The future of customer service – AI aided responses

By now many businesses have embraced the power of online chat support. The generation that grew up with the internet prefers it to any other form of communication, it’s fast and efficient, and the service providers are getting better and better at it.

It is hard to imagine a web-first business without chat support

There’s much to improve however: a good chat support operation requires a lot of man power, because otherwise it either seems like a ghost town or the queue times get so huge your customers leave before they get an answer – which is one of the leading causes for churn.

It’s a hard problem to tackle: you must have it, and you must staff it. In other words either you do it well, or you shouldn’t do it at all.

Furthermore, as we change the way how we think about the internet as users are ditching desktops in favor of their phones, with Facebook pages becoming more and more important to companies, simply having a web based chat is not going to cut it: you have to talk to your customers through messaging applications.

Why is this important?

At the end of the day consumers are suffering because businesses are slow to adopt to these changes – perhaps companies see it as a scaling problem. Chat support on for example Facebook Messenger is a great new channel and opportunity for companies to meet their customers. We still, however, see companies try to actively HIDE their chat support instead of putting it directly in front of their customers.

In the long run, businesses that won’t embrace this change will lose out to those who take it as given. It’s a disruption that’s happening in every market from shopping to banking.

So how do you tackle this customer support disruption?

First, you should be acquainted with your data: what are your customers asking your support about? The old solution to this was – update the FAQ (and send your impulse buyer down the labyrinth of badly organized web pages. Yuck.)

If you have a good grasp on your data, you should know that the power rule applies – a small handful of topics make up a significant portion of your chats.

So if you have a good support team, they’ve prepared some canned responses – helping these customers while maintaining a manageable chat queue.

But that’s not going to cut it if you’re talking with your customers on several channels – not if you want to maintain quality (nothing is worse than bad customer service).

At a critical size, you have to scale your operations – support more languages, hire more people. There’s always outsourcing, but that’s not a one-size fits all solution (again, quality).

So what if we told you that you can automate those messages? That in 2016, we are at the stage where our machines can hold down a conversation – as long as they had good learning material (a database).

So, where we’re going with AI in customer service?

Sure, an AI won’t be able to know everything about your product and run your support on it’s own – but it can be taught to answer the most common questions, easing the load on your support staff.

What’s more, since it memorizes all of your previous chats, it can recommend answers to your agents: and that’s much faster than any human copy-pasting canned responses.

This also has the added benefit of training the AI to be more accurate in its responses in the future.

Of course how much you can benefit from this technology depends on quite a few factors – but typically we’ve seen cases that 10-20% of the conversations can be automated upon integration, and it’s feasible to automate 80% of responses in some niche-specific cases.

If you would like to learn more about chatbots please read our guide on chatbots.