What Chatbot Experts Predicted and What They Actually Deliver Today
Early last year, a new headline gripped the tech world: chatbots, many claimed, were dead. Despite initially enthralling everyone from sales gurus to CIOs, it seems the technology hadn’t quite lived up to the big promises it made. Or maybe our expectations were simply a little too high.
As we move into 2019, attitudes towards chatbots remain mixed, with some companies still betting that they’ll ultimately prove their worth. But is this perspective justified? Here’s a look at what the experts said chatbots would do for us—and what they’ve actually done.
What They Claimed: Chatbots would replace human workers.
Early on, some painted a picture in which chatbots replaced human workers en masse. As recently as fall of 2018, some tech insiders continued to speculate about the potential of intelligent machines to eliminate jobs. For some, this was a point of concern; others saw it as a potential to save on labor costs.
So have chatbots changed the face of the labor market?
Overall, it’s complicated. Chatbots may have reduced the need for humans to work shifts outside of normal business hours, such as after-hours and weekend support. And companies may overall need fewer front-line employees to serve their customers.
But humans are still essential to many customer-facing functions. Like many groundbreaking technologies, chatbots’ true value has come not from replacing humans but from augmenting them. This is especially true for teams where quickly executing repetitive tasks is a key concern, such as customer support.
“By using bots to deal with lower-level inquiries, support teams can spend more time answering complex questions that are more valuable to the business,” explains Mike Murchison, CEO of Ada Support.
What the Future Holds
Some of the most exciting chatbot use cases, in fact, are the ones that intentionally factor in human input.
At Tenable, a leading cybersecurity company, bots play a crucial role in directing helpdesk questions to the correct experts. If the bot can’t answer an employee’s question with existing knowledge base articles, they can use the bot to contact a Subject Matter Expert.
“There’s an ‘Ask An Expert’ button that posts to a different Slack channel, and all of the channel members are SMEs who could potentially help with the question,” explains Bill Olson, a Product Manager who helped design this intelligent helpdesk.
“If two people ask for expert help, their questions will appear in that Slack channel. Another employee can go into that Slack channel, see the two questions, and say, ‘I know the answer to this one.’ They can hit the ‘Claim’ button, which will open a thread inside of Slack so they can have a conversation [with the person who asked the question].”
Other companies like Nutanix use chatbots to facilitate approval workflows such as approving the provisioning of virtual machines. By pulling these processes into a chat app—where today’s employees spend the majority of their time—bots can make it easier for human workers to accomplish everything they need to do more efficiently.
What They Claimed: Chatbots would create better experiences.
Perhaps the most exciting claim about chatbots was that they would totally redefine common experiences like returning an item or requesting time off of work. They’d offer instant, intelligent help to both employees and customers—and at a lower cost to businesses.
From the start, however, designing a useable chatbot interface proved challenging.
“There are technical and UX problems that limit the efficacy of a text-based, conversational UI,” says Dave Feldman, Vice President of Product Design at Heap.
Without sufficient AI to power things like Natural Language Processing—where users can talk to a chatbot the way they would talk to another human, instead of with rigid commands—it can be hard to feel like you’re having a high-quality experience with a chatbot.
Similarly, in order to actually improve experiences, chatbots have to alleviate some of the work that would otherwise fall to humans. It’s nifty if you can ask a chatbot to reschedule your flight, but if a human still has to input the request into a system or make the change manually, the chatbot is relatively useless.
Achieving this requires a fairly sophisticated degree of automation and integration, something that enterprises still struggle with. Only 16% of enterprises have deployed multiple automation use cases at scale, according to a Capgemini study.
This lack of automation may be to blame for the failure of high-profile chatbot projects like Facebook’s M.
“Facebook’s goal with M was to develop artificial-intelligence technology that could automate almost all of M’s tasks,” writes Alex Konrad of Forbes. “But despite Facebook’s vast engineering resources, M fell short: One source familiar with the program estimates M never surpassed 30% automation.”
What the Future Holds
Thankfully, this is one area where the technology is actually quite promising—especially as more automation platforms see bots as a fundamental part of their promise to help lines-of-business staff work more efficiently.
“To understand how bots and automation go hand-in-hand, you have to jump into the day-to-day lives of your target users,” says Ee Shan Sim, a product manager for automation platform Workato. “You have to ask, ‘Okay, if I were a sales manager, what functionality would I want?’ Or ‘As a project manager, which of my daily tasks could a bot make easier?’”
This process also involves understanding the way language impacts the user experience—especially for first-time chatbot users.
“[With chatbots], user inputs are required, but you want them to be intuitive. The challenge is finding a balance between how powerful a chatbot’s automations should be vs. how intimidating it is to the user to go through the workflow. [You have to] continually ask, ‘What makes sense for a first-time user? What’s going to look weird to them?’,” Sim continues.
“There’s a misconception that chatbots aren’t good enough to be customer-facing,” says Murchison. “In reality, customers are more likely to interface with a bot, because they know they’ll get an instant answer.”
What They Claimed: Everyone would love chatbots, and adoption would skyrocket.
Experts predicted that because of their potential to help cut costs and deliver top-notch experiences, they’d be the hottest new enterprise tech. In fact, a 2017 Deloitte report indicated that 67% of professionals expect chatbots would outperform mobile apps in the next five years.
As businesses have realized that (like any new technology) chatbots come with their own challenges, adoption has slowed. Many companies have looked to tech industry leaders like Facebook, who have given up on their chatbot initiatives, and followed suit or at least scaled down their chatbot efforts.
But others aren’t ready to abandon ship. It really depends on what line of work you’re in—for example, 95% of content management professionals surveyed said they still planned to adopt chatbots by 2019. Similarly, the banking sector continues to debut high-profile chatbot projects like Bank of America’s Erica, who managed to attract 1 million users in just three months.
What the Future Holds
As we move into 2019, chatbot adoption will probably continue to boom in some sectors and slow in others. Some experts believe that adoption ultimately boils down to how well you communicate the purpose of the bot to your prospective users, whether they’re customers or employees.
“[Bank of America] had email campaigns for some time saying Erica is coming; here’s what it is [and what it can do],” says Emmett Higdon, director of digital banking at Javelin Strategy & Research. “They did a good job prepping the audience for its introduction.”
Experts also agree that successful projects like Erica have high user growth because of how well-integrated they are with other services, like content libraries, search tools, and AI services. So as the automation and cognitive technologies surrounding chatbots improve, adoption will, too, as long as companies can keep pace with user demands.
That’s one thing experts broadly agree on: whether a chatbot is customer-facing or internal, its long-term success depends on how well it can anticipate what users want—and then go above and beyond their expectations.
For example, Higdon imagines a scenario where a customer asks Erica how much they spent on Uber last month. To really improve adoption and retention, the bot can’t just name a dollar amount.
“[It should be able to say] ‘By the way, that’s twice as much as you’ve spent in the last three months, is there something wrong here?’ Something that gets the customer to go ‘Hmmm’ and think more about their financial health overall,” he says.
Moving Forward With Chatbots: Patience Is the Answer
Overall, chatbots may have evolved differently than we expected. They haven’t turned into a revolutionary tool as quickly as many thought they would, leaving businesses disenchanted and disappointed with marginal improvements.
But as Intercom CEO Eoghan McCabe points out, this lack of buzz around chatbots is par for the course when it comes to emerging technologies.
“I don’t think there’s ever been a new technology that hasn’t followed that cycle,” he comments. “We insiders who get so excited about the future will always jump on the hype and excitement ahead of its practical reality. Virtual reality, self-driving cars—[all of these] technologies will get less sexy before they get real.”