Pitfalls to Watch Out for When Implementing a Chatbot

Chatbots are trending in the world of customer service, and with good reason. They promise companies better customer involvement, retention, and satisfaction.

So what exactly is a chatbot, and how can your company benefit from one? There are several things to keep in mind when implementing a chatbot of your own. Let’s take a look at a few different types of chatbots and some of the problems that can arise when adding this technology to your company’s tool belt.

Source: Andy Kelly on Unsplash

What Is a Chatbot?

As the name suggests, chatbots are automated programs designed to chat with your customers and provide quick answers to a variety of issues. Some bots are equipped with a large capacity for problem-solving, while others can handle the simplest of functions.

1. Website Bots

One of the most common types of chatbots is the automatic pop-up window that often appears on a website. Usually, these bots initiate conversation with a friendly message like, “Hello! Let us know if there’s anything we can do for you.”

PRO: Helpful to New Customers

This approach can be effective, especially when someone visits your website for the first time and may need some help finding what they’re looking for. Website bots give the impression that a friendly, professional employee is waiting on the other end of a computer ready to assist customers. Seeing that someone (or something) is available to help right away is encouraging.

As an example, Amazon recently rolled out Amazon Lex—a chatbot companion to its voice-enabled artificial intelligence (AI) tool, Alexa. Amazon designed Lex to be an advanced chatbot skilled in machine learning.

This Amazon Web Services (AWS) tool allows developers to easily create a website bot that can help customers meet an end goal in just a few prompts, all without using the manpower of an actual customer service employee. Consumers get help from what feels like a personal assistant, and companies save some time and money. It’s a win-win.

CON: Redundant or Unhelpful Messages

Depending on the level of knowledge you’ve programmed a website chatbot to understand, this helpful pop-up window may begin to feel more like a complicated way to use the search function of your website.

If a customer knows what they want before entering your site and can navigate there, a chatbot can feel like an annoyance—especially when bots respond to questions with links to other parts of your website or with messages like, “Sorry, I’m not sure how to help with that. Feel free to email or call a customer service representative.”

Source: h heyerlein on Unsplash;

2. Twitter

Social media is quickly becoming the most popular platform for chatbots. On Twitter, for example, many companies use bots within the direct messaging feature, allowing users to get prompt feedback. Another way to use chatbots on Twitter is to manage your company’s account with a bot, using an algorithm to generate automatic tweets and replies.

PRO: Prompt Replies

Most of the time, Twitter bots reply to messages instantly, helping customers feel heard. Even if you’re able to provide only a simple response in a direct message or refer them to another source, it’s useful to provide some guidance right away and let users know you’re there.

CON: Limited Responses

On the other hand, Twitter bots are often limited to a set of standard responses, which can be frustrating if a customer is looking for a specific answer.

For example, if a customer were to message a clothing company about processing a return, they might get stuck with a generic answer until an actual representative can reach out. No one wants to engage with a bot that provides a few simple answers and defers them to customer service with all other inquiries.

Twitter bots can also be problematic when they reply to tweets out of context. A notable example of a Twitter bot failure is Microsoft’s Tay. The company initially created Tay as an experiment in conversational learning, but it quickly spiraled into a social disaster. Twitter users taught Tay to praise Hitler and denounce feminism, among other offensive responses, in less than twenty-four hours, proving that the bot was not ready for public trial.

3. Facebook Messenger

Last year, Facebook launched a chatbot platform for brands to keep up with competing companies such as Kik and Telegram. In most cases, these chatbots are used for quick-reply scenarios. Instead of opening a dialogue, messenger chatbots offer designated prompts.

Source: Alex Knight on Unsplash

PRO: Quick Conversions

This platform has worked well for converting chat into sales and clicks. With companies like Domino’s, customers can order pizza directly through the app with just a few responses or with an emoji when preferences have been saved.

The Guardian is another good example of providing easily accessible info. Using the site’s bot, readers can select top stories, look for trending categories, or search key words to find relevant articles.

Fandango makes it easy to plan a movie night by offering a messenger extension that shows up when chatting with friends in the app.

Rather than navigating to a separate website, users can take advantage of these quick and easy prompts.

CON: Imperfect Algorithms

Not all Facebook chatbots are created equal. Some brands still opt to let users initiate conversation in an open dialogue, which can lead to unanswered questions, vague responses, and irrelevant information.

This is problematic for users who may be unfamiliar with the functionality of a specific brand’s chatbot. For example, in testing various brands, I asked Starbucks, “How do I use this chatbot?” and never heard back. I also asked Forever 21 for information on current sales and received an answer the next day about phone cases, with a signed name at the end of the message. This inconsistency creates confusion.

Another example of Facebook chatbots failing is the creation of Bob and Alice. Facebook created these bots and allowed them to converse, which led to a bizarre exchange between the two. While the chatbots appeared to be speaking gibberish, the broken English turned out to be some sort of secret language within the algorithm that only the bots could understand.

Until Facebook Messenger algorithms are perfected, these chatbots are better used as quick-reply help centers than discussion-based models.

There are several other platforms available for creating your own chatbot, but these three cover the basic pros and cons of the technology as a whole.

Can Chatbots Replace Human Interaction?

The biggest pitfall in chatbot implementation is trying to use a bot as a replacement for human interaction. Giving a branded chatbot all the responsibilities of a customer service representative will leave your customers exasperated.

As Tay, Bob, and Alice proved, AI is still far from successfully replicating human speech patterns. Tech experts have warned against teaching AI too much for fear of the theoretical robot uprising. Getting basic information from a bot is comforting. Feeling like a bot is passing as human—not so much. Plenty of sci-fi movies and episodes of The Twilight Zone have taught us what happens when machines get smarter than their creators.

Source: James Pond on Unsplash

If you’re looking to create a chatbot that can mimic human interaction as a means to streamline your customer service, you’re likely to fall short. This type of development not only requires a lot of time and money but it’s also almost impossible to get right. Even Siri and Alexa, with all of their voice recognition learning, have serious limitations.

But this isn’t necessarily bad news. In fact, many companies are bolstering phone communication systems to provide better customer service, proving that going completely digital is not the answer.

And when it comes to high-dollar sales, consumers still want to talk to a human being during the buying process—if not over the phone, at least through email. The same goes for customer service that requires more in-depth information for industries such as health care and travel that cause consumers to spend more time in the decision-making process.

Why Make a Chatbot?

While chatbots may not fulfill all your customer service needs, they can add value to your brand in a limited capacity. By creating a bot that helps customers place an order, schedule a reservation, or perform some other basic task, you make life easier for them.

As chatbots become more prevalent, consumers will also expect your brand to utilize them. They can be tricky, but bots can boost your credibility when done right.

Another reason to implement a chatbot is that chatbots can be fun to talk to. Consumers are intrigued by a chatbot that can engage in a decent back-and-forth conversation.

Take SmarterChild for example. In the early 2000s, ActiveBuddy, Inc., created this chatbot and implemented it in AOL Instant Messenger. SmarterChild could tell the weather, the score of a basketball game, movie showtimes, and a few other tips, but the main appeal of SmarterChild was to see how it would respond to asinine questions and prodding.

Source: Drew Graham on Unsplash

Although SmarterChild didn’t live on to see the likes of modern-day chat, it was definitely a forerunner in the AI industry. Its small-scale functionality paired with a dash of sassy back talk served as a great example for chatbots of the future.

What Is the Future for Chatbots?

So, what does the future hold for branded chatbots? Too smart, and they’ll take over the world. Too basic, and they’ll fade into obscurity. The tech industry must continue to find a balance between conversational learning and basic functionality for bots to succeed.

The rising generation will be especially dependent upon AI and instant-gratification technology. For these consumers, chatbots will become paramount to daily interactions with brands. Today, you can order pizza by typing an emoji. In the not-so-distant future, perhaps chatbots will allow consumers to create websites, sell goods, and more.

In the meantime, as you implement your chatbot, be sure to develop one that has enough knowledge to be useful, fits the platform it’s been designed for, and follows a well-structured algorithm. Getting it wrong can do just as much harm for your brand as not trying at all.

This field of technology has a lot of room to grow, but chatbots are definitely worth implementing on your company’s website or social media platforms.

Note: The opinions expressed in this article are the views of the author, and not necessarily the views of Caphyon, its staff, or its partners.

Author: Shea Drake

Shea Drake is a freelance writer whose focus is MarTech, IoT, and Augmented Reality, particularly in how they’re shaping businesses both large and small. When not writing, she enjoys photography, Marvel movies, and traveling. You can find more of her writings on Twitter, she’d love to hear from you.

Leave a Reply

Your email address will not be published. Required fields are marked *