Why We May Be Thinking About Chatbots All Wrong

Last Sunday, I wanted to order a pizza. I could’ve called my local Pizza Hut or used the company’s website, but I decided to try Facebook Messenger instead. I already had three conversations going, so why not add one more chat window?

My customer account for Pizza Hut linked to Messenger in seconds and didn’t require any new payment information. One large pepperoni, please. Same as last time? Yes, please. But instead of dealing with a stranger or another checkout screen, I was typing to a chatbot instead. No small talk. No dropped calls. The bot offered me a few promo items, but otherwise, the whole exchange was quick, easy, and largely indistinguishable from a human interaction. Is this still my address? Yep. The receipt came via email.

The company credited with selling the first item on the internet is now making the e-commerce even easier. Pizza Hut recently announced its new chatbot ordering feature as part of a massive social media rollout that debuted a few months ago. Conversational commerce is hotter than a Samsung battery right now, and the largest pizza chain in the world knows it, following competitors and contemporaries alike into a space still very much evolving.

But what if businesses are approaching chatbots all wrong?

What is the secret sauce?

Magnus Jern, president of mobile solutions company DMI, recently told the BBC that when chatbots try too hard to be natural, it diverts from the purpose of conversational commerce. Jern helped launch IKEA’s Anna chatbot in 2005, which was recently retired after 10 years. “In the beginning, we tried to impersonate a person, and we found that there was no reason to do that,” he said.

But the move is a curious one, especially when chatbots are on the rise. Earlier this year, KPCB’s Mary Meeker referred to them as the “secret sauce” of messaging in her keynote on digital trends.

However, academic research has suggested that consumers don’t want robots that can talk like humans. Some would argue, instead, that all we want is a smoother ordering process. A Harvard Business Review report from 2010 found that “loyalty has a lot more to do with how well companies deliver on their basic, even plain-vanilla promises than on how dazzling the service experience might be.”

In his book, Influence, Dr. Robert Cialdini, a professor who teaches psychology and marketing at Arizona State University, concluded that we’re more motivated (to act, purchase, click, etc.) when choice is limited.

Also, there’s a difference between talking to a human and a bot that technology may never be able to reconcile. According to a study published by JAMA, conversational agents like Siri or Google Now simply don’t understand the difference between “I’m dying” and “I’m dying of hunger” in a crisis.

In other words, ordering a pizza via chat isn’t so unique anymore, but predicting how that conversation might look in the future is a bit more challenging.

The commerce of chatbots

The term “chatterbox” was coined in 1994, the same year Pizza Hut filled its first online order. Today, the company is the largest pizza chain in the world—with roughly half of its orders coming through digital channels and more than 60 percent of those via mobile devices, per internal data.

“We all have to become students of human behavior,” Baron Concors, Pizza Hut’s global chief digital officer, told a MobileBeat audience in June.

Conversable, the Austin-based company behind Pizza Hut’s chatbot technology, is turning the study of human behavior into a thriving business. The software company is partnering rapidly with major brands like TGI Fridays and Whole Foods, using conversational messaging for self-service and on-demand content.

For some companies, customer service is one long conversation. For others, the conversation ends with a pizza delivery.

The chatbot ecosystem is exploding. Facebook now supports over 11,000 chatbots, plus a dedicated store. Apple recently debuted its own iMessage app store with iOS 10. And messaging apps, well suited for brand chatbots, have never been more popular. WhatsApp, for example, now has over 1 billion users. WeChat and Viber have hundreds of millions.

In June, Tommy Hilfiger announced its own chatbot designed with the help of Facebook’s Creative Shop and bot creator Msg.ai. According to TechCrunch, the social giant caught flack for hosting too many clumsy bots from outside developers. The partnership with Tommy Hilfiger lets the company reclaim some control over its new chatbot platform, while heeding the call for online concierges among high-end fashion brands.

As Tommy Hilfiger himself told TechCrunch, “We are obviously distributed in our own stores and in department stores, but going directly to the consumer is really part of the motive and the future of the omni-channel process.” Gigi, named after supermodel Gigi Hadid, will answer customers in a more natural style since, as CMO Avery Baker argued, no one wants to feel like they’re talking to the corporate animal anyway.

Calls for conversation

For some companies, customer service is one long conversation. For others, the conversation ends with a pizza delivery.

“Even the most digitally tuned-in customer will want to know that they are connected with someone who can put themselves in their shoes,” said Simon Hunt, director of customer experience at Firstsource Solutions, a business-process outsourcing firm based out of India.

In August, the travel app Skyscanner estimated a layover of 413,768 hours to a shocked consumer looking for a cheap flight. When the man posted about the error on Facebook, a Skyscanner rep responded with a clever and lighthearted comment that eventually went viral and generated press coverage. To clarify, the technology screwed up, and then a human came in to clear things up.

“Bots are easy. Conversations are hard.”

Conversocial’s CMO Paul Johns told Digiday that such unscripted rapport is a growing trend. Proving a resolution is great, but opening up a meaningful dialogue may be even better. Per CeBit, 71 percent of people who receive a quick response from a brand on social media are likely to recommend that brand to others. It’s no surprise that chatbots are being considered to automate the job.

But despite these developments, there’s still a weird tension surrounding the conversational commerce movement. People want quick, straightforward service, but they also seem to value human empathy. Is it possible for a chatbot to provide both, even if consumers know they’re talking to an algorithm?

Conversations are hard

Ben Lamm, the CEO and co-founder of Conversable, probably said it best: “Bots are easy. Conversations are hard.”

In the race for creating tech with personality, businesses seem stuck on naming their bots after anything other than a tool. Anna. Gigi. Facebook even has a bartender bot named Shaky.

But chatbots don’t care what we call them, and, let’s face it, expressing our trust in AI impacts my comfort, not their effectiveness. That’s why we’re still anthropomorphizing machines.

“Giving something a human name is a way of exerting control over it,” writes Adrienne LaFrance in The Atlantic, “a reminder that it works for you, that it exists within a human construct, even when the machine itself is wholly indifferent.”

It’s hard to pinpoint how much conversation and functionality is necessary, but it’s safe to say that functionality is ultimately what will drive revenue.

Chris Messina, the developer who coined the term “conversational commerce” about two years ago (and also came up with the word “hashtag”), gave a talk at the MobileBeat conference where Pizza Hut made its bot announcement this summer. Messina showed the audience his own bot, an integrated messaging platform, and went over the evolution of the bot movement.

Toward the end of his presentation, Messina went over a few rules for both ethics. “A bot should be able to describe itself,” he said. “What it does, how it handles information, if there’s a human on the other end monitoring stuff. Bots should have a similar type of disclosure statement.”

That’s all well and good, but for the most part, the only rule I care about is if the bot can get my order right the next time I want a pizza.