Chatbots are evolving in leaps and bounds and are now able to be much more than just glorified FAQ sections. They also possess a much nobler history than many of us are aware of. So, before we explore how far they’ve come and all the exciting things they can do CFD brokers, let’s take a quick stroll down memory lane.
The Turing Test
Chatbots have actually been around since the very dawn of AI research. In 1950, mathematician and computer scientist Alan Turing proposed the now famous Turing Test. Designed to assess the degree to which machines could exhibit human-like intelligence, the test worked by having human participants carry out text-based conversations with computer programmes.
Human evaluators would then observe these conversations without knowing which of the two parties was the programme. If the computer could fool the evaluator into believing that it was the real human in the conversation, then it was said to have passed the Turing Test.
One of the first ever chatbots was actually an attempt to take Turing up on his challenge. Created in 1966 at MIT by Joseph Weizenbaum, ELIZA is the great grandmother of the kind of chatbots that we’ll be discussing further on. It was able to conduct strikingly human-like conversations without having almost any context about the human world or any ability to comprehend it. It was partly created as a critique of the test itself, the idea being that Turing’s test could be gamed by clever engineering. ELIZA worked by identifying and ranking keywords in its interlocutor’s sentences, then producing appropriate responses by referring to a script. One of ELIZA’s scripts was modelled on the psychologist Carl Rogers’ method of interacting with his patients. Like a Rogerian analyst, ELIZA would elicit further information and keep the conversation going through the use of open-ended questions.
ELIZA’s generation of chatbots were only able to perform parlour tricks. Today’s bots exist in a radically different milieu of almost total interconnection and big data, where understanding what human beings want and being able to make it so is an enormously valuable commodity.
At Devexperts we’ve been developing both client- and broker-facing technologies since 2017. Our expertise in building the components that literally allow brokerages to tick, has given us an insight on the impact that chatbots are set to have on the retail trading space.
The online trading industry is perpetually on the search for new viable channels to attract, engage and retain clients. In our view chatbots, or digital intelligent assistants enhanced with AI, open up the largest one of all to them. According to Statista; WhatsApp, Facebook Messenger and WeChat are currently the top three messaging apps in the world, boasting 4 billion users between them. They’re not just popular compared to other chat apps, they’re fast-becoming the most popular apps, period.
In 2018, Apptopia studied worldwide app session data (excluding China and other countries using third party android stores). Their research showed that over a three month period users spent more time on WhatsApp than on any other app. WhatsApp users racked up an astonishing 85 billion hours between them. WeChat came in second, Facebook third and Facebook Messenger fourth. Like email a generation ago, chat apps are now the communication channel of choice, allowing businesses almost frictionless interaction with new and existing clients.
What Can it do?
Correctly integrated, today’s chatbots can relay information from and interact with a number different brokerage systems, even performing actions on behalf of the client. It all depends on how comprehensively the chatbot is integrated into the brokerage’s systems. The more data you give them access to, the more they can do for you. A minimal integration allows your chatbot to function as a first line of customer support, quickly answering routine questions that human teams are usually inundated with, then escalating other more sensitive enquiries to human agents when the need arises. This alone saves costs, improves service and frees up human agents to respond to more pressing issues. When allowed access to your brokerage’s price feeds, a chatbot can be used to inform clients about asset prices as well as providing them real time alerts should a symbol they are monitoring make a sudden move. Does your brokerage subscribe to any third party providers? Chatbots can also be made to play nicely with these systems, allowing the news and analysis you pay good money for to be delivered directly to your clients, without them having to interact with anything other than their chosen chat app or smart speaker.
It does not end there; more comprehensive integrations yield more complex client-bot interactions. Connection with the user’s trading account allows the chatbot to inform clients about their account balances and other metrics such as trade history, PnL, even warning them when their margin level is approaching 100% and eliciting further deposits.
Fully integrated, today’s digital assistants can perform the role of personal broker, opening and closing positions on behalf of the client through text or voice-based instructions. One of the most stubborn pain points that almost all online brokers feel is how to simplify the process so as to open up the experience of trading to a wider client base. Everyone ought to have access to markets, but not everyone is destined to be a chartist. A well-programmed chatbot bridges this gap, allowing you to offer full trading services to clients who never have to look at a trading platform. In fact, they never even have to leave the chat application that they are already using.
Make the most of your data
Like all AI systems, chatbots gets smarter with access to more data. They learn from their interactions, allowing brokers to offer a radically more personalised service than is currently possible through other means. Marketers pay lip service to the idea of tailored services all the time but in practice anything bespoke at scale is a logistical nightmare. Chatbots promise to deliver this holy grail of all business; personalisation at scale. In the human world this means your local barista remembering your preference for unsweetened coconut milk. In practice it’s enormously labour intensive, usually requiring a human being for each and every bespoke interaction. In the digital realm, our usage data allows machine learning algorithms to give us more of what we want at no extra cost to the business. Online brokers are traditionally very protective of their information and are highly conservative when it comes to allowing outside access to it. This attitude is about to change as the industry starts to understand the competitive advantages that AI systems confer. Particularly in the area of sales and retention, providing access to normally hoarded data pays dividends.
For instance, an unfunded, unverified account repeatedly asks your brokerage’s digital assistant for Tesla price quotes. Following a 10% drop in its stock price it alerts the client of the dip and provides the means for them to verify and fund their account before allowing them to buy said dip. In other words, the effort of regularly following up on the account is completely automated and does not require a single minute of a costly human labour.
Cue more stats
Back in 2016, Oracle conducted a survey of 800 decision-makers, who revealed that 80% of them were either expecting or actively preparing for their companies to be using chatbots by 2020. That same year Markets and Markets Research estimated the value of the chatbot market at $703mln. Fast forward to 2019 and Markets and Markets re-estimated the size of the market at $2.6 billion, with expectations for it to grow to $9.4 billion by 2024.
In the same year, Juniper Research estimated that by 2023 chatbots could help businesses across the banking, retail and healthcare sectors realise savings of $11 billion annually, up from $6 billion in 2018. In 2019 the firm also estimated that by 2023 successful chatbot interactions in the retail space would surge from 2.6 billion in 2019 to 22 billion by 2023.
The trend is clear to observe and it’s very much a case of get on board or be left behind. Not because this is some new fad or flavour of the month, but because the competitive advantages a well-integrated chatbot can confer on client and broker alike are difficult to deny and impossible to imitate.
Successful platforms quickly lead to network effects that are almost impossible to recreate once they get going. There was once no real viable competitor to email marketing until social media, YouTube and podcasts. Similarly, the combination of the World Wide Web, the ubiquity of the smartphone and affordable mobile data have led to a chat phenomenon that we feel is only getting started. Chat is fast becoming the quickest, cheapest and most personal way to touch base with your customers. Through next generation chatbots, the kinds of interactions you’re now able to have with them at scale are limited only by your own ingenuity.
Devexa is the name of the broker bot Devexperts engineers have been working on. She allows for a truly bespoke service for all clients, irrespective of their level of experience or the size of their account balance. She also offers brokers a multitude of options for keeping traders engaged and informed in a manner that would have been impossible just a few years ago. For example, Devexa can learn that a client mostly trades gold and oil and is then able to recommend a trading competition that’s currently running on those assets, or a webinar on how global geopolitical tensions are likely to impact their price.
Trading signals and ideas can also be automatically communicated directly to the client via chat, informing them about the assets that they have historically shown the most interest in. They can then find out more details about the strategy or even trade the asset in question directly from within the app. It’s like going back to the good old days of having a personal broker, only it’s a bot that’s informing you and getting your orders filled. Future iterations of Devexa will be able to handle more complex instructions, where she will convert a request such as “I want to buy 1 lot EURUSD when the market drops to 1.02” to enter a buy limit order for 1 lot at this rate.