February 6, 2018

Dawn of the Chatbots: What Do Consumers Want and Expect?

Written by
Mark Newton

Unless you’ve been living under a rock (or in a wifi deadzone) for the past few months, you’ll no doubt realize 2018 will be a year of massive technological hype. Whether it’s artificial intelligence, blockchain, augmented reality, or chatbots, you’ll likely have seen endless articles about how each will fundamentally revolutionize the industry in which they’ll be used – which, you’ve also learned – is practically every industry out there.

Of these well-hyped technologies, chatbots are probably the most successful in terms of their common use today. While AI is still in its relative infancy and blockchain continues to confuse anyone without a computer science degree, chatbots provide a practical, identifiable service that builds upon existing expectations and infrastructure.

Major companies and corporations such as Unilever, CNN and Facebook have already incorporated chatbots across a range of roles, including customer service, marketing, and content distribution. The advantages to these early adopters are clear, they can provide 24/7 on-point customer interaction – for a fraction of the cost of human operators. For example, a Tableau report stated that 29 percent of customer service jobs could be automated, saving $23 billion annually in the process.

However, regardless of the hype, do consumers actually want to interact with chatbots and AI? To quote the great Dr Ian Malcolm (of Jurassic Park fame) have developers become so “preoccupied with whether or not they could that they didn’t stop to think if they should”?

Do Consumers Want Chatbot Customer Service Assistants?

There are several studies online that have attempted to gauge consumer attitudes to chatbots, although bring a hefty pinch of salt, as usually the companies that actually manufacture bots also commission and create the studies. However, regardless of any flowery conclusions they might make, the pure empirics are not overwhelmingly in favor of chatbots.

For example, a study by chatbot developer Ubisend revealed that only 35 percent of the 2,000 respondents surveyed wanted to see more companies utilize the technology. Although, admittedly, only 25 percent of those surveyed knowingly used a chatbot.

Other surveys produce similarly low results: a Code report revealed only 14.4 percent of British respondents wanted to use chatbots. The most optimistic figure I could find regarding attitude towards chatbots came in an Aspect Software Research study, which cited 44 percent of US respondents would prefer to use a chatbot, but only if the company could get the experience right.

The Ubisend study also reveals some interesting statistics in regards to how consumers would like to use chatbots. The majority of consumers prioritize getting speedy information that solves their immediate requirements, such as finding out the opening times of a store or business. In this sense, it seems most consumers are using chatbots in an identical way to traditional search engines. This potentially makes the intelligent chatty element of the chatbot somewhat redundant.

This coincides with findings from the Code report. When asked what consumers wanted from digital tech, only 30.8 percent wanted to see innovative, unique tech, while 43.9 percent (the most popular response) merely wanted to see improvements to pre-existing services they regularly use like customer service, helpdesk, and other software they interact with like customer loyalty gamification software.

Despite this, a noHold study found that although consumers asked direct questions, they did so in a more conversational human way, perhaps showing customers warming to the idea of communicating with an intelligent chatbot. The study found that, on average, chatbot users used 5 words to frame a question, rather than the 2 words generally used with a search engine.

Are Shy Chatbots Actually Better?

The chatty elements of advanced chatbots could be more of a liability than a benefit to image conscious businesses. After all, having a computer program communicate for your company, especially a high tech one that learns from interacting with customers, could result in a PR disaster. One only has to cast their minds back to Microsoft’s 2016 AI experiment, Tay. Within 24 hours, a chatbot designed to mimic a fun, pop culture loving teenage girl was turned into an internet troll (warning: the story linked here has some potentially offensive language).

Companies will be eager to avoid this kind of event. Oddly, human employees with bills to pay could be, perhaps counterintuitively, easier to keep on-message than a fully fledged ‘learning’ AI.

During their six-month experiment with chatbots, CNN came to similar conclusions. They used various chatbots across different channels and services to suggest news stories to followers. The experiment on the whole was a success, and CNN plans to expand the program, but they also realized the need to tailor each chatbot to a particular audience and fine tune its output. Talking to The Drum, CNN’s senior vice president and chief product officer, Alex Wellen, explained:

“[T]hey are putting CNN right alongside the people that matter to them most — the family, friends, and colleagues they message each day — and we want to earn that trust. We want the conversation with CNN to feel personal and non-intrusive… In other words, the chatbot can’t be too chatty.”

Indeed, 47 percent of those who responded to a LivePerson survey advocated skipping the conversational element altogether.

Conclusion

Much of the negative response to chatbots encountered in the above surveys can likely be attributed to teething issues and lack of interaction. Having said that, the responses also reveal some interesting ways in which consumers want and expect to interact with chatbots. Most clearly, they want technology to be a tool – providing them with a service and immediately answering requests – before it becomes a gregarious pal. A chatty bot that takes ages to get to the point is perhaps the antithesis of what consumers actually want.

With this in mind, chatbots will likely be best used in a basic customer service role, providing tangible answers to direct questions. For example, a chatbot attached to a job board or applicant tracking system could provide information on salary expectations for the role, or Glassdoor reviews of the company in question, thereby assisting the candidate meaningfully. Used in this way, chatbots could be a quick and easy way to customer or candidate satisfaction.

So, companies would do well to temper their enthusiasm for a cheap customer service representative. Clearly, consumers do not want to meet a gaggle of affable AI chatbots anytime soon. In fact, 88% of the respondents to the Aspect Software Research survey expect a chatbot conversation to eventually lead to a human one, somewhat negating the savings to be made in replacing human customer service representatives with machines.

Ultimately, it’s also important that the technology community does not become too self-centered and congratulatory regarding its advances. It needs to keep in touch with the end consumers and deliver products they actually want. Although much of the tech community might push for smarter bots and deride more basic examples as nothing but a glorified phone tree, it seems that when it comes to customer services, the customer – for the time being at least – is more in favor of the latter than the former.

Mark Newton is a Senior Editorial Manager for SmartRecruiters.com in Berlin, Germany. In addition to writing about the HR and tech worlds, he also contributes freelance articles concerning technological advances in the environmental and conservation NGO sector.