How Chatbots Can Save Your Team’s Sanity (and Money)
- Chatbots can answer the questions your customers ask all the time — freeing up your team to handle the more interesting work
- For inquiries that require following a script, chatbots perform better. But, for inquiries that require nuance and logic, humans are a better choice.
- Chatbots have come a long way in sounding natural. Technological advances mean they can better deduce the intent behind your question, no matter what exact words you use.
- Chatbots can also be trained, so they keep getting better and better.
If you’ve ever worked in customer support, you probably hear the same handful of questions like a broken record filled with songs you hate. Your eyes are probably rolling right now, as you think about how long you’re open or whether you ship to the middle of nowhere.
Your team today probably handles all of those same questions. They might even have a copy-and-paste script with all the answers, where they just ctrl+c and ctrl+v into the email and send it off.
But it doesn’t have to be this way.
You’ve probably heard about chatbots, and wondered what the hype is all about. As it turns out, chatbots can handle these routine inquiries pretty well. Technology has advanced enough so AI can understand the intent of a person better than a human can in many cases. It has advanced enough so that it doesn’t actually matter what specific words people use to ask the questions.
The chatbot can deduce the intent of the question — just like your customer service reps can figure it out when they’re on the other end of the line.
We’re here to give you some background on chatbots and how they work, the technology behind them, and a few of the benefits.
Important disclaimer: We’re not talking about building Skynet. We don’t believe that robots will replace humans, nor do we want them to. What we’re saying is that robots can handle the boring shit you don’t want to do — like answering those same 5 questions — so you have the time to tackle all those brilliant ideas you never have the time for.
What is a Chatbot?
Before we jump straight into the deep end, let’s start with the most basic question. A chatbot is an AI-powered robot built to answer questions for humans. Businesses add them to their website for lead generation, customer service, or more likely, a mix of both.
If you’ve ever visited a website where a chat window with a picture of a generic, but pleasant-looking person with a common name, pops up and asks how they can help you, that’s almost certainly a chatbot.
Other times companies are transparent about the fact that they have a bot, because the chat window is named something like “BusinessBot” and the photo is a robot.
There might be simply space to type in your question, or there might be a menu for you to choose an option that best describes your inquiry.
Think Alexa or Siri, but without the synthesized voice and access to all your entire life’s information. In any case, the chatbot is designed to reduce the amount of rote communication your team has to respond to and make it easier for your customers to get in touch.
Chatbot or Human? Which Makes More Sense For You?
Robots can actually handle routine tasks much better than humans can — like that list of FAQs.
A chatbot (or more accurately, the AI behind it) can follow a script and respond to a template of text-based input much more efficiently and effectively than any human could.
Take our customer service rep from the introduction. Let’s say they receive an email asking about the status of an order they placed. If you’re lucky, the customer included enough information to look up an order. But there’s a good chance they didn’t. So, your employee has to reply, ask for more information and wait to receive it before actually looking up the order.
Then your employee has to find the information and send it back to the customer. All the while, the customer on the other end is getting agitated because they haven’t received the information they need.
While this example is exaggerated a bit for dramatic effect, it’s a scenario that plays out every day.
Even if it only takes 5 minutes and is relatively painless, it only takes a few repeated inquiries to waste an hour tracking stuff down info that a chatbot could find instantly (and for cheap or free).
With a chatbot, the customer simply copy-and-pastes their order number into the bot and instantly receives a shipping update. The customer is happy because they get a fast answer, your employee is happy because they don’t have to answer the same question for the thousandth time, and the chatbot doesn’t have feelings.
In fact, 69% of customers actually prefer to use chatbots because they’re so darn fast.
Remember how we said the point of a chatbot isn’t to cut humans out of the equation? Here’s why: Humans are better at tasks that require logic, judgment, and subjective reasoning. If multiple answers could be equally correct for a given question, a human should take it.
In an ideal world, your chatbot would handle the routine stuff (order status), while your customer service reps would take the more complicated questions.
In addition to freeing up your team to focus on supporting customers in a way that requires nuance, experience and logic, there are a few other benefits of chatbots:
- Bots don’t get frustrated, they don’t put you on hold, and they don’t need to queue you up to talk to you. This means a chatbot can handle frustrated customers and maybe even curb some of the frustration that customers have about waiting or not being responded to quick enough — or just looking to pick a fight.
- Chatbots can detect sentiment, so they can understand when their automated responses are pissing the customer off and kick the inquiry to a human
- You’re getting more for your money. Regardless of how much you’re paying someone to answer questions, you’re paying them every time they pick up the phone or respond to an email. Why not use a robot to handle the easy stuff, so you can pay the same for more nuanced support inquiries?
Technology to Power the Chatbot
As mentioned previously, chatbots are an obvious example of AI advances at work. Thanks to innovations like Bert and BigBird, the programming that powers the bot is getting much better at understanding natural language processing.
Natural language processing, also referred to as NLP, is basically the ability of a computer to understand the intent behind the words typed into the box. It’s kind of like how you know pop and soda are the same thing (even if soda is the correct term).
The tl;dr version of NLP is that it renders “Google-Fu” effectively useless. It doesn’t necessarily matter what words you’re searching for. Whether you’re looking for “pop” or “soda,” you’ll still end up at soft drinks.
If there is an example where the AI can’t tell if you’re looking for pop as in soft drink or pop as in music, it can send your query to a human to double-check. And in these cases, machine learning steps in — teaching the bot to never mix up the two meanings again.
Training Your Chatbot
To have an effective chatbot, there always has to be a human to train it. Just as you have to train new employees to understand the processes and procedures in your company, a human has to teach a chatbot to understand how your company handles each question.
A common misconception we hear often is that chatbots operate in a very linear way, meaning you have to ask in a specific way with specific phrases for it to understand, otherwise it glitches and your computer starts on fire.
But thanks to NLP, that’s no longer true. The way chatbots work now is they get to the root of a question, understand the topic the inquiry ties back to, and maybe a couple of hints about which direction is going. From there, the bot can synthesize what the customer actually means, and finds the answer.
For example, a client of ours is an e-commerce business that creates custom merchandise. They asked us to update their chatbot to answer questions about holiday shipping.
To train it, we fed the bot a long list of questions people would ask about holiday shipping, like “will my purchase arrive in time for Christmas?” and “How long will it take to get orders in December?”
The bot will be able to pull out keywords like “Christmas” “holiday” and “December” and serve up information about holiday shipping.
When someone asked a question that doesn’t quite fit so neatly into one of these categories, we reviewed the question and the answer the customer received, and told the chatbot whether it was giving the proper answer. From here, machine learning takes over, and the chatbot gets smarter and better prepared to answer the next inquiry.
Chatbots aren’t about outsourcing your team or getting rid of it. It’s actually the opposite. Chatbots are about making life easier for your team.
Using a chatbot to answer the routine stuff takes some of the more tedious work off your team’s plate — freeing them up to tackle the big and bold.
If you’re interested in learning more about chatbots, or building one, let’s chat. We love building custom features for your website — and we love saving you time and money even more.
Get in touch here, and we’ll be in touch quickly to continue the conversation.