Table of Contents
So, what is a chatbot exactly?
To understand more about what chatbots are, you need to first understand something about machine learning.
To start off with, programming was simply a way that a human told an algorithm to act. It was an If-then scenario. So, if someone presses A, they get a chocolate bar. If someone presses B, they get a muesli bar, for example.
This was okay, as long as it was fairly simple to explain the “If” situation to the algorithm. In the early days of computing, this was more than simple enough. But what happens when it is hard for the human to explain the “If” situation as a result of sheer volume or because it becomes too complex?
In these cases, you need an algorithm or bot that can create new algorithms itself. These algorithms are created with random combinations and links. This is where machine “learning” actually starts.
Now, in one way, machine learning is similar to human learning in that it involves the use of different pathways. The difference is that the machine is essentially guessing what the answer is, without understanding anything.
Let’s say, for example, we want a bot that can recognize a balloon. It’s pretty simple, but remember that there are many different balloon shapes, sizes, and colors. For a human, it’s not that difficult, but that is because we have experience with balloons and know how to recognize them.
Can you think back to the first time you saw a balloon? You wouldn’t have known what it was. Someone would have had to tell you. You would, as you grew older, have seen other balloons and so you were able to recognize the different shapes, colors, etc.
Your bot or algorithm also needs someone to say, “This is a balloon.” It needs to be fed different examples of balloons. Then, the bot needs to be tested. You would show it pictures of balloons and other items, and it would have to say if each item were a balloon or not.
Not all of the bots would do well with these tests. In fact, not many of them would. Those that did poorly would be scrapped. Those that scored better would be used as a model for building anew bot. Different versions of those configurations are tried, and the testing begins again.
So, in each generation, the bots that get the answers right the most often are used to create the next iteration. This is what enables them to get better at recognizing balloons over time. The more data they have access to, the more accurate they become.
They “learn” but not quite as we do – they might, for example, interpret a beach ball with a handle attached to as a balloon.
So, it is not intelligence as we understand it. That said, the bots do become quite complex in design, more complex than would be possible for a human creator to make. And they are able to process volumes of data that we just are not.
They might not always be right, but they usually will have a high degree of accuracy. Combined with the speed at which they can process data, this makes bots incredibly useful.
Chatbots in Day to Day Life
You might not have realized it, but you have come across bots before. Whenever you run a Google search, or look for something on YouTube, it is a bot that comes up with the search results. Speaking of YouTube, you know that list of recommended videos that is lined up for you? That’s the work of a bot as well.
But bots can do so much more than that. In fact, the uses are almost limitless, so in this post, we are going to talk about chatbots in particular.
What is a Chatbot?
These bots undergo the same kind of programming process that we spoke about earlier. Obviously, though, it’s a much more complicated set of questions asked when they are “learning.”
Chatbots are designed to respond to either text or voice communications. As the eBay sales bot says, “It is as effortless as talking to a friend.” Most chatbots today have been programmed to recognize natural language.
What that means is that they should be able to understand you, even if you phrase things a little bit differently.
Of course, some bots are better than others. Nordstrom came up with a chatbot to help you find the best deal that matches your preference, and it was hailed as quite a success.
The chatbot, in this case, asked a few questions to find out exactly what kinds of things the shoppers were looking for. This included questions about the person they were buying for so that they could narrow down the results more.
Many customers enjoyed using and found it extremely useful. As a result, Nordstrom gained points for customer service and got a lot of good publicity.
Are Chatbots Always the Answer to Boost Sales?
Unfortunately, not all projects have been as much of a success. For a business, this can be a nightmare. Businesses should not rush to get their chatbots out to market, just for the sake of being on trend or they might risk alienating their clients.
I am going to use a personal example here. I recently tried out the eBay sales bot. I wanted to like it – they offered me a free gift with my purchase using the bot. My first experience was pretty dire.
I typed in, “Kuretake Gansai Tambi” because I was looking for water colors by that brand. The bot came back to tell me that it only spoke English and programming language. Fair enough, so I typed in “watercolor” and was told the same thing again.
That annoyed me, so I logged out. I went back in later and tried inputting watercolor again, and this time I was shown some results. (Clearly, the bot considers all the searches you have run in the same session.)
It did show me some results, but, to be honest, it was rather clunky. I had to press a button to see the top matches, and then press another button to see more matches. I realize that this sounds a bit like whining but, for me, it is easier just to go onto the site and run a normal search.
However, I won’t give up on using the bot just yet. It does have a feature that allows you to input your clothing sizes and that could be a useful time saver in future. You do also get to favorite items so that the bot can learn more about your personal tastes over time.
So, yes, it might prove useful later on. Is it the answer all the time? No.
And that is what companies do need to realize about chatbots. If customers get frustrated with the bots, it looks bad for the company and could result in lost sales. The only reason I even went back in to try again was that I was writing this, and I wanted to give the company a fair chance.
Would I use it again? I doubt it. I wasn’t blown away by the suggestions it offered. Other users might not have the same negative experience so they might love it. For now, though, I’ll run my own searches, thanks.
How Can Using a Chatbot Benefit a Client?
Boosting sales and giving shoppers a taste of what it is like to have a personal shopper is not all that bots are good for, though. They can also be used to deal with customer queries.
Say, for example, I wanted to find out if the company sold a sweater in green, I could easily get that kind of answer from a chatbot. If the chatbot then also came up with links to the item, it would be an ideal sales opportunity.
Most of us want answers to our questions as quickly as possible. Being able to provide them instantaneously could give a business a significant advantage over its competition.
It is quick and easy, and they don’t even have to put its phone down. Because the bot can process so much information quickly, it would also be able to tell me straight away whether the item was in stock and what sizes were available.
Let’s go back to the sweater in green for a minute. Let’s say that I don’t buy it straight away. I might want to think about it, for example. But somewhere at the back of my mind, I will remember that the company did have it in green. When I was ready to buy, guess where I would go?
How Can Using a Chatbot Benefit a Company?
The company benefits because clients are happier. Clients have a more positive experience and so view the brand more positively. The benefits go beyond happy clients though. How many calls or emails do you get a day asking pretty basic questions?
Now imagine if all those queries could be handled by a chatbot. Imagine all the resources that would open up as a result. Your staff would not have to get horribly bored dealing with the same old question time and time again and would have time to get some real work done.
Also, you would always know that the clients were getting a consistent message. The bot will answer the same question the same way every time. The bots are incapable of making things up so if they don’t know the answer, they will say so. A human staff might make something up to prevent themselves looking foolish, or inadvertently give the wrong answer.
It is true that bots are not perfect – there could still be issues that creep in, but bugs can be fixed fairly quickly. The bots that we are using at the moment are a lot more advanced than the first one created back in 1966, but they are still really only at a fraction of their potential.
As machine learning begins to advance more and more, what the bots are capable of doing will as well.
What to Expect of Chatbots in the Future
That chatbots are here to stay is certain. They make it possible for businesses to handle basic customer service issues simply and reasonably efficiently, and also provide the companies with the ability to gather a lot more data on their clients.
Chatbots will increasingly be used to take over more traditional roles in customer service such as call centers. They will also become a lot more sales oriented, making it possible for clients to find answers to questions and make it easier for them to shop as well.
One area that has been underutilized at this time but that is sure to become more important in future is that the bots are able to collect and process a lot of data with each customer interaction.
Take, for example, my interaction with the eBay bot. It could record the conversation and note that I asked about Kuretake. (It would, of course, have to recognize this as a brand, so there is some work to be done here.)
In future “conversations” with the bot, it could remember that I previously showed a preference for that particular brand and show me other products from the same company or products that would work well with Japanese watercolors.
There are many permutations of what the company could do with all the data. Maybe they could use the data to boost sales. Or use it to ascertain the possible market demand for a new product.
The key here would be that there would be a lot of raw data that could be used strategically in a number of different ways. Savvy companies would increasingly take advantage of the data made available to them in this manner.
Chatbots could become every marketer’s dream – they have a direct line to the end consumer, are able to process vast amounts of data, and clients are willing to provide data quickly and easily using this communication channel.
As chatbots become capable of performing even more functions, it will be very interesting to see how they develop in the future.
*Articles on this blog contain affiliate links that GMP receives a small commission from at no cost to you. We only recommend tools we trust. And never ones we don't trust.