New research conducted at Penn State reveals that customers do not like interacting with human-like bots, and that anthropomorphic features (human names, witticisms, personality cues, etc.) may actually exacerbate frustration on the customer end. The best customer service bots don’t need to be conversational, for example, just effective.
Per S. Shyam Sundar, James P. Jimirro Professor of Media Effects, co-director of the Media Effects Research Laboratory and affiliate of Penn State’s Institute for CyberScience (ICS):
“There’s a big push in the industry for chatbots. They’re low-cost and easy-to-use, which makes the technology attractive to companies for use in customer service, online tutoring and even cognitive therapy — but we also know that chatbots have limitations. For example, their conversation styles are often stilted and impersonal.”
“We see this again and again that, in general, high interactivity can compensate for the impersonal nature of low anthropomorphic visual cues. The bottom line is that people who design these things have to be very strategic about managing user expectations.”
The main takeaway of the study is that bots with human-like features but low interactivity (responses that do not match user’s queries, lag time between responses, etc.) result in lower satisfaction than bots with low interactivity and no human-like features. In other words, giving an ineffective bot human traits actually makes customers more frustrated than just interacting with an ineffective bot.
When bots have high anthropomorphism, users expect a human-like level of interactivity; this makes meeting user expectations a difficult task. However, when a bot with low anthropomorphism has high interactivity, users are pleasantly surprised: the bot exceeded their expectations for its ability to effectively interact with humans.
This research has massive implications for brands leveraging bots for customer service or any other consumer-facing interaction.
The Case Against Intelligent Bots
Natural language processing algorithms are not advanced enough today for truly smart bots. No matter how advanced Siri gets, iPhone users can still trick ‘her’ into saying nonsensical things. Likewise, no matter how advanced a customer service bot is, users will still encounter issues that the bot cannot understand, much less solve.
Knowing this, brands have a choice:
- Continue to give bots human qualities, knowing that there will be scenarios that bots will not ‘understand’
- Avoid using bots altogether
- Leverage bots for narrow and specific use cases without trying to emulate human intelligence
Clearly, the first option is a bad choice: customers end up even more frustrated than they were when they began. Many brands, however, have more difficulty deciding between the next two.
While avoiding bots altogether may seem like the path of least resistance, the reality is that brands that do not leverage automation for customer service are at a serious disadvantage from a cost and time perspective. Staffing live agents for every interaction is extremely costly, particularly considering that many customer service requests are repetitive; agents spend the majority of their time answering the same few issues over and over again.
This leaves the last option — to use a combination of automation and human intelligence to give customers and agents the best possible experience at scale.
The Best Customer Service Bots Don’t Have to be Intelligent; They Just Need to Work
This new research illuminates a simple truth: customers want bots to be efficient, not human. For brands, this is actually good news. The best customer service bots should be leveraged to collect information, answer frequently asked questions, and intelligently route issues to agents based on capacity and specialty. When it comes to the more complex issues, agents can step in and take over where the bot left off. By allowing issues to be resolved through bots alone or through a combination of bots and live agents, customers get the best of both worlds: instantaneous responses, rapid resolutions if the bot is able to solve the issue, and the option to speak with an agent if the bot is not equipped to understand or solve a query. This fluid handoff between bots and humans ensures that customers are guided down the most efficient route possible towards a resolution.
Ultimately, brands should be relieved that the best customer service bots do not need to live up to anthropomorphic standards. Customers understand that they’re interacting with a machine, and all they want from that machine is for it to be effective and efficient.