Hi. My name is Amy. I am a former Chatbot hater.
It wasn’t until about a year and a half ago that I saw how much value they could add.
Don’t get me wrong. When I need to increase my blood pressure, I visit a home improvement site like {Redacted}. Chatting with their emo bot with more existential-angst than Sartre is guaranteed a 20-point systolic increase.
However, these new AI conversational bots? They’re not all that bad. In fact, some of them are pretty useful. So, let’s talk about what you need to know about AI chatbots and what makes a good one.
FIRST, WHAT IS A CHATBOT?
A chatbot is a computer program used to conduct online conversations. Chatbots were originally text-based. These days, you can interact with them via voice, touch, text and video.
WHERE ARE CHATBOTS USED?
Chatbots are used in almost every industry on websites, devices, and mobile apps. As you’d imagine, the most common chatbots are Customer Service chatbots. With that said, more and more sales organizations are using them effectively. They’re beneficial because they’re available 24/7/365.
HOW DO CHATBOTS WORK?
The user writes a short sentence or two, and the bot responds to what the user typed, said, or touched. The String (number of interactions per conversation) is usually between 1-5 (yes/no type conversations) or ~20 (more detailed conversations.) Incidentally, if you’re a marketer using chatbots in your business, measure your String numbers. Folks use them for reporting and benchmarking, but they’re also helpful in quickly spotting problems.
AI Chatbots can respond with yes/no answers, and they’re also able to provide human-sounding detailed answers. They excel at things like answering frequently asked questions and providing links and resources for something within your site.
If the chatbot can’t handle a question, it can redirect the user to a live agent, your knowledge base, or wherever you tell it to go. (When you build your chatbot, you’ll design a conversation flow that outlines how you want things to work.)
WHAT TYPES OF CHATBOTS ARE THERE?
THERE. ARE. SO. MANY. NAMES. FOR. CHATBOTS.
Totally understand why marketers ask this question, especially if you’re comparing vendors. (Many use different terms/language or have special names for their chatbots and behaviors.)
First, let’s talk about the difference between rules-based and AI chatbots.
Rules-Based Chatbots – aka Menu Bots, Flow Bots, Button-Based Bots, Radialbots. These Old School bots guide the user through a specific process (think decision tree or flowchart), giving them several options to choose from along the way. For example, the chatbot greets you with “WHY ARE YOU HERE TODAY?” Underneath are several answers, such as “I need to talk to a sales rep” or “I need more info on installing my product.” These bots are rules-based and typically can’t answer anything outside their pre-determined rules. This type of chatbot does not learn; is VERY limited; can’t hold a conversation for long; are noticeably not human; and their success depends on how good you are at pre-programming them. So, if you know what questions people will ask, they can be useful-ish.
Note: I am biased against these bots because they drive me bonkers. However, it’s important to note that we have clients who use them successfully for specific tasks. For example, if you have a lot of traffic coming at once after a radio/TV ad, and the majority needs to be pointed to a specific landing page or process, they can be very helpful. Also useful if you are inundated with service calls after a new product launch and need to get people to the right place to watch a video or something. Stuff easily assignable to a flow chart or yes/no response tree is where they excel.
AI Chatbots – AI chatbots use Natural Language Understanding (NLU) algorithms and deep learning to mimic human responses. They are designed to understand user intent and context based on how/what/when the user writes, touches, or says. Because they improve over time, some of today’s chatbots feel so much like you’re talking to a real person that you forget you’re talking to a machine.
AI bots need to be trained, and the rule of thumb is that the more you train them, the better they’ll be. It’s important to remember that old-school chatbots don’t learn, whereas AI chatbots do, so the more data they collect and analyze, the better they can become. This typically allows them to provide more accurate and up-to-date answers than traditional chatbots.
AI chatbots are better at context too. This often means they are better at escalation, which is one of the biggest problems with conventional chatbots. (People get frustrated with the never-ending loops they can get caught in while using a chatbot.)
Within AI chatbots, there are several categories. The most common are utility chatbots and informative chatbots.
Utility Chatbots – these do things like help you control your lights and other smart home devices (Ex. “Alexa, Reading Room Light On.”)
Informative Chatbots – they’ll help you when you ask what the weather is, if a flight/train or something is on time, or if their package has shipped.
WHAT ARE THE BENEFITS OF AI CHATBOTS?
There are many benefits to chatbots….
Always available. 24 hours a day, 7 days a week, 365 days a year. This is helpful when you can’t afford around-the-clock human support.
Versatile. They’re good at all sorts of things, from pointing users in the right direction to pre-qualifying leads before you send them to sales reps and live agents.
Free up human hours. AI Chatbots do a lot of the gruntwork so your team can focus on complex tasks and strategies instead. Again, they excel at repetitive requests and easy-find answers (ex: is Item X in stock?)
Decrease customer response time. They can reply to many users at once, and they’re much easier to ramp up than humans. (No hiring!) This increases customer satisfaction and can improve engagement scores.
Enhance Personalization. Able to create bespoke journeys for your users. Many companies use them to guide their users through a specific process.
Reduce errors. Can create strong branding. Because they’re trained, they’re often more consistent in what they say than humans.
AI chatbots make excellent junior analysts. Because they collect and analyze oodles of data, they can provide you with a constant stream of insights and recommendations. Yes, some of their findings are complete garbage, and another chunk is ho-hum, but it’s worth it anyway, as they tend to find trends and patterns that humans don’t see.
WHERE DO MARKETERS USE CHATBOTS MOST?
It’s easy to see how Customer Service benefits from AI-enabled chatbots, but where should you use them if you’re a marketer?
Improving your lead capture.
Chatbots provide another outlet for customers to “talk” to you. During your conversation, you have an excellent opportunity to collect something of market value (email address, mobile # for texting, etc.) and to ask some questions about their intent and preferences. This allows you to provide a better experience while visitors chat with you and in your follow-up triggers, text messages, and calls. Plus, if you are using dynamic personalization on your site, it allows you to gain insights into what they want/need from you, so you can change your site to be more reflective of that for current and future visits.
With AI chatbots, your conversations are stored so you can look at each conversation individually and as a whole. You can look at all your chats in their entirety or in groups and clusters. The data you can collect is a veritable goldmine for sales, merchandising, content, and more. Not only can you look at data by what people say, but you can also verify it (or not) based on their behavior on your site afterward. This will help you figure out where/how people struggle.
Are you using a chatbot in your marketing efforts? Have any tips you’d like to share? Tweet @amyafrica or write info@eightbyeight.com.
This article is Part 1 in the Chatbot series. Read Part 2 and Part 3 here.