Chatbots: After The Hype… What’s Next?
Everybody is talking about chatbots. They have become that cool new toy on the market that companies put high on their wish list. Chatbots promises to be faster, better and cheaper than their human alternative.
Research firm Forrester reports that globally 57 percent of companies are already using chatbots or plan to introduce one this year. These include chatbots for frontend use as well as backend use. It’s clear, that chatbots are being hyped more than ever but what is the reason for that? And are they living up to the hype surrounding them?
What started the chatbot hype?
During the last decade, we have become increasingly familiar with the concept of chatbots. Their popularity has been increasing extremely fast over the last couple of years and is in high demand in 2019. However, while a hyped product typically is followed by vast growth in a short period of time, chatbots are different, since they have been around for quite some time now.
In fact, the first proof of superficial contact between man and machine took place in 1966 as the chatbot Eliza was created. Fast forward to MSN, which is a web portal developed by Microsoft and launched in 1995. Here, we also encountered a chatbot that would enter conversations with its users and ask about simply anything. People were ecstatic. This was the start of the technological breakthrough that people had been waiting for. The bot would give standard answers to messages that were recognized by non-complex sentences or even just keywords. For a long time, chatbots’ purpose was to do just that – and even though these features and applications on platforms boosted the chatbots’ reputation wildly, these bots were not smart.
However, this was only the beginning. Since then, the development of chatbot technology has taken off, and the bots have become increasingly smarter. Chatbots, such as Apple’s Siri or Google’s Alexa, have become virtual assistants, that are accessible directly to consumers.
Following the technological advancements of the latest years the chatbot trend has grown a lot and when Facebook launched a platform for developers of chatbots, the trend grew even faster. At this point everyone wanted chatbots, expectations increased extremely fast and messenger began overflowing with bots. But then it all kind of fizzled out. But why?
Establishing chatbot ground rules
Typical for emerging technologies, such as the technology of the chatbot, is that it is extremely hyped when first exposed to the world. The ‘Gartner Hype Cycle’ provides a view of how emerging technologies will evolve over time. At first, product innovation will trigger a peak of inflated expectations and create hype. Afterwards, it will reach the stage of disillusionment where companies will hit the wall and either break it or make and hopefully go through the enlightenment phase. In the end, the product will reach the ‘Plateau of productivity’ where it shows its true nature.
In relation to chatbots, a lot of companies rushed to jump on board to get ahead of their competitors. “Despite enormous advantages, most chatbots fail because of an unclear purpose and poor planning”, warns Wang Xiaofeng, an analyst at Forrester, adding that companies should first establish clear business goals and key performance indicators. And in a lot of cases, companies have failed to establish ground rules for their bot, which we will demonstrate later in the blog post.
There are other factors contributing to the failure of a chatbot. One factor might be that companies have set targets that are simply too high for the capabilities of chatbots. Another factor might be hurrying the launch date in order to be the first between competitors to check the chatbot marker ✅ This is a big mistake as first-time success is vital for any hope of the consumer reusing the chatbot.
When new technologies make bold promises, it is challenging, to say the least, to discern the hype from what’s commercially viable.
Bots for the sake of bots
When bots were the talk of the town – or the internet – companies asked themselves if they truly needed a bot and for what purpose? – or at least they should have asked themselves that. It was a classic case of the nature of hype: it tends to ignore essential questions like these. In the past couple of years, bots were being blindly applied to problems where they just weren’t needed. Either that or they weren’t developed well and didn’t serve their purpose well.
In theory, every company can benefit from a chatbot, but the bots don’t have to find their purpose in the customer service department by default, as there are other areas within a company that just as well can benefit from using AI. In order to implement a bot, you must have the patience to build a bot that can do what you want it to do.
Let us put it simply: building a bot – just for the sake of it – will most likely do more harm than good. However, when done right, implementing a bot can work as an amazing improvement to a company’s customer service, if they view and use it as an extension, and not a replacement of customer service assistants entirely.
Bots vs. Apps?
There was an exaggerated assumption that apps were ‘over’ and that they would just be replaced by the new, fancy bots. However, this was quite a misconception. It’s important to view bots and apps as two separate entities designed to serve different purposes rather than trying to make bots solve the problems that we preferred being solved within an app.
The reputation of chatbots hit an all-time low when they took over Facebook Messenger. According to Silicon Valley’s blog The Information, the 100,000+ AI bots on Facebook Messenger hit a 70% failure rate in completing simple user requests in February of 2017 and after this discovery, Facebook decided to reduce the effort they had put into chatbots on Messenger. This is a result of developers failing to narrow down their bot to one prominent area of focus. Instead, they tried to make the chatbot an ‘all-rounder’ that could partly do everything, yet not successfully fulfill ONE request.
Most bots are built on a decision-tree logic, where the bot’s automatic response depends on spotting particular keywords in the user input. The advantage, as well as disadvantage to this, is that it’s very easy to list all the cases that they are designed to cover. A bot is merely a reflection of the capability, preciseness, and patience of the creator of the bot, as well as their ability to predict the users’ needs and their inputs.
Chatbots replacing human assistants in customer service
A common misconception regarding chatbots during the past years when chatbots were on the rise was that chatbots were going to completely take over the customer service. However, this was not the case. Platforms tried to do this, but as a result, customers could only receive help on very simple questions. There were heaps of chatbots that were given the task of answering every question, disregarding the level of complexity, which resulted in confused customers with unresolved problems.
The secret of a successful application lies in the perfect combination of technology and people. It is vital for successful implementation to remember that chatbots should only contribute to the customer service experience. Instead of thinking of chatbots as a replacement of humans in customer service, it should be viewed only as an extension – at least for now. Consider the chatbot as a function that will help to drastically reduce the response time and lead time of the conversation. When customers have a simple question, they expect a fast response. This is where the chatbot can replace people, as they can do this better than humans. Leave the more complicated questions for the customer service employee, and employees can be utilized more efficiently in order to take special care of the customers with these complex questions.
Where we usually see a lot of opportunities and possibilities with AI and chatbots, Gartner now predicts that the hype of chatbots has passed its peak and is leaving the phase of “Peak of Inflated Expectations” and entering the most defining phase in terms of the future of chatbots; “The Slope of Enlightenment”. This is where all the possibilities lie, and where one must draw learnings from practice and mistakes. Now we are discovering that companies are creating chatbots that are extremely well made and that provide impressive benefits and results.
What does the future look like?
By now, the chatbot has been through the hype cycle. As of now, people are having conversations with their bots in their homes. The bot is ordering pizza for us, it’s our DJ and it’s telling us whether we should bring an umbrella to work in the morning. But what’s next? Have the chatbots lived up to their full potential and the hype we’ve created for them through the last decade?
No, there is a long way to go. The faith in chatbots must be restored, which will happen when companies implement it in the right manner. Platforms that want to have a chatbot must find the perfect balance between technology and people. When this is done, customers will be accustomed to relying far more on the chatbots as “real-life assistants” or simply as the experts.
According to Facebook IQ data, there has been extreme growth within the activity of chatbot discussions in a period from January 2017 to January 2018.
It is already clear that the world of chatbots is experiencing growth – and it is increasing fast. In fact, many countries are already experiencing vast improvements in their industries from using chatbots. Along with other Asian countries, Singapore is performing extremely well with the use of chatbots, and the number of chatbots that are implemented has been increasing for the last few years. This is thanks to their government which, for a while, have been investing heavily in the tech sector.
“In the Asia-Pacific, Singapore is probably the most advanced in terms of investing in chatbots, companies are more willing to experiment and like to explore emerging technologies. Customers here are also digitally savvy and have high expectations.” Wang Xiaofeng, Analyst at Forrester.
In Singapore, banks such as Citibank, DBS, and OCBC have already launched chatbots or have plans to do so soon. Singapore has had continuous commitment to keep R&D spending at 1% of GDP. They recently pledged to invest 19 billion Singaporean Dollars into scientific and technological research as part of its Research Innovation and Enterprise (RI)E 2020 plan.
Singapore isn’t the only country going through this transformation. China, taking the lead, along with many other Asians countries are investing heavily in chatbot technology, and in most cases, the chatbots are used within the banking industry. The purpose of the chatbots, in this case, is usually to give customers easier access to their bank information or financial advice, instead of customers having to spend half a day at their local bank office, waiting to be assisted.
Google Trends search for the term chatbots from 2014 – today.
We’ve been through the Hype Cycle and now, we are ready to make real progress. We’re moving away from a stage where everything was black and white, and we can start to examine the middle-ground.
The overestimation and the anti-climactic era are completely normal for emerging technologies and is by no means an end to the work that is to be done with the bots. Companies will continue to invest heavily in chatbots, and developers will continue to improve the technology, such as Natural language Understanding (NLU), behind chatbots.
Do chatbots belong in customer service? Definitely! But that’s not necessarily the area your company should implement it in. Customer service can never be fully automated, due to the need for empathy and human connection. It is all about finding the perfect balance between technology and people. The trend line of chatbots is evolving every day and the forecast for the future of chatbots is looking bright.