한국어
한국어
Columns

Why do social robots fail? (2/2)

Part 2. Balancing expectations and cost, and the weight of the word “robot”

By Seungbin Jeong · · Updated

In Part 1, I traced how robots whose features and marketing raised expectations went on to take different paths. Some kept selling at relatively low prices even after reviewers judged them to fall short, and one moved away from the home market and shifted its business to institutions. Meanwhile, some home robots sold at high prices eventually had their services shut down.

Balancing expectations and cost

What this comparison drew my attention to is the relationship between users’ expectations, their actual experience and the purchase cost. By cost, I mean the amount an individual pays out of pocket to buy the product. Students and teachers who use Buddy at school, for instance, use a product an institution bought, so they do not bear the purchase cost themselves.

The idea that the gap between expectations before purchase and the actual experience affects satisfaction is not new. In consumer behavior research, the expectation-disconfirmation theory Richard Oliver proposed in 1980 has explained this relationship. Looking at two further conditions, price and who buys, offers a clue to why the robots in Part 1 took different paths.

Even with the same gap between expectation and experience, users may react differently depending on how much they paid out of pocket. Someone who paid little may accept some limitations, while someone who spent a lot may find the same limitations far more disappointing. This is a hypothesis the case comparison suggests.

I believe three things should be considered together from the earliest stages of product development: what expectations users will form about the product, who pays and how much, and what experience the product can actually deliver. Packing in more features than needed and inflating the marketing can raise users’ expectations. A high price calls for an experience that convinces users it was worth the cost. If the target is institutional buyers, it is not enough to look at why the institution would pay for it; the benefit to the people who use the product on site matters too.

LOVOT is a case of deliberately setting the direction of expectations. At its 2019 launch, the robot alone cost ¥349,000, with a separate monthly fee on top. Even so, according to its maker, more than 18,000 units were active as of January 2026. As noted in Part 1, GROOVE X did not present LOVOT as a robot that works in place of people. Rather than housework or assistant duties, it focused on the experience of living together and forming an attachment. Buyers come to expect emotional connection rather than the ability to perform tasks. In the end, whether that justifies the high price depends on how satisfying and lasting the experience turns out to be.

The weight of the word “robot”

Applying expectation-disconfirmation theory to social robots suggests a hypothesis: the very idea of a robot may raise users’ expectations and widen the gap with their actual experience.

Performance that would have been good enough for a speaker or tablet with the same features can become a reason for disappointment once the product is called a robot. When people see something that talks with them and moves on its own, they may expect it to understand the situation and respond flexibly.

The cases hint at this possibility. Companies add features because they think a robot should be able to do at least that much, and users judge what the robot can do from those features. Once something is called a robot, the comparison extends to existing products that already perform those functions well. As we saw in Part 1, Jibo was compared with smart speakers, and Pepper with in-store tablet kiosks and staff. Miko, too, was described as closer to a tablet dressed as a robot.

Existing research offers clues for examining this hypothesis. In 2019, Horstmann and Krämer at the University of Duisburg-Essen in Germany surveyed 433 people and interviewed 13. Few had encountered a real social robot, and indirect experience through films and other media appeared to shape their expectations of robots.

In a 2016 study, Kwon and colleagues at Cornell University addressed the problem that people may expect capabilities beyond what a robot actually has, based on its appearance. The researchers described the gap between expectations of a robot and its actual capabilities as the “expectations gap.”

These studies do not directly show that the label “robot” itself creates higher expectations than ordinary products do. I have also yet to find a study comparing expectations of robots and ordinary products under the same conditions. I leave this as a question for future work.

The robots that have been sold or used over relatively long periods had clear roles for users to expect. LOVOT and NICOBO focused on emotional connection rather than doing people’s work for them, and Pepper was used in settings with a clear purpose, such as classes and research.

If users can see clearly what a robot offers them, couldn’t that also temper the vague expectations the word “robot” creates?

Where design comes in

Marketing strategy and pricing are usually decided by the marketing department and executives. But what role a robot takes on, which features to include or leave out for that role, and what form and behavior it uses to engage users are areas where designers can be deeply involved. The Kwon et al. study mentioned above also shows that a robot’s appearance can affect what users expect it to be capable of.

I see the form of Qoobo, with no face and no legs, only a tail, as a design decision that sets the range of users’ expectations. Instead of a robot that looks as if it could do anything, its makers built a product that connects through limited movements and responses. The form itself tells users what range of behavior to expect.

Imagining what a family that paid $899 to bring Jibo home would ask it during the first week, and what answers they would get, is also the designer’s job. Designers need to picture concretely, and then test, which features people would keep using a few days after first talking to it out of curiosity, and at what moments they would run into the robot’s limits.

Examining these moments of use from the early stages of development makes it more likely that a team will spot in advance where the expectations created by features and marketing will turn into disappointment. Sharing those findings with engineers, marketers and decision-makers allows a more realistic discussion of which features to build and what experience to offer.

Deciding a robot’s features does not end with choosing the technologies that can be implemented. It also means considering what users will expect when they see the robot, what they will actually experience, and how much they would be willing to pay for that experience.

I believe that connecting users’ expectations, their actual experience and cost is one of the important roles industrial designers should take on in social robot design.

Summary

  • To improve the commercial success and sustainability of social robots, users’ expectations and actual experience should be considered together with who buys the robot and what it costs.
  • The name “robot” and a robot’s appearance may raise users’ expectations and widen the gap with actual performance.
  • Stating a clear role and purpose can help narrow the gap between expectations and actual experience.
  • In designing a robot’s features, form and behavior, industrial designers can anticipate users’ expectations and test, from the earliest stages of development, where those expectations will diverge from actual use.

Scope of this analysis

This essay is based on an exploratory analysis comparing public sources on the 32 social robots documented in the archive. I distinguished between discontinuation, withdrawal from certain markets, manufacturer shutdown and current availability, and I did not judge the gap between expectation and actual experience for the six robots for which I could not find independent reviews or research. The relationship between price and business continuity discussed here is a trend observed in the cases, not proven causation. Nor can continued sales alone establish profitability or commercial success.

Research cited