Chatbots are getting a lot of attention in retail these days. At the NRF Big Show in January, several vendors, startups and established alike, featured chatbots in their booths at the show. They could do everything from interact with you about new products and what you liked about them, to help you figure out the opening hours of your local store.
You can order a pizza via a chatbot on Facebook Messenger, and several retailers are experimenting with chatbots around common customer service inquiries such as "What should I buy?" With a little rigging, you can set up Alexa to order your favorite drink for you at Starbucks. And yes, I count Alexa as a chatbot: a natural language interface, whether text or voice, and some artificial intelligence and a lot of computing power behind it to identify the most likely "best" response in an interaction.
I get it. I do. There are a lot of questions that consumers ask, like "What time does your store open?" that rudimentary natural language processing should be able to handle. It would satisfy both the customer and the retailer - the customer, who gets an answer quickly and simply, and the retailer, who doesn't have to have someone available to answer that same question over and over again day in and day out. Everyone wins.
But there are three things that retailers need to watch out for when using chatbots, or they will find themselves in dangerous territory.
1. Full Disclosure
Retailers need to let consumers know they are talking to a bot and not try to pass it off as a real human being. It's okay to personify it - Mall of America called theirs "ELF" this holiday season. But don't pretend to consumers that there is a human on the other end of the chat.
This is especially important if the conversation goes south. It's not really a coincidence that most of the chatbots implemented in 2016 were focused on service for the front end of the customer experience - for helping them select products versus finding their stuff. When customers contact retailers because they have a problem, they want to know that someone is listening to their problem and is there to help them. When they feel they're being ignored or misunderstood, they get frustrated.
User experience companies have a term for how that frustration translates when consumers are interacting with e-commerce sites: "rage clicking,"when a user clicks a button over and over and over again because they're mad that they're not getting the response they expected.
If a retailer passes off a chatbot as a real person, and the consumer figures out that it's not, it could easily lead to either the chatbot equivalent of rage clicking, or where consumers deliberately mess with the chatbot in order to get responses that could potentially be embarrassing to the brand
2. A Path To A Human
Just like the people who always hit 0 as soon as they run into the phone tree when calling, chatbots need to allow people to exit the conversation by being handed over to a person. The few demos that I've seen will turn users over to a human, but they manage it behind the scenes.
So, for example, when you ask the chatbot what time the store opens on Monday, and the chatbot replies 10 am, and you ask, "Even on Presidents Day?", the chatbot might not be able to answer that question. But rather than tell you "I don't know. Let me transfer you to someone who does," the bot designers try to be sneaky and preserve the illusion that you are talking to a real human being.
That's all fine and good when the options for using chatbots are actually few and far between. Consumers who use chatbots in a retail setting today are most likely opting in to the experiment. But as chatbots become more ubiquitous, retailers should never force consumers to use them even when they don't want to. And back to the rage-clicking result above, retailers should always give consumers the option to exit from a chatbot conversation into one with a real person if they decide they've had enough - even if the chatbot itself has not yet sensed that the person it's interacting with is getting frustrated.
3. Higher Trained Human Help
It's true that approximately 90% of customer service calls to a call center for a retailer focus on "Where's my stuff?" And the bulk of that could be answered easily by a chatbot. But that means when a customer does punch out of a chatbot conversation and ask to talk to a real person, it's because the basic information isn't cutting it. This means the last thing you want to do is then dump them into a conversation with an employee who is only trained and enabled to handle basic information.
So far, most of the content I've seen about chatbots has focused on making it easier for consumers to interact with machines - helping them get help when no human help is available. The industry has been fairly quiet about the impact on retail jobs. Right now, the impact is zero because most of this is experimentation and relatively small scale. But when bots take off in retail - and the money-saving implications are big enough that I have no doubt they will take off - what will most likely be forgotten is that the remaining jobs will need to become much higher value to compensate. That means a different hiring model, training model and pay scale.
Retailers have a tendency to overlook this trade-off. They also have a tendency to get themselves into a vicious cycle when it comes to technology vs. humans when interacting with customers: retailers implement self service technologies because they say customers hate interacting with employees. But the reason why customers hate it is not because they want self-service, it's because the live service is so poor, thanks to high turnover, no training and a lack of incentives. Consumers tend to want live customer service, but they want high quality service. When a retailer cuts labor in a store and replaces it with kiosks, then the employees who are left had better be more helpful than the kiosk. The same is true for chatbots.
4. The Bottom Line
Chatbots are here, and there will be a lot more of them in retail by the end of 2017. But retailers need to be careful about how they embrace them. They need to give consumers options, and they need to make sure that their labor model adapts to more bot-driven interactions, so consumers don't get frustrated.
Nikki Baird is a managing partner at Retail Systems Research, a market intelligence firm focused on trends in retail.