Missed an episode of Chatfuel School—our weekly YouTube show where we cover hot topics in the world of bot-building and marketing? Don't worry; we took notes for you.🤓 The topic for May 13, 2020 was how you can use an AI chatbot to grow your business. We brought in Chatfuel customer, AI expert, and co-founder of Janis.AI Josh Barkin to share his insights. Here's a recap of what you missed.👇 (Sign up to get a reminder when the next episode premieres!)
Why isn't your business using artificial intelligence yet? 🤖 Here's the most common reason: This technology seems too intimidating and out of reach for SMBs. It sounds like the stuff of Hollywood movies, something that requires a whole team of experts and a multi-million dollar budget. That's exactly why we brought Josh Barkin on the show this week—to bust these common misconceptions.
In reality, businesses of any size and budget can use AI technology. It's an incredibly powerful tool for improving customer service, increasing sales, and standing out from the competition. Contrary to popular belief, anyone can build an AI chatbot for Facebook Messenger to do all of the above, and Josh joined us to explain how.
What AI is (and is not)
Artificial intelligence is the technology that allows computer programs to become smarter over time with machine learning. It's the technology that lets chatbots, voice-enabled speakers, smartphone operating systems, and more to understand human input and respond accordingly. It helps them pinpoint the intent behind a user's message, instead of returning only strict, rule-based responses.
Josh gives the example of a non-AI chatbot for a pizza restaurant. 🍕 Let's say one user tells the bot, I'd like to order a pizza. Another user asks it, If I order a pizza, how long will it take? Both inputs include the key phrase order a pizza, but the intents are completely different. The first user wants to place an order, and the second has a question about delivery times. AI would allow the bot to learn the difference in cases like these, so it can give both users the right information.
However, because AI has become such a common term, Josh pointed out that it's important to manage user/client expectations. Building an AI-powered chatbot doesn't mean you'll have the equivalent of Siri for your business overnight. "People often think, "Oh, it's like Siri—the bot's going to be immediately responsive and will know everything." Or they have this Hollywood interpretation of what AI is," says Josh. "But," he clarifies, "It doesn't quite work that way. You have to train and monitor your AI, and retain your customers through that process."
We had one more question about terminology for Josh. If you've ever taken a peek into the world of AI chatbots, you may have also heard the term NLP. We asked Josh what the difference is between AI and NLP, if any. "NLP is just the type of AI, it's conversational AI—the type of artificial intelligence businesses can use," he tells us. But Josh advises marketing agencies to avoid advertising with this acronym. "NLP is more of a bot-builder's term. I wouldn't encourage a chatbot agency to use it in their pitch. AI sounds sexy; NLP doesn't." AI is the more recognizable term, and the one that's more likely to get potential clients excited about the possibilities.
How artificial intelligence can benefit your business
If you already have a chatbot for your business (if not, find out why you should!), AI can help it respond to users more accurately. But how does that translate into tangible benefits for your company? Josh lists four ways:
#1. Funnel users into transactions 🌪️
Think of the pizza example above. A simple, rule-based Facebook Messenger chatbot may have identified the phrase order a pizza and sent both users to the order page. Since that doesn't answer the second user's question, he would probably get discouraged or frustrated and abandon the process altogether. That's a lost sale!
Instead, a Messenger chatbot with well-trained AI could understand the intent behind his question and get him the right information. Then the bot can ask that user if he's ready to order, and help him do that if so. It's all about timing, and giving users what they need while they're warm and engaged. AI can help your bot deliver the right response at the right time to satisfy customers and get more sales.
#2. Reduce support costs 💸
You can build a chatbot to respond to FAQs on Facebook Messenger in minutes. But if you take the extra time to incorporate AI into that bot, it'll answer customer service questions even more accurately. You can still incorporate live chat into either version of the bot to help it resolve more complex inquiries, when needed. But with AI, your bot will eventually be able to handle a large percentage of them itself. That means you'll save time, money, and resources. Your support staff will be available to work on other tasks while your chatbot automates responses to common queries.
#3. Automatically store actionable user data📄
An AI chatbot won't just respond to free-form inquiries from users. It can also save data from them that you can use in the future. Josh gives the example of a user asking a bot, Do you ship to London? An AI management tool like Janis (more on this below) will automatically save London as a value for a parameter called city. From then on, you'll know that this specific user will want their items shipped to London, which you can incorporate into messaging or retargeting ads in the future.
Parameters are the same as user attributes in Chatfuel. They're tags that contain bits of data about users which you can use to customize parts of the flow and reengagecertain segments of users later. Your bot will collect parameters from users throughout the conversations, building data-rich profiles just using the messages they send.
#4. Delight customers with more accurate responses, 24/7 😄
Finally, another benefit of adding AI to your bot is pure customer delight. Asking a chatbot a question on Facebook Messenger and having it understand and respond accurately is exciting. That kind of personalized experience is extra impressive and will set you apart from the competition. It'll make a prospect or customer's interaction with you a memorable one. Plus, a chatbot can provide this kind of experience at any time of day or night, 24/7.
Why you should add AI to your existing chatbot
👉 If you haven't built a chatbot for your business yet, we'll help you get started in the next section.
👉 But if you do already have a bot, you might wonder if adding AI to it is necessary.
If you've already designed a neat flow with buttons and quick replies every step of the way, why would you need AI? If users will be choosing pre-set responses instead of typing in their own messages, why would you need software to help your bot understand their intent?
Because users will still want to type in their own responses at least sometimes. Bots that offer buttons or quick replies as response options about half the time get the most engagement, according to a study by DashBot. The rest of the time, users tend to prefer typing free-form responses into the chat interface. Josh explains that "as people use bots more, they become familiar with the button UI and realize that it's actually helpful. But if the user is new to Messenger bots, and you're just trying to capture that lead, AI is really helpful." So even if your chatbot is set up with buttons and quick replies, adding AI lets it help those users who may still type in their own responses.
Because without AI, free-form replies will trigger the dreaded default answer. It's an automatic message users get when the bot doesn't know how to handle their input—like Sorry, I didn't get that. It's discouraging to users and is often a conversation-ender. "And if you lose [the user]," Josh tells us, "you're losing money, because you're paying to acquire customers and you want to get conversions and return on ad spend."
➡️ Artificial intelligence will help your bot understand these free-form replies more often, and send the default answer less often. That means more satisfied customers, more time saved by your team, and more conversions and sales for your business.
How your business can harness the power of AI
Whether you've already used Chatfuel to build a no-code bot for your business or not, Josh shared how you can get started with AI.
🤔 If you don't yet have a bot for your business:
No worries! It's easy to get started with Chatfuel. Our no-code platform lets you build a Facebook Messenger chatbot without previous experience or coding knowledge.
- First, sign up for your free account.
- Then, take a deep-dive into bot building with our free online course. Or, find out how to build a basic Messenger bot in two minutes.
- Check out our help documentation for guidance on using specific Chatfuel features.
- Take a look at some of our tutorial videos to learn things like how to map out your bot's conversation or track its performance.
- Visit our blog to learn things like how to avoid common chatbot mistakes, or how to dazzle customers with pre- and post-purchase experiences via chatbot.
- Then, follow the instructions below to incorporate AI into your chatbot.
🤓 If you already have a bot for your business:
Incorporating AI into your existing chatbot is easier than you think! We recommend using Janis.AI to help. Josh is the co-founder of Janis, which is designed to "help business manage the path to automation." It's a no-code platform (like Chatfuel) to help businesses integrate AI into their customer experience.
- Basically ,the Janis team has taken the AI technology from Google and made it accessible to businesses of all sizes.
- Then, they created an easy way for businesses to connect this technology to Chatfuel.
It all starts with the free Chatfuel template for Janis.AI. (Find it in your Chatfuel dashboard.) "It has pre-trained AI to help your bot understand the most common things users say, plus a framework you can build on," says Josh. Think of it like an AI starter kit. It'll help you start getting value from an artificial intelligence chatbot faster. Then, if you want to invest more time and take it to the next level, Josh recommends checking out their user guide. If you have additional questions, Josh invites Chatfuel users to tag him (@Josh Barkin) in the Chatfuel Facebook community.
Josh's best practices for AI chatbots
Here are Josh's top tips for creating a powerful AI chatbot:
Have your bot introduce itself 🤖
Don't try to convince users that your bot is a human. Instead, have it clearly introduce itself as a chatbot right away. And while you're still in the process of training its AI, have it share that fact too. It'll give users more realistic expectations of what the bot can do. Simply sharing this information can help them be patient if your chatbot doesn't understand them on the first try.
Customize the default answer with the three-strikes rule ⚾
Don't set the default answer as, Sorry, I didn't understand that. It's a conversation ender! Instead, Josh recommends using the three-strike rule for your Messenger bot while you're still training its AI:
- STRIKE ONE: The first time your bot doesn't understand a user's response, have it ask them to rephrase their message. The chatbot will have a better chance of understanding the input if the user shortens or simplifies it.
- STRIKE TWO: If your bot doesn't understand the rephrased message, Josh recommends having it display a gallery of options. Each card can be a category of tasks the bot can help with, like returns/exchanges, product recommendation,etc. This menu might help the user navigate to the information they need within the bot.
- STRIKE THREE: If that doesn't work, Josh advises that the bot ask the user if they want to be connected to a human agent. If they say yes, it can connect them via the Live Chat plugin.
How agencies can make more money selling AI bot services
If you're a marketing agency or chatbot agency selling AI-powered chatbots, Josh has advice for you on how to sell this service and generate even more revenue:
Have two demos for clients 👯
We asked Josh: What's the best way for agencies to convey the value of AI automation to their clients? He recommends running two demos for prospects: one of a Messenger bot with AI, and one without. That way, the client can see how AI works and choose the package that's right for them. Josh suggests thinking of AI as an add-on service to chatbot building. "It's not crucial, but it's a nice-to-have. It's just like offering either web design, or web design and SEO. Bill it as a service, just like you'd bill for SEO services."
Convey value in terms of resources saved 🌟
Another way to show clients how valuable AI is: Translate the benefits into tangible savings. Help them see how much time it can help them save in the long run. For example, they won't need X employees handling routine inquiries for eight hours a day, every day anymore. Use case studies if you have them (or fair estimates if not) to show how much hassle and resources your AI chatbots could save prospective clients.
Set client expectations 💭
Josh reminds us that many clients will have the "Hollywood idea" of what AI is and how it works. Make sure to manage their expectations appropriately. Set a goal for the percentage of free-form inquiries their bot should be able to handle by a certain date, if they choose the AI package. Let them know that their AI chatbot won't instantly be able to perfectly handle every inquiry it gets. Explain how, over time, the bot will become more and more independent of your human staff (and theirs).
Charge for regular management 🗓️
Josh points out how lucrative offering AI to clients can be. Your agency can charge a setup fee, plus a retainer to continue training and maintaining the AI in their bot. So instead of one-time revenue from a single project, you've now won recurring revenue from ongoing business.
Choose a niche 📍
Training AI for a particular niche will save your agency time, Josh advises. Let's say you serve only clients in the real estate industry, for example, or just healthcare, just retail or only banking, etc. You can train AI for that one, single niche and use it for multiple clients. Because if each client is in a different industry and you try to train AI for each one, Josh explains, "that's overwhelming and doesn't get you closer to your automation goals. Choose a bunch of clients in your niche, and all of them can benefit from the time you invested [in training the AI]." With Chatfuel, you can connect one "AI brain" to multiple Chatfuel bots, which makes it easy to copy a trained AI into bots for various clients.
Use AI chatbots to build your business
- If you run an ecommerce business, an AI chatbot can help you get more sales.
- If you run a brick-and-mortar business, an AI chatbot can help you get more customers in the door.
- If you run a marketing agency or chatbot agency, AI chatbots can help you get more clients and charge more, too.
And no matter what type of business you run, AI chatbots for Facebook Messenger can help you save time, money, and effort, and delight prospects with a helpful, engaging experience. Build a bot and use Janis to integrate artificial intelligence into it with Chatfuel today!
You can watch the full recording of Josh's appearance on our Chatfuel school web show below.👇 Catch new episodes of Chatfuel School on YouTube every week. (Sign up to get notified when the next episode premieres!)