Connected devices like Amazon’s Alexa are training us to interact with the Internet of Things to do things like play music and set the room temperature. Like mobile apps seven or so years ago, voice assistants and chatbots are the latest mobile gold rush.
Artificial Intelligence, machine learning and chatbots together make up an exciting yet nascent space. Even within six months of opening their platform to developers and brands alike, Facebook Messenger has 40,000 chatbots within it. Some chatbots are good, like Hi Poncho, Duolingo and CNN, while a vast majority are, well, not so good.
The space is young, but bot best practices are starting to emerge. Based on user usage and feedback, the best bots exhibit these core traits: purpose, learning and personality.
Brands looking to launch bots should remember to keep those traits in mind to avoid getting a meh reaction from customers
A bot is not a brand’s Web site or app. The most used bots have a singular purpose, which is solving a particular customer problem. For example, the Whole Foods bot’s single purpose is to provide recipes. It has no store locator, no daily specials; just recipes.
A brand can pick one problem where their customers would appreciate a virtual assistant always at the ready. For example, the Duolingo chatbots are great language tutors (in French, Spanish and German). The driver, chef and police officer chatbots help a person learn a new language their way through many practical real-world situations.
The exciting aspect about bots is that through artificial intelligence, the bots ‘learn’ and improve with more customer interactions.
The good news is that all brands know how talk to its customers already. A bot can be trained with customer care call logs, FAQ pages, social media feedback and common search queries. And just like humans who are constantly learning to get better, training is never over for a bot either. The Kayak bot, while great in many respects, does not yet understand ‘Thu night’ or ‘morning flight.’ But with the tenets of learning and improving, Kayak likely will get better.
The popular weather bot Hi Poncho manages to elevate a simple weather service into entertainment because of a quirky personality. It’s morning weather update never fails to make me smile and it makes me apologize when I’m rude to it. I could use many other weather sites and apps but it has become my slightly weird but lovable weather friend.
Brands looking at bots is all about test, learn and iterate. There’s quite a bit of thought on personality, purpose and continual training that goes into a rich chatbot experience.
It’s all about the bot’s raison d'être, training and personality