Published in Hypertext, 12th June 2020
By Brendyn Lotz
One a scale of one to 10, with 10 being most annoyed, how annoyed are you when you discover you are chatting to a bot and why is it 10?
Joking aside, chatbots can be incredibly useful especially if your company often has to deal with frequently asked questions or can leverage artificial intelligence to deal with customer queries.
But building a great chatbot is not as simple as creating canned responses and pushing it live. We’ve seen some truly incredible chatbots such as rAInbow, a chatbot designed to be a companion for victims of domestic violence.
And while chatbots are not all that new, chief executive officer at CLEVVA, Ryan Falkenburg (pictured) says that when done right, chatbots can be a valuable addition to a business’ customer service offering.
The trick in getting your chatbot to be more helpful and less annoying is to insist that certain features be built into the application.
“Used effectively, chatbots can be a valuable addition to any customer service offering. All too often, however, chatbots have limited functionality, leaving customers frustrated and desperate to talk to an actual human,” says Falkenburg.
So what are the features you should insist be built into your chatbot?
While humans may not mind talking to a machine, Falkenburg says that the trick is to make it feel as if they are talking to a human. Part of this is accomplished through something called natural language processing.
This concept is rather complex but to explain it in the simplest terms possible, natural language processing is where an AI is able to interpret what you say with a high level of accuracy and decipher your intent.
For instance, you could say “I ordered 12 eggs and only received six” the chatbot should be ready to offer assistance in the form of offering to rectify the order or process a refund.
We’ve simplified this greatly here and we recommend reading a bit more about natural language processing. This Medium post from Dr Michael Garbade is a good jumping off point.
Expert advice not just assistance
Now that you have your chatbot able to understand what customers want it’s time to decide what sort of interactions you want customers to be able to have with the bot.
As Falkenburg explains, chatbots are great question and answer machines but to leverage real value out of the tech you should insist that the chatbot can dispense expert advice as well.
“Advice is very different from assistance. I go to my pharmacist for assistance. I go to my doctor for advice. Increasingly customers that engage with chatbots are looking for expert advice and very few are getting it,” says Falkenburg.
This is a fine line to tread however. For instance, a law firm may not want a chatbot to dispense legal advice for myriad reasons but it may help if the chatbot is able to direct the person to the sort of legal professional who could assist them.
By now you’ve likely encountered a situation where a chatbot directs you to a useless piece of info you’ve likely already seen or tried on your own.
Worse, the chatbot may ask you to complete a customer service form which puts you back at square one.
The CLEVVA CEO says that chatbots which can diagnose a problem and then solve it hold massive value. The trick however is that they need integration with back office systems.
Say for example a customer needs to cancel an order or reschedule a delivery. A great chatbot would be able to facilitate that seamlessly action refunds or changes without the need for a human to step in.
Finally, and perhaps most importantly, when a customer is transferred to a human, make sure they don’t have to repeat themselves.
“You should insist that your chatbot is able to work in sync with other channels, and can hand over the customer journey to others so they can continue, in context, from where the chatbot left off,” says Falkenburg.
This should honestly be a consideration across your entire business and not just for chatbots.
To sum up then, your chatbot should function as an expert within your organisation and not simply a piece of code running in the background.
“Be more demanding when it comes to your chatbot provider,” says Falkenburg. “There really are no more excuses for not being able to deliver these four basic chatbot requirements.”
To view the original article, click here