a chatbot.
RULE #1: Have a goal for your chatbot and some very clear objectives.
The goal: Create a digital clone of the CEO of Peterson Teixeira Company.
The objectives: (shown below)
- CONSULTING — Provide expert business/marketing advice
- MARKETING — Promote my content, my services and my business
- CUSTOMER SERVICE — Answer any doubts about my company’s services
- DESTROY SALES OBJECTIONS — Close deals automatically whenever possible
- CREATE RELATIONSHIPS — Display my personality and answer personal questions
- LEARN ABOUT THE CUSTOMER — Gather data on how my target audience talks and thinks when it comes to consulting and marketing services
Simple, right? Well, on paper, it is. But in real life, not so much. Because in order to do all that you need a solid framework to build a machine that can actually be a copy of you, and deal with all these requests like if it were a T-800. So let’s talk about that.
Let’s see how you can create a Chatbot.
THE FRAMEWORK TO CREATE A BUSINESS CHATBOT POWERED BY MACHINE LEARNING
The company’s robot was built using API.AI, a very very very good A.I. platform for chatbot building. You can create ONE robot that can connect to Skype, Slack, Twitter, Kik, Facebook Messenger, Telegram and a lot more options. I found out afterwards that Google bought API.AI in September 2016. Makes sense, because the guys behind API.AI. aren’t playing around. The API.AI platform has been evolving pretty fast, launching payments and unlimited queries shortly after they were acquired by the tech giant. The product is pretty great for businesses, no doubts about that.
PS: They also have a very good Forum too, to help developers, entrepreneurs and marketers solve any problems regarding the platform or any other doubt that you may be having about chatbot creation.
But you can find other professional chatbot platforms here.
Anyway, the next section was created by my own hands and I haven’t read how others built their own chatbots. Since I needed a “framework” for building the most perfect business chatbot possible, I just created one. One that fits how I organize things and how I think. It’s a very solid way to structure a robot, as you’re about to see next.
OK, the way my current framework chatbot structure works, is exactly like this:
Menus for everything that the chatbot can do. If the user wants to talk about business, then there’s a menu for that, if he wants to talk about casual stuff, then there’s another one for that. If he wants to talk about my services specifically, then there’s a menu for that and so forth. There are also menus inside each topic of a subject so the user can know which questions he can ask (if he is out of ideas).
- BUSINESS SUBJECTS
Any business subject that the user wants to talk about that I personally can talk about. This can be things like SEO, Blogging, Social Media, Sales etc. But it can also be something very specific, like just Facebook/Instagram/Wanelo etc.
- BUSINESS QUESTIONS
Any random question about business that the user makes that is not necessarily inside a subject. It’s just a “random” question which is about business and that people usually ask me, but that doesn’t deserve its own section. A standalone question about business.
- PERSONAL SUBJECTS
Just like business subjects, personal subjects is also mandatory because this is where journalists will dive in when interacting with your robot. For example, they might want to address your childhood to find out several things like whether you were bullied or not, whether you had a perfect family or not, whether you had money or not etc. Have you noticed how big winners in life and business always get their life’s story told by many? Why do you think that happens? Answer: Because people care about that. It helps people connect with you in a more profound way if you went through the same tribulations. It’s easier to become someone’s mentor or role model once you reveal such things.
To reinforce this point, just observe how Kelsey Humphreys STARTED the interview with Tony Robbins:
PS: Did you notice what the interviewer starts doing at the 58:20min mark? Exactly.
- PERSONAL QUESTIONS
Any random questions about me that the user makes about my personal life. Examples: What videos I find funny, or what music or movies I like (among other topics). This is where journalists will also play ball with your chatbot, helping promote it while training it at the same time. Therefore, this section is crucial. Because if you pay attention to interviews and if you deconstruct them, you’ll realize that it’s always usually about asking A MIX of personal questions + business questions. Someone may ask you random, isolated but interesting questions like “what are your morning rituals” or “What biohacks you use?” (possible questions that entrepreneurs like Tim Ferriss likes to ask, for example) while others may ask you something more “ordinary”. Nonetheless, it’s interesting to address such things because personal questions are eye candy for journalists, podcasters, and fans. They are curious about these things a lot.
- BUYING OBJECTIONS
This is the same as Sales Objections. Any objection that the user makes that is in the way of me closing a sale. This will address objections that are common in the industry AND anything about my company as well.
Any random messages that a user may send in the conversation. Things like “awesome” or “very interesting” or some other random message OR question like “can you answer complex questions?” also fit in this category. You’ll populate this category a lot when people start talking to your chatbot.
You’ll understand this part once we dive in into the robot’s framework. But for now, here’s what you need to understand: some answers are broken into two, three or more parts and some of these answers are very interesting, like a list of professional Marketing Tools. And sometimes, people will choose to access the bot to get ONLY the latest resource of that list (the most recent one). Examples: “What is the latest funny video you watched?” or “what is the latest Marketing tool you found that you now recommend?”. You’ll need a dedicated section just to serve this kind of request.
- BANNED WORDS
Any word that you would like to ban from your robot. Think like curse words, for instance. Any word that you consider unethical for a business conversation or any word that displays lack of morals in your opinion, put it in this section and ban it.
- BANNED SUBJECTS
You’ll understand this part (is pretty simple actually) after we start talking about the framework. But basically, this section exists to block subjects that you’ll NEVER address. Example: “Let’s talk about quantum physics” or “I wanna talk about Pokemon GO”. Get it? Some subjects you will never talk about, not even in real life. Remember: The more you proactively train your robot the better.
- BANNED CRAZY QUESTIONS
Artificial Intelligence currently cannot handle common sense. Example: what happens if the user asks “is the water wet?”? That’s obviously a crazy question. Something people shouldn’t even ask. Therefore, is good to block any questions as ridiculous as this one whenever you get the chance, just to make sure that your chatbot is getting stronger and stronger. But here, you’ll need a lot of human interaction to reach a good maturity level.
All right. Now let’s cut to the chase.
DECONSTRUCTING A CHATBOT
A professional business chatbot can only be built after you have a clear understanding of how people talk through messaging apps and services. You need to deconstruct how requests are made, the humans behind it and their true intent in order to build something solid. Language, people and the message itself are key factors to consider when building a digital clone.
Therefore, here’s what you need to know to deconstruct chatbot conversations:
- Chatbot Talks: The “Categories” of Talking Humans
- Chatbot Destruction: Thinking About Message Variations Any User can Provide
- Chatbot Infrastructure: The Design Framework of A Robust Chatbot Explained
- The Legendary CEO Chatbot: The Massive Power for Businesses
OK, time to master this.
CHATBOT TALKS: THE CATEGORIES OF TALKING HUMANS
Humans are completely unpredictable most of the times and you’ll notice this more than you expect while building your own chatbot. Basically, humans are always in one of these categories below when they’re talking to your chatbot:
The Investigator, The Monkey, The Random Friend and
- THE INVESTIGATOR – This user is talking to your chatbot just to find out more about it. He may be another entrepreneur, a competitor, a CEO who is evaluating your new chatbot ideas, a Facebook employee who is testing your bot to check if you followed guidelines etc. Anyway, this user is not that interesting. He’s just deconstructing your chatbot. But you’ll need to “protect” your chatbot against these folks. More on that later.
- THE MONKEY – This user is talking to your chatbot just to find ways to break it, you know, just for fun. He doesn’t care about you nor your business and he just wants to embarrass your robot and your brand and take a screenshot of his “achievements”. He doesn’t want to help, he just wants to break things. You also need to create protections against these folks as well.
- THE RANDOM FRIEND – This user is valuable. He is the person you simply drop in a chat with your bot just so you can study how ordinary human behavior works. You’ll need this guy to understand how people type, what abbreviations they use and why, how they like to ask/request for certain things on a daily basis and other important points. He’ll help you evolve your chatbot’s conversational tone. This is why I call this user “random friend“. Because he helps you notice gaps in your chatbot that you might wanna address to make it look more like a real T-800 instead of a silly bot. You’ll need random people just talking to your bot, without them even knowing they’re talking to a bot OR what the bot is about.
- THE CUSTOMER – This is user is the one you want talking to your chatbot all day long. The more interactions and messages exchanged, the better. This user can be a prospect with doubts about your services/products, a customer who wants to drop a compliment or a complaint about your business, a reader of your blog who wants to find out more about you, a journalist who wants to interview you in automated format and similar users who are actually there to interact with your brand and who have real interest in your business. You’re building EVERYTHING in your chatbot just to provide for these folks. Engaging this type of user is any company’s end goal.
OK. Now that you know what users to expect in practice, let’s open your mind and see what “bombs” they can drop in your chatbot while talking to it. You’ll see next, how hard it is to build a true CEO-like chatbot that can handle complex business requests made by real users and customers.
CHATBOT DESTRUCTION: THINKING ABOUT MESSAGE VARIATIONS ANY USER CAN PROVIDE
Here’s some reality for you: Although you may end up talking to users who just want to “break” your chatbot once in a while, the truth is that natural human conversations with sincere and normal users are very hard to predict. Very hard. Why? Because people’s personalities influence the conversation and their responses in ways you won’t be able to foresee. Therefore, receiving different messages/answers and questions, variations of the same intent, are very very common.
But let’s talk first about DATA VARIATIONS, then we’ll talk about CONTEXT VARIATIONS.
Here’s one example, any user can give a response/message to your bot like this one below:
The user just messaged back with an image. Great.
You probably expected a text response, but your expectations were frustrated.
So this is what data variations is all about (to A.I. everything is data). The “X” possible ways any user can talk to your robot sending data, which you must be able to predict so you can build something that is ready to handle each case.
For example, have you thought about the possibility of a talk happening via emoticons?
Here’s another possibility (that youngsters use a lot):
Therefore, here are the ways a simple user can talk to your bot (data variations):
- Using text
- Using audio
- Using symbols
- Using emoticons
- Using a photo/image
- Using videos
This is something you’ll have to address.
Because a user can drop a “!” or a “-” in the conversation, and depending on the context, he’s just trolling your chatbot. But NOT always. Example? Sometimes the user wants to reinforce a previous message, like when the user drops a “?” for example. Thinking about everything is key to make all users understand they’re talking to a very robust chatbot.
So here’s a reminder for you:
But here’s some bad news: sadly…this is just THE TIP of the Iceberg.
So let’s talk about context now for a moment.
For instance, here’s a real life example that happened with my business involving a random user who was talking to my bot, and he wanted to know more about Google Analytics:
THIS is what I call context variations, like shown above:
CHATBOT: “Hey what do you want to know about Google Analytics friend?”
That’s nice. He doesn’t want to know “what is the average bounce rate” or something else. No. He wants to know “everything”. Great isn’t it? Fun stuff. But you see, in a sense, the user was right in his answer (he just answered my question) and my clone was mistaken in assuming that people will always give “the right” answer (a specific answer that clearly shows an intent).
This is what I was talking about. Predicting this kind of behavior is extremely hard even for the best bot builders out there, and you need users to keep dropping those messages in the conversation so you can make your robot stronger because you won’t be able to foresee all these answers all by yourself.
So let’s go a bit further now.
Question: what would happen if someone just dropped a math problem?
Are you getting the point?
Because now I made it very simple:
Anyway, that was just to make you think about unpredictable user input in different ways.
Now let’s crack the code: Let’s deconstruct the chatbot framework.
CHATBOT INFRASTRUCTURE: THE DESIGN FRAMEWORK OF A ROBUST CHATBOT EXPLAINED
The Framework is crucial. Very crucial. Because it defines how fast you can evolve your chatbot. A poorly designed chatbot infrastructure will definitely result in pure chaos as soon as your bot starts growing its number of intents.
Keep this rule in mind. Therefore, make sure you’re creating the right infrastructure to accommodate all training data properly, because the infrastructure is the HEART of a chatbot. Make no mistake.
Now that’s clear, here are the core areas that together form my Chatbot’s infrastructure:
- Business subjects
- Business questions
- Personal subjects
- Personal questions
- Buying objections
- Banned words
- Banned subjects
- Banned crazy questions
CHATBOT INFRASTRUCTURE: MENUS
OK. Now I’ll show you just part of PETERSON TEIXEIRA’s chatbot infrastructure so you can understand why all these areas were created, and why the framework is as it is. Pay attention to the next image:
PS: This is only part of the enormous list of intents.
Question: What these menus do, exactly? Feel free to guess.
Here’s the outcome of a few of them when the user is out of ideas and wants directions:
Are you getting the picture now? Is quite simple, isn’t?
For instance, here’s the outcome for when the user wants to have a normal conversation:
Menus have two main goals:
- show users HOW they can talk to the chatbot
- show users WHAT they can talk about with the chatbot
That way you can make users explore your chatbot with more far precision. That’s why menus are important. Because menus serve the user with instructions that your bot is ready to handle. But this is just one part of the puzzle, my friend. Before creating any menus, listen to what I have to say next:
Why? Why display an incomplete menu Peterson?
The answer is simple: Because you want users EXPLORING your chatbot, and feeding it unpredictable messages and requests so you can make your chatbot stronger. That’s why. As a consequence, you’ll be making the user imagine your chatbot as something a lot more robust, forcing him to talk in different ways which to you is great because you’ll be getting a lot of precious user data (predictable requests, unpredictable requests, complex questions etc).
Therefore, make the Menu with lots of options but also, make the Menu incomplete.
Because if the user knows your chatbot is robust, and if he knows it can address several topics, but at the same time he also knows that the menus were designed to be incomplete, he’ll likely try to explore your robot. That works beautifully. Because by taking advantage of people’s curiosity, you’ll force them to give you data on situations you haven’t thought about, and when it comes to machine learning chatbots, the more the better.
Ok, but this was on the conversational side of the chatbot.
Let’s explore the chatbot background and see what happens “behind the scenes”.
CHATBOT INFRASTRUCTURE: BUSINESS SUBJECTS AND BUSINESS QUESTIONS
By clicking on “The Menu (business talk)” on the API.AI infrastructure, here’s what this part of the robot you just saw in action looks like:
These are all possible variations for THE SAME intent. Get it?
Now let’s see the answer:
This answer is the result of one of the intents you saw, which gives the user this:
But I want you to pay attention to something.
Have you noticed that the business menu gives several subjects as options? Below you can see business questions and business subjects as I show you below in the API.AI platform:
But this is only PAGE #1.
The next page of this section of the chatbot is like this:
From the images above you can realize a few things:
1 – Business questions aren’t the focus because they are usually too random and superficial
2 – Business subjects are what I want the user to engage with so I can learn more about him
3 – Not everything that the bot can talk about is listed in the business menu
And like I said before, a SUBJECT is way different than just a standalone question. Here’s what “Blogging” can become to a user who decides to talk about it:
Are you getting the big picture? Whenever someone says “I want to talk about [ SUBJECT ]”, if the chatbot is capable of talking about that subject then he’ll dive in into that subject. That’s how the bot talks when someone just wants to talk about Blogging, or SEO, or Native Advertising etc.
So basically, whenever a user wants to talk about something you’ll either:
- #1. label it as a subject and make the user enter that subject so he can ask his question
- #2. or treat it as a standalone question and answer it right away
OK. Now you probably have noticed a “small problem” here:
“As a user, If the question is a standalone question, then I’ll receive my answer instantly. Good. But once I get into a subject with the chatbot (like SEO, Blogging, Advertising etc), the bot doesn’t give me any more instructions leaving me a bit lost of what can I do or how can I continue the conversation.”
Because when a user enters a subject (I want to talk about Blogging), he doesn’t get any other instruction. The chatbot (on purpose) won’t give him any further instructions except to “ask whatever he wants” to force the user to talk as if he were talking to a human. Therefore, things like this will happen:
This response that my chatbot gave is a SUBMENU. A menu inside a subject to assist the user so he can know what he can talk about regarding that topic, and also, to make him wonder how robust the chatbot really is.
Because this will push the user to spit new data (new questions, new requests, new synonyms etc) to the chatbot in an attempt to break it, or to just talk to it as normal as possible. Either way, this strategy allows you as a company to improve your digital clone much, much faster.
Here’s the “Blogging – Menu” so you can have a closer look:
Some of these variations above came from users. All these examples indicate that the user wants to know HOW he can address the “Blogging” subject. You must do this for every single subject that fits your brand so the user can understand that you’re taking certain topics seriously. If he knows that by typing “Let’s talk about Facebook” or “I wanna talk about Twitter” generate an instant and proper response, then he knows that you trained your clone specifically for that topic as well, with all your personal insights and advanced knowledge. Therefore, it will increase engagement and retention rates.
OK. Now you know this:
Business questions – Random, standalone questions that users make about business.
Business subjects – Whenever the user wants to ask several questions on one topic.
Menus/Submenus – A way to display what the robot can talk about (overall and one topic).
This leaves us with the following:
Message – Random, standalone messages.
Personal questions – Random, standalone questions that users make about the CEO’s life.
Personal subjects – Whenever the user wants to ask several questions on one personal topic.
Buying objections – When the user is at a sales point, but he has some objections.
Updates – When the user already knows the bot and wants to know/check a latest resource.
Banned words – Bad Words that you must block because you’re a serious business.
Banned subjects – Subjects that the robot isn’t supposed to talk about at all.
Banned crazy questions – Questions NO ONE should be asking because it makes no sense.
Since I already covered business questions/subjects, I’ll now address personal questions/subjects because there isn’t much difference. That way you can fully understand the overall chatbot talking structure faster.
CHATBOT INFRASTRUCTURE: PERSONAL SUBJECTS AND PERSONAL QUESTIONS
OK, just like I said before, certain personal questions are just “random”, right? Any personal question is just a single curiosity your customer has about you. That’s it.
Here are some simple examples:
These are a few questions I was asked in the past (I know, some are pretty weird but whatever). So if a fan or a customer asks my digital clone such questions, he’ll get the same answer as if I were there, answering these questions in person:
IMPORTANT HINT: Make your chatbot interesting AND funny. If your digital clone is not interesting or isn’t making people laugh, then you’ll lose. Why? Because there’s not much reason to talk to a chatbot if you stop to think about it. Which is why you need to find your own “secret sauce” and mix business with your personality.
Now here comes the part where this gets REALLY INTERESTING:
You can tell a lot about a person, just by knowing what makes that person laugh.
No wonder this is a topic that many customers (that become friends later) like to ask me. After all, what makes the man behind the brand, laugh? What do I like to watch as entertainment in my spare time?
Well, by far, what I like to share with those who ask is the funny stuff I watch online whenever I have some free time. So here’s another example of a personal question, that you should address so the customer can know a lot more about you, your sense of humor etc:
Show other people your human side, your true side, and people will become friends faster. And we all prefer to do business with those we are friends (when those friends are extremely professional, of course). Simple. Powerful. Effective.
But the good news?
You can hold the conversation with more ease, when you’re talking about fun subjects like funny videos. For instance, this first video is just the tip of the iceberg. If the user keeps engaging and responding yes, then he’ll start getting more answers within the same question. This can also be categorized as “episodic content” because you’re talking about one thing but you’re dividing it into consumable chunks:
If the person clicks on the links above, he’ll watch both of these videos below too:
Any CEO finds something funny, otherwise he’s just weird (let’s be honest). If you can share the things that make you laugh with other people, like customers, investors, entrepreneurs and businessmen, using your chatbot, then you have a very good way to create relationships without being present. Because people will know a lot more about you even before they know you. You probably remember that teenager in your school, years ago, who used to LOVE a celebrity without the celebrity even knowing about that teenager’s existence. The same principle happens here. You share who you really are, like in live interviews, and people will end up knowing more about you which will give you more business and friendship possibilities. Trust me on this.
Make your company INTERESTING.
And what is a company? A company is just a group of PEOPLE.
Therefore, answer the following:
1. What do you think is very funny and entertaining?
2. What do you consume for entertainment that you would share with customers/journalists?
3. What you like that would make others know a little bit more about your personality?
These are all points where people can connect with you and your business. By exposing this side, you display creativity, confidence, personality. Everything a business needs in the Era where automation and bots are taking over. For example, here’s the video behind the 3rd link which is something I found really funny (and really impressive too):
PS: Do you recognize this guy? HINT: He’s from The Walking Dead.
And if you go on, you’ll keep getting responses about my personal taste for entertainment-like videos:
Why is that important? Why showing this kind of personal info is an advantage for a business?
ANSWER: Because today, a brand must display human personality.
Don’t you remember what I said about The Power of A.I. + Marketing? With such level of automation and intelligent technology in our fingertips, for such low cost, it’s pretty obvious that many companies will start using A.I. and automation in a very wrong way.
Consequence? This will create a bigger level of chaos since email spam was born. Do you remember the 90s? Email open rates were around 95%. Now? It’s around 3%. Why is that? Simple: because bad marketers ruined everything.
The same will happen to chatbots and other A.I. blessings. This is why you need to inject your personality and create “human connection points” while using these tools, so you can avoid getting labeled as “another chatbot” or “more noise” by users.
All right. Enough of this.
Now, a brief talk on personal subjects.
This is very similar to “business subjects”. But here you’ll want to address FIRST the subjects journalists like to approach the most, like your hardships in life, your childhood etc. To make your chatbot look more robust faster, handle personal subjects that book writers/journalists can benefit from.
Deconstruct interviews of famous people and see what subjects the majority of interviewers approach. This is a very good opportunity for getting high-quality backlinks automatically. Not many businesses are exploring this.
For example, I was asked before about my childhood.
So here’s a demonstration involving only 2 questions inside the same subject (childhood), just so you can see for yourself what curious people may ask you in the future:
If you just got an urge to ask the chatbot directly this question, then you can imagine how these kinds of questions can push fans and customers to engage with your brand. This is why personal subjects are a powerful move in A.I. chatbots. Some companies may want to do this differently, and instead of creating the clone of a CEO, they might create “personal profiles” inside the chatbot so the customer can know the team that is about to serve him.
Possibilities are endless.
The number of connections you can make now, are endless too.
Therefore, explore this resource.
OK. Now this is behind us, let me address Updates.
CHATBOT INFRASTRUCTURE: UPDATES
You saw that when the user was asking for my favorite videos, the answers were delivered in chunks. But why? Well, because it makes the chatbot experience better and it makes the chatbot look more like a human.
So why another section named Updates? Simple. Because it allows a user/customer/fan to access one resource that’s inside one of your lists. Taking the previous example of funny videos, here’s why you need the Updates section in a robot:
The same thing can be applied to:
- Show me the latest professional marketing tool you found
- Show me the latest article about A.I. you read and shared
- Show me the latest motivational video you watched
This is why this chatbot section exists. Simple, right? Very simple.
Now let’s talk about Buying Objections.
CHATBOT INFRASTRUCTURE: BUYING OBJECTIONS
This is where you feel like you’re building a powerful, non-stop salesman that can crush any objection that a prospect may throw at his way. Yes, it’s good news, but calm yourself down. There are some rules you need to take into consideration because in my experience, crushing objections will only work via Chatbot if:
- RULE #1: You convey trust and honesty in your answers
- RULE #2: Your robot displays your real-life morals and principles
So let’s talk about some real life examples.
Here’s what you can tackle upfront to make your brand look more powerful and robust when facing a sales situation with a prospect who’s talking to your company’s chatbot, but has doubts about your professional services or products:
Let’s try one of these objections:
Interesting, right? Now THAT works.
Because you’re giving actual information on how you deliver your product/service to the prospect. The information he wants, the information he expects you to have, and that is necessary to close the deal. So if you can answer some specific objections upfront with a solid answer, you’ll get one step closer to the sale.
And since the prospect knows that the chatbot represents you, he’ll know that the info being provided is trustworthy (especially when it comes to common sales objections and FAQs). After all, why ruin your reputation, right?
But giving another try with a different objection, here’s the result:
We can play this game in many levels. For example, maybe the prospect talking to your bot is a manager or executive and he wants to know more about your consulting process, and how it will impact his routine:
Leave no questions unanswered.
The next objection is a good one and solopreneurs usually get the next one a lot. To be honest, the next objection is actually a pretty basic one, because real decision makers who know the game won’t even go there. Only those who don’t understand the consulting business or how online marketing works will ask this. Putting that observation aside, here’s another example of an objection your chatbot can handle:
Are you understanding the power behind a good A.I. digital clone now? And these points were just a few samples. Just a few. So if you know the sales objections your customers may say when about to close a deal, then you can anticipate their objections and turn your chatbot into a powerful salesperson.
All right. You got the sales objections part.
Now let’s shift the focus a bit to something you probably didn’t notice.
Question: What I didn’t show you so far that is extremely important for creating a remarkable chatbot experience? Can you guess? Give it a try.
Natural conversational flow.
And this is where the section Message comes in to play.
CHATBOT INFRASTRUCTURE: MESSAGE
For example, assume the talk between you and a user is exactly like this:
Did you notice his reply: “Alright, good”?
THAT is a random message, a common reply from a user to almost everything you’ll say (like slangs). Which is why is not healthy to put a special reply inside every single intent for this kind of ordinary response, or otherwise, you’ll fill your chatbot with repeated answers in all intents, like this:
PETERSON: “Robot, if the user says ‘alright, good’ after you just answered him something about SEO, make sure to respond ‘Great'”.
PETERSON: “Robot, if the user says ‘alright, good’ after you just answered him something about Blogging, make sure to respond ‘Great'”.
PETERSON: “Robot, if the user says ‘alright, good’ after you just answered him something about Neuromarketing, make sure to respond ‘Great'”.
See the problem? Not good, right? Too much work for the same results. You’ll waste tons of time for very little outcome doing this. So what’s the smartest way out to handle this? Answer: The Message section. You’ll handle standard messages that may or may not be inside an intent, separately.
So this is a piece of the whole structure:
This way any random replies like “awesome” or “wow thanks!” or common questions like “How this conversation works?” can be handled by this section, helping you provide a very high-quality experience without the need of putting any repeated effort. Because whenever people “just talk” or say “everyday words and ordinary sentences”, the Message section handles it.
Here’s a few other examples:
Another example? A user may want to talk about something like SEO, but mistypes it.
Then he apologizes for it:
THAT part was handled by the Message section.
Since it’s a common sentence, in other messages/questions/request the user may say the same thing. So why replicate the same answer to every intent? It’s just useless effort. Instead, you just grab it all with the Message section and you move on with the conversation.
Well, now the last pieces of the puzzle:
Banned Words, Banned Subjects and Banned Crazy Questions.
CHATBOT INFRASTRUCTURE: BANNED WORDS, BANNED SUBJECTS, BANNED CRAZY QUESTIONS
You probably know what these are all about, right? It’s pretty obvious. Nonetheless, here’s a quick example to make things clear:
QUESTION: Can you guess from the image above the following:
What is the banned subject?
What is the banned word?
What is the random message?
Simple, right? This is why you need to feed the chatbot with sections like those two. Because eventually, users will talk what they shouldn’t be talking. Either by choice or by accident. And to finally end this, here’s the last section you wanna have in your chatbot infrastructure so you can block questions like this one below:
OK, done. Finished.
Now you probably understand the whole professional bot structure. With this kind of infrastructure, building a robust digital clone becomes a lot easier. Any user intent can be placed in a category so it can be handled properly.
Now let’s talk about business a little bit more.
Next, I’ll show you a little bit more of the business side, and the true potential behind a digital clone.
THE LEGENDARY CEO CHATBOT – THE MASSIVE POWER FOR BUSINESSES
Most CEOs a few years ago were more on “the political side” of business. Meetings, business conversations and other political-like activities. They didn’t use to do “the technical work” much, like putting their hands in the dirt to do article/video marketing like big names like Gary Vaynerchuck or Grant Cardone do nowadays. But now the marketplace has changed. We have CEOs who know and talk about SEO, Marketing, Affiliate Marketing, Consulting and many other subjects. Many do it now because they know that for a company to get better attention nowadays, a hand from top executives is necessary. Besides, many powerful CEOs like the ones I just mentioned, do the work because it’s their passion and because it fits their life’s purpose.
Welcome to the age where CEOs are working more than never before.
So here’s what a chatbot can do for your business if you create a digital clone of the CEO:
- he can destroy sales objections and handle real negotiations
- he can give expert advice in all areas he masters
- he can can answer questions about the company
- he can can answer questions about the processes
- he can can answer questions about products and services
PS: I call my chatbot “he” to be respectful. My digital clone deserves it.
Now, I’m not gonna lie to you here.
In order to get to this level of robustness, you’ll need to invest a lot of time and resources to build something really solid. Because it’s not any chatbot that can handle everything above automatically. You’ll have to know your customer really well so you can predict any questions he may throw in the middle of the conversation, so you can then give him the right response.
But you want some piece of reality?
There’s good news and there’s bad news, so here it goes.
BAD NEWS: If your chatbot fails in 3 messages, people will judge your robot as a weak one.
GOOD NEWS: If the user who “broke” your chatbot understands “the behind of scenes”, and the amount of work involved to build a chatbot, he’ll give it another chance in the future because he knows that a broken chatbot will be a completely different one in the same week (assuming of course, the team behind the chatbot is taking its training seriously).
Anyway, let’s talk business.
Next, I show you an example of someone who is a bit interested in what I sell:
OK, that’s the basics. But maybe, after a while talking, I decide to make the chatbot trigger a few questions to the customer and his project so I can gather more specific data that may help both of us in our first meeting, and he then, finally answers this:
With those questions answered I now know what to expect. This is what you need to start doing to make things easier for both sides. It’s even easier for your company to create a better proposal for customers as well. Creating a very good clone has only benefits for your brand, because customers will be talking to it before doing or buying anything, which is a Win-win situation for both. You filter customers, and customers filter you. And using 1/3 of the time.
But let’s see another example.
What if the customer wants to go a little deeper and check if you’re a “perfect fit” for his company? Well, you probably know that he may ask you something very simple like this:
If your company only likes to work with certain niches, then is better if your customer knows this in advance, before both of you engage in any business meeting or conversation, right? So that’s a good filter to have. Why waste anyone’s time? Time is money my friend. Therefore, a very professional business dismisses unnecessary meetings or customers it can’t or don’t want to help. Just be real. Period.
The customer may also want to ask you, in what your company specializes (another good filter):
Or maybe the customer is testing your knowledge and posture, or maybe he doesn’t know the basics behind Marketing. But either way, he may ask you something like this too:
Another good point is talking about your readings. Books. Articles. The content you consume to upgrade yourself. Because high-level customers, like entrepreneurs and C-level executives may want to know what you’re reading to know what kind of content the CEO is feeding into his own mind.
Because based on someone’s readings, you can tell a lot about that person. For instance, by knowing the 14 books that inspired Elon Musk you can understand (and expect) why Musk started companies like Tesla and SpaceX. But you also know that he likes to read and share very well-researched, and advanced articles that deconstruct A.I. With information as simple as these, you have a better profile of who you’re dealing with. And for B2B customers, this is very important.
So let’s take another example:
Now, from the output of my digital clone above, you probably know now the following:
- BUSINESS MINDSET (HINT #1) – (LINK #1 + LINK #2)
I enjoy reading very advanced content on A.I.
- BUSINESS MINDSET (HINT #2) – (LINK #3 + LINK #4 + LINK #6)
I enjoy reading the consequences of A.I. in the marketplace
- BUSINESS MINDSET (HINT #3) – (LINK #5 + LINK #8 + LINK #10 + LINK #12)
I enjoy reading/writing the consequences of A.I. in Marketing
- BUSINESS MINDSET (HINT #4) – (LINK #7 + LINK #9)
I enjoy reading what great experts and entrepreneurs are talking about A.I.
- BUSINESS MINDSET (HINT #5) – (LINK #13)
I enjoy reading what problems and solutions A.I. is generating
- BUSINESS MINDSET (HINT #6) – (LINK #11)
I enjoy reading what the media is talking about A.I.
- BUSINESS MINDSET (HINT #7) – (LINK #14 + LINK #15 + LINK #16)
I enjoy reading what A.I. companies most investors are following
Do you think knowing this kind of information helps a B2B customer? Definitely. Not only what you are reading defines you, but it also reveals a lot about you and the business you’re running and how you’re running it. The same question could have been made to find out what are the recommended readings about Marketing, SEO etc. Customers will want to know if the content you’re consuming is advanced enough so they can predict whether or not you can serve them well.
The content you read shows customers HOW SMART YOU ARE, WHAT YOU KNOW ABOUT YOUR MARKET, YOUR BUSINESS and YOUR INDUSTRY. And all this helps the customer and thus, also helps you close THE SALE.
Now let’s assume a different situation.
In this next example, the customer wants a preview of your expertise for his own niche.
A tiny free consultation. So here’s what it would look like via chatbot:
Let’s take another example:
But what if the customer wants to know more about a specific subject from you that will be required in his project AFTER your company gets hired? He may want to test your knowledge upfront. So here’s another example:
That’s it. All the variations above are real possibilities for businesses with chatbots. All you have to do, is design it carefully so it can Sell, Educate, and Create relationships. There’s a lost of work involved, but know you have a glance of the power behind a digital clone.
A SHORT CONCLUSION
A business chatbot is not something you can build overnight. Please have this in mind. You’ll spend hours and hours injecting episodic content in your bot, hours and hours thinking about all possible reactions and queries your users may make, and hours and hours trying to predict human behavior to deliver a very good experience.
The path for a robust chatbot?
One word: Training.
So here’s another final business advice: you’ll have to reach out to normal people and people from your target audience, and make sure to involve them in a chatbot interaction just so you can power up your digital clone with N variations for the same intent. There’s no other way. You’ll need help. Massive help. And lots of data from actual users.
Just remember that the framework, the chatbot structure, is VERY IMPORTANT:
- business subjects
- business questions
- personal subjects
- personal questions
- buying objections
- banned words
- banned subjects
- banned crazy questions
ONE LAST THING: There will be times where you’ll feel lost.
When you reach that level, and you don’t know where to look to improve your chatbot, just go to places like Quora, Forums, and Online Communities and look at the questions being made by users so you can see what are their main requests (and also HOW these requests are being written). Comments in websites are another good source of information for this too. Next, just feed those same questions to your chatbot. Is that simple.
That’s it, my friend.
Now you have the tools to create a digital clone that can handle customers at scale.
Just embrace the work. Because trust me, it’s all worth it.