AI and the ‘Frictionless’ Customer Service Experience

This McMinute is ‘two-McDonalds-for-one,’ co-authored  by Susan and her daughter, Nora, who teaches and conducts research on human-computer interaction (HCI) at George Mason University.

Well-oiled Online Machinery

Ever since the dawn of the digital age, which has brought us to a state of abject dependence on devices, designers have worked ceaselessly to remove what HCI professionals refer to as sources of friction: design features and steps that slow, complicate or thwart digital engagement. Of course, some friction must be strategically added to the system to promote safe use, especially the “are-you-sure?” backstops meant to prevent us from deleting our data or signing away our privacy. But overall, the goal of the system is to enable smooth and speedy digital interactions. AI is adding fresh lubricant to the process by facilitating and consolidating searches, delivering us rapidly and efficiently to relevant websites, and unleashing trouble-shooting agentic bots that can take the work out of getting what you want online. The path to digital fulfillment is now pretty slick.

It’s true that some companies deliberately plant “sludge” in the system, making it difficult for customers to return items or extricate themselves from the clutches of unscrupulous sellers. But for the most part, companies have been the ones driving toward an AI-enabled online utopia that aims to make things easy and automatic for everyone—a ‘win-win’ meant to save money for businesses and wait time for consumers.

Those of us with a long history of working on customer experience strategy will recall that in the last few decades of the 20th century, consumers were identifying long phone waits as one of their major beefs. Early advances in customer service were based on modeling how much satisfaction ratings might change as a direct result of reducing a five-minute phone wait to two, ideally with the creation of a self-service interface that also increased profitability by reducing the number of phone reps. Before automation and ultimately AI, that was the principal source of friction.

How’s it Going for Everyone?

Although navigating multi-tiered service menus can still be time-consuming, phone waits have, in fact, been significantly shortened by automating delivery of information about things like account balance or power outages and shifting to online appointment-making or pick-up. And if we really need a human in the loop – and we are able to reach that queue – automated callback options can spare us the torment of synthetic hold music and advertising captivity.

But if you ask consumers how they feel about customer service today in a world rapidly being reshaped by AI, the feelings you hear are both varied and equivocal. Automated interactions certainly create some efficiencies, but those efficiencies also introduce new opportunities for friction when the feature actively malfunctions or snares you in a loop of automated troubleshooting from which escape seems impossible. A gratuitous option meant to achieve efficiency is experienced by customers as a time-consuming irritant. And the proliferation of more complex and customized service interactions begets new sources of friction in an ever more complex machine. A truly frictionless system is difficult to construct, in science and in society. And ironically, failed automation can become more stressful than the problem a customer was trying to solve in the first place.

Humans to the Rescue

It turns out that humans are still better at solving complicated problems than AI agents, which makes the friction of having to interact with a stranger on the phone the smoothest path to certain kinds of resolution. But there is yet another critical factor in all this: human nature. It turns out that in systems involving humans and computers, removal of all human contact also removes sources of pleasure that a stickier sort of life may offer. Like talking to other people. There is an inherent tension between AI-enabled automation and the pursuit of a frictionless, human-less, customer-satisfying service model.

What matters in service and problem resolution is not just what we feel about the outcome but how we feel about the process—the how and the who of it. We are bio-evolved to feel gratitude and self-worth when we are helped by humans, even people we don’t know. Sometimes, especially by people we don’t know. Getting help from another human being triggers all kinds of chemical rewards from oxytocin release to dopamine hits. And being helped by a caring stranger, even one whose job is to help us, can still create strong positive feelings about ourselves and our society. For an interesting perspective based on a “moral psychology” perspective, see Haidt, J. (2000). “The Positive Emotion of Elevation.” Prevention & Treatment, 3(1).

Among customer service experts, it is canon that having a problem solved graciously elevates satisfaction above having experienced no problem at all. Sometimes, a little friction in the system feels good—and it burns less than the anxiety of trying to figure out how to do things without the help of another human being. There is an argument to be made that on some days, having even simple service rendered by another human being, feels better than DIY with the aid of an automated menu.

‘Man v Machine’: Achieving Harmony, Maintaining Balance

In a world of AI enablement, optimizing digital performance is no longer the technical challenge it was 10-15 years ago. The online service infrastructure can be made to work very well, in ways that would have defied imagination even a decade ago. The extraordinary capabilities of AI are also inviting a subtle shift in the field of HCI scholarship away from the data science, software-centric question, “How do we program digital services so that they function well for humans?”  to a slightly different question, rooted in the social sciences:  “How do we help humans function well in a system that presses them to offload on computers so many of the functions that once helped define their human-ness? It’s a profound existential challenge, and our service interactions with computers are a critical component.

The new frontier of AI-governed service automation has not yet been given much focused attention in published studies because it’s still early days and there’s much about AI to study. Well-controlled, truly realistic studies on how people prefer to get service are tough to conduct. NAXION has been conducting our own research on how consumers experience customer service in a post-AI world. To add an ironic twist, we’ve also used this initiative to compare how humans and synthetics view the impact of AI on customer service. We’ll have more to say about the results of our research in the coming weeks, including how different generations feel about AI in an online service environment.