It was only a matter of time that businesses would become interested in chatbots.

 

Customers are becoming harder to reach through advertising, apps and search. Even when you do reach them, you either need a human being to interpret their requests, or an extensive set of automated tools. Both those options can be expensive, and if they can’t service the user’s request on the spot, then you may lose them as a customer for good.

 

It’s a basic marketing problem: where does our audience spend its time? How can we reach them when they’re there? When the answers were simple, like ‘at the football’ and ‘put up a billboard’, marketers could breathe easy.

 

But now, marketing, IT and sales are entwined, and the best place to find your audience – virtually any audience, for any product or service – is online, and increasingly, in online chats. This makes the chatbot an obvious, and highly effective, way to provide customer sales and service – if your technology’s good enough.

 

Speak to me!

At its simplest, a chatbot is an automated system that can respond to human voice or text inputs in real time, and we’ve had them for decades. Modern chatbots, on the other hand, rely on sophisticated natural language capabilities to understand the meaning of what a customer is saying, and offer appropriate responses in real time, like making a restaurant booking, scheduling an appointment or offering a product for sale.

 

Some digital assistants, like Apple’s Siri, use natural language engines built around carefully constructed rulesets (like grammar and syntax). It’s a heavyweight approach that requires constant upkeep.

 

Facebook and Google have gone down a different avenue. They’re constructing deep neural networks (DeepText and SyntaxNet, respectively) that use bulk data analysis to understand how language works. Conceptually it’s similar to how neural networks learn to recognize cats – by accessing and analysing huge datasets (that is, millions of images of cats).

 

Ones and zeros

This means the neural networks’ success is largely determined by their access to as much data as possible. In the case of chatbots, the data required is natural human speech – which gives any company that runs a successful social media or online sharing platform a huge advantage. And Facebook and Google, as it happens, have access to some of the world’s biggest online conversation spaces, forums and chatrooms.

 

Another advantage of the neural network approach is that the network doesn’t need to be told the rules of the language it’s analysing, as it learns them on the fly. This makes it easy for the system to learn more languages, as they don’t require linguists or other experts to first carefully describe the language’s structure and rules.

 

What’s the message?

How is this relevant to users? As always, the technology is trickling down and there are already developers releasing chatbots into the wild. They’re easy to use, provide ready access to huge and largely untapped audiences, and will only get better (more natural, and more connected to services and other functionality) as time goes by. They can make using your services easy, natural and even enjoyable. Now might just the right time for your business to say ‘hello’ …

Leave a Comment

Click to leave a comment