One day in July 2016, Steven Weinstein spent 10 hours and 43 minutes on the phone.
He wasn’t taking calls at his local PBS station during pledge breaks, but was helping a Zappos customer, where he works as a call centre agent. Weinstein took only one bathroom break during the marathon call, which set a new record at the online shoe retailer.
“After we had placed the order, the business part was over and then the connection just got even stronger,” Weinstein noted in a video. “We talked about everything from vacations to restaurants to places we’ve been to.”
Why did he stay on the phone for so long? The answer lies in the company's view on developing the customer experience. Zappos encourages its call centre agents to focus on building a rapport with customers rather than solely (yep, that’s a pun) selling them shoes.
Zappos, a division of Amazon, says this helps differentiate it from other e-tailers. It worked for Weinstein’s incredibly chatty customer, who told him she’s “never been treated like this by any company, anywhere, ever before.”
However, as unique as this particular interaction was, this type of customer experience could become increasingly rare as companies flock to chat bots.
Chat bots defined
In its 2016 report, The Disruptive Chat Bots, Deloitte defines a chat bot as “a program that mimics conversations with people using artificial intelligence techniques such as natural language processing (NLP), image and video processing, and audio analysis. Most importantly, these chat bots learn from their past interactions, improving their responses over time to more accurately accomplish their tasks.”
You’ve undoubtedly interacted with a chat bot by now — whether you knew it or not. And the more you interact with them, the more they 'learn' to perform their tasks.
Maybe one popped up when you visited a company’s website, then answered your questions in a live chat session. Or you may have had a brief phone conversation with one when you called a company for customer service. You might not have even realized the bot wasn't a real human.
Perhaps you used a chat bot with your smartphone, connected devices or a variety of apps. Bots are being integrated into software to help you with everyday tasks like texting, instant messaging and performing online searches, plus additional voice commands to get directions, make a list, listen to a song, watch a video or even to check your bank balance. Yes, bots are very busy these days.
The business case
Chat bots can be used in various ways by businesses. They can answer phone calls, respond to emails, send reminders and notifications, answer customer queries and follow-up on sales leads.
But to come to life in the first place, chat bots must be programmed correctly. Marketers design conversation scripts and framework scenarios for the chatbots to follow. Developers create platforms and language recognition so the bot can become intelligent. IT departments integrate the chat bots into existing software and apps so it is a seamless experience for the user. From concept to reality, deploying an effective bot is a big project — but the ROI can be impactful.
It’s easy to see some of the benefits of bots for businesses. They are often faster and more accurate than humans because they can search through several pools of data in mere seconds, meaning shorter customer wait times for help. Bots also work 24/7 and don’t require lunch breaks, vacations or sick days. Some bots can be programmed to work in various languages, and none of them ever get tired, bored, frustrated or rude with customers.
As Deloitte puts it, “chat bots are set to cause immediate disruption in customer support.”
Chat bots in action: Who is using them?
CNN and The Wall Street Journal use chat bots to deliver customized news content to readers. Passengers of Dutch airline KLM can get booking confirmations, flight status alerts, check-in notifications and boarding passes through a chat bot that communicates in nine languages. Closer to home, TD Bank began testing a Twitter direct messaging (DM) chat bot in June to help customers report lost or stolen debit cards and make other general inquiries. And if you're feeling hungry you can just place an order at Pizza Hut with its Facebook Messenger and Twitter DM chat bots.
Gartner predicts that by 2020, more than 85 per cent of customer interactions will require no human involvement at all, since chat bots have been adopted remarkably fast already. The proof is in the numbers:
- Deloitte says KLM’s chat bot logged more than one million customer interactions just half a year after its launch.
- Although there were no chat bots available on Facebook’s Messenger app in February 2016, just five months later there were more than 18,000.
- There are now over 20,000 chat bots available on Kik, the interactive messaging app built and based in Canada.
Is it a bot or not?
Some chat bots are so good at mimicking human communication they get mistaken for people. As a company called KnowledgeVision recounted to TechTarget, so many customers specifically asked to speak to its virtual assistant ‘Caitlyn Kelly’ that it had to set up a separate phone line and voicemail for ‘her.’ (The firm vetoed a suggestion, however, to give Caitlyn her own LinkedIn profile.)
But on the whole, those success stories are rare. According to Forrester’s 2016 report on The State of Chat Bots, “simple chat bots work well (within) a narrow set of potential responses and commands … but most chat bots disappoint.”
Chat bots disappoint us when they don’t understand sarcasm, humour or historical and cultural references. They also fall short when they use awkward and unnatural language, don’t handoff customers to human help as quickly as needed, or can’t handle complicated queries.
There can also be outright disasters. Less than 24 hours after Microsoft launched its Tay chat bot on Twitter last year, it had to be pulled offline because users ‘taught’ it to tweet racist and sexist phrases.
These tech fails sometimes still happen because a chat bot’s AI software needs a lot of time and data in order to learn and improve. That’s why most companies today have employees overseeing their chat bots, to soothe just one frustrated customer or nip a Tay-like catastrophe in the bud.
Eventually, chat bot technology could advance far enough to replace people in certain jobs. That will allow us, in many situations, to solve our problem or complete our task as efficiently as possible through an automated AI system. But it may lower our chances of meeting someone as patient and dedicated as Steven Weinstein in the course of a random phone call.