IS IT OKAY TO ASK IF YOU’RE TALKING TO A HUMAN OR A BOT DURING A CUSTOMER SERVICE INTERACTION?

Attributed to PV Kannan, Co-Founder and CEO, <24>7.ai

Recently, I read an article by Jon Mooallem (aka “Mr. Know-It-All) in WIRED about the polite way to ask if you’re talking to a bot or a human. It’s a tongue-in-cheek article, but it’s a legitimate question that I think deserves an answer. Personally, I have always advocated for bots identifying themselves, and I believe companies should be transparent about when you’re interacting with a bot. As these become more prevalent in interactions between consumers and brands, being transparent about bots becomes even more important in building trust.

It’s been 18 months since I published my first article on LinkedIn, about chatbots for business. Since that time, I’ve seen numerous articles on bots and a couple of things have become abundantly clear 1) There’s still a lot of confusion in the market about what bots are and what they can do, and 2) bots are not just a fad. They are here to stay, and they are transforming business. Bots vary greatly in intelligence and sophistication, and as businesses look to deploy them, it’s important to understand how they differ.

For customer service, bots must be transactional. These “intelligent chatbots” are able to integrate with enterprise systems, leverage big data, and use artificial intelligence to help customers resolve issues or conduct transactions and get things done. These are also referred to as “virtual agents” because they can help consumers conduct transactions as much as a human would. (The term does NOT refer to a human agent who works remotely). Virtual agents use sophisticated natural language processing and artificial intelligence to understand, anticipate and act on consumer intent. This is very much like the process that an intuitive human would use, and sometimes the results are a bit uncanny.

In the past year, we’ve been implementing a number of bot services with large enterprises, so I thought I would share some of our learnings: 

• From the consumer’s perspective - It’s becoming increasingly hard to tell if you're talking to a bot, however there are some ways to tell. For example, if the response is instantaneous, you’re likely talking to a bot. Human agents need a moment to think about their replies and type them out. The interesting twist though, is that some conversational AI software is now building in a delay to make it even more convincing that there’s a human on the other end. Again, we encourage companies to be transparent about this.
• From the agent’s perspective - Human agents are telling us that they’re getting asked more often “are you a bot?” AI technology has made human agents faster, more efficient and more consistent because the technology predicts and writes responses for the agent. Agents can customize responses of course, but the bot can do 90 percent of the response writing for them.
• From a large enterprise perspective – Our large clients have always put experience before cost savings, so NPS impact is closely watched when implementing bots. If the whole transaction gets done in less time than it would have with a live agent, it actually increases NPS. Also, when the context of the previous conversation is passed along, customers acknowledge great experience. Companies are concerned about user adoption especially when we are transparent it is a bot, however, what they’re finding with bots is that users are engaging very well and less than 5 percent ask for a human agent right away.

Creating “Near-Human” Experiences

Because of artificial intelligence, it’s now possible to analyze a company’s tremendous amount of customer data to create a personalized experience. This includes transaction history, preferences, shopping patterns, etc., and when combined with natural language processing, can be used to understand what customers mean—not just what they say. In other words, companies can use AI to determine consumer intent.

A large telecommunications company that I work with recently deployed intelligent virtual agents to personalize its customer interactions at scale. Using virtual agents, the company is increasing monthly chat volume of hundreds of thousands of chats per month and driving incremental sales by 10 percent in the first year. Projected savings are in the tens of millions annually, while CSAT runs at 85 percent or higher.

This is possible because AI analysis of their existing chat conversations (with human agents) was used to identify their customers’ top intents (reasons for making contact). This could include things like renewing a membership or making changes to their account. When virtual agents leverage intent, they become incredibly powerful and they behave more like a human agent would, anticipating customer needs and offering appropriate options.

In the End, Does it Matter if it’s a Bot?

Even when consumers know they’re talking to bots, they might prefer to interact with them instead of human agents. Often, if they’re just looking to do something routine, they can get results much faster than they would with a human. Our research has shown that when consumers can complete a task by interacting with a bot, there is no dip in NPS. In other words, as long as they can accomplish their business quickly and efficiently, they don’t care.

Of course there will be a dip in NPS when a bot has to hand it off to a human agent, but that is to be expected whenever a consumer has to be transferred to a different channel. Even if the bot can’t resolve the issue, however, it can hand over the conversation over to a live agent with the context of what’s already been discussed, and that dramatically improves the customer experience. The bottom line? It doesn’t matter whether the consumer is talking to a bot or a human, just as long as they can quickly and efficiently connect with a company to get stuff done.

With technology advancing so quickly, the day will soon come when consumers will find out they’re talking to a human agent and will ask “can you transfer me to the bot?”