Last week, The Information reported that AI bots were experiencing a 70 percent failure rate, and the told-you-sos of the world started to point fingers. In the tests it conducted, the Silicon Valley blog said artificial intelligence-powered chatbots could successfully complete only 30 percent of requests without human assistance.

With the attention large technology companies are giving this medium, hordes of developers have been building a bot for this and a bot for that, targeting large audiences, creating inflated expectations and hope for big payoffs. Yet the implementations we have seen to date are underwhelming. In 2016, we saw unparalleled market consolidation and a general activity burst in the AI and chatbot space. You could argue it started in April 2016, when Mark Zuckerberg introduced Businesses on Messenger at F8 by saying: “I don’t know anyone who likes calling businesses.”

Why is there such a disconnect between the hype the industry has generated and the relatively slow uptake? (Of course, some form of disconnect is normal for every hype.) I believe we are dealing with a dilemma here, which lies in the nature of bots themselves and when they are truly useful.

Looking for chatbot value

Bots are about simplifying man-machine interaction. Conversation is the most natural form of communication for humans. Bots promise to make any engagement simpler: No more app or page downloads on a possibly spotty network just to satisfy a one-time need. No more account registrations with yet another password to memorize. Bots simply fit into the life you already have. Just open your favorite messaging app, search the business name (or scan their messaging code off an ad), start chatting, and you get the responses.

Marketing professionals are typically the first to embrace consumer innovation, as is the case with chatbots. Marketers want to bind you to their brand; they want to impress you with creativity and appeal. Yet a bot gives you a simple text interface or perhaps one enriched with some basic GUI elements like images or buttons. It’s hard to craft engaging content in a standardized messaging bubble that makes your customers want to come back to or engage with your bot. Furthermore, you give up control to the environment in which the bot lives, e.g. Facebook, or … well, mostly Facebook. Contrast that with your app or website: These are environments entirely under your control. We are giving that up when we move to messaging platforms.

As a consumer, when you’re in the market for a new product, you start with discovery. You browse and compare offers. But unfortunately, the amount of information you need to consume in this discovery phase of your customer journey does not easily fit into messaging bubbles. And that brings us to the bot dilemma.

The bot dilemma

Bots are great for the occasional, simple question: “I know what I want, give me the answer as quickly and conveniently as you can.” Yet when shopping around, consumers often don’t know exactly what they want. They want to browse for offers in a space with the freedom to explore, and a conversation is not the best way to go about that task. They want to be wowed, convinced, surprised; they want visual and experiential appeal from the company that they might do business with. Bots cannot easily give them this.

Furthermore, if consumers need something frequently, they often prefer an app, not a bot. The more often we make a customer engage with our brand, the more they tend to want a rich and satisfying UI coming with it, which a bot cannot provide. And adding new bells and whistles to messaging frameworks, such as carousels with buttons and images, will make them too similar to existing mobile solutions without providing the additional value we want from customer service bots: bypassing menus and getting right to the answer to a question on the customer’s mind.

So if we bring everything together, the dilemma becomes obvious:

  • Bot platforms like Facebook promise an audience of billions, which attracts marketers.
  • Due to the simplicity of their interface, bots are best for occasional, simple, directed questions.
  • An important aspect of the marketer’s job is to generate demand from new customers.
  • Marketers strive for a rich experience, and complete control over it, but a conversational interface alone makes for a somewhat poor experience.

Or, in a nutshell: Bots have attracted marketers, yet marketers don’t want to be where bots work best.

The solution: customer service

Now let’s revisit why Mark Zuckerberg launched his bots program to begin with: He wants to be the place where you have conversations not only with your friends, but with your businesses, too. When do conversations happen? Oftentimes after you made a purchase. That’s when “frequently asked questions” arise, not just before you buy. “Where is my order?” “What is your return policy?” “Do you also have item X in stock?” “What’s the status of my claim?” Everything starts with a question, and bots are a fantastic fit to handle these questions.

Customer service is the area where bot developers should flock, as bots have a clear value proposition in the contact center, from reducing cost through automation to improving the customer experience through 24/7 quick access to basic information. Both marketing and customer service are about 1:1 conversations, to serve and optimally increase wallet share of the customer.

Bot developers that focus on customer service automation will be able to assign a clear monetary value to their work versus those that build the nth weather bot or questionable utilities or games that work better in the environment of a native mobile app anyway. Bots excel where marketing and customer service meet.

Tobias Goebel is the Director of Emerging Technologies at Aspect Software, a call center improvement company.