This fear of rejection in the online setting is a very real factor that designers in a two-sided matching market need to take under consideration. • Emily Young/UCI

Are You Too Popular? Unexpected Insights from Online Dating

June 17, 2021 • By Keith Giles

Market psychology can be the key to success when designing an online platform. As Professor Behnaz Bojd of The UCI Paul Merage School of Business puts it, “We need to recognize the differences between one-sided and two-sided matching markets. For example, e-commerce is a one-sided market, whereas labor markets are two-sided. Employers are looking to attract top talent and jobseekers are looking for the best companies. Both sides of the table need to agree to choose one another for the relationship to be successful.”

Together with Professor Hema Yoganarasimhan of the Foster School of Business at the University of Washington, Bojd set out to uncover some of the psychology behind two-sided marketplaces. “We wanted to examine the effect of a user’s popularity rating on their demand in a mobile dating platform,” says Bojd. They have published their results in an article, “Star-Cursed Lovers: Role of Popularity Information in Online Dating,” which is forthcoming from Marketing Science.

Market Psychology

As we’ve established, e-commerce represents a one-sided market. Customers need a product, and they shop around until they find what they’re looking for. But the providers of goods and services do not need to evaluate their customers in advance. If a customer has a valid credit card, the transaction is approved.

Two-sided markets—like dating websites, college admissions departments and job search sites—present challenges that are not present in one-sided markets. “There are more than 2,000 dating apps out there now that generate over $3 billion a year in revenue,” says Bojd. “Designing information systems that factor in the unique qualities of two-sided markets isn’t as easy as cutting and pasting features we find on e-commerce sites. We need to take the time to understand the psychology of users in a two-sided market if we hope to avoid unintended consequences.”

The Games We Play

In their study, Bojd and Yoganarasimhan examined a popular dating app and a game within the app where users could rank a set of four potential candidates based on limited information. As Bojd explains, the game works like this: “There are four women and four men in a virtual room and each person is asked to rank the opposite gender from one-to-four in only 90 seconds. They see only limited info about each person—their picture, name, age, location and popularity rating—and matches are calculated using the tried-and-true stable matching algorithm,” she says. “If I’m matched with Person A, I can decide if I want to send a message to them, and they can reply to my message or not.”

Surprising Results

Knowing that a potential partner is popular should increase their appeal. Yet, the more popular someone was in the game, the less likely it was that another person might choose them.

As the game was played, volunteers rank-ordered members of the opposite sex and are then matched based on a Stable Matching Algorithm. After the matches were selected, users were allowed to message and chat with their match after the game.  “We quantified the causal effect of a user’s popularity (or star-rating) on the rankings received during the game and the likelihood of receiving messages after the game,” she says. “ Surprisingly, we found that popular users received worse rankings during the game but received more messages after the game.”

In other words, the results differed depending on the popularity ratings assigned to each person. “People were avoided more if they had 3 stars versus if they had a 2- or 1-star rating,” says Bojd.

Factoring in the Fear

The results appear to go against common sense in some ways, with outcomes that often are the opposite of what one might assume. Why is this? “We link these results to the possibility of rejection that users may have feared,” says Bojd. “This fear seems to have led to strategic behaviors to minimize potential rejection.”

People are used to referring to star ratings for products online. These ratings help them decide whether to purchase are not. The ratings systems on certain online dating platforms may be having the opposite effect. “Sometimes people like to approach popular people and other times they avoid it,” Bojd explains. “Popular people are physically attractive and have other desirable characteristics, but there are other factors in play. It’s not that some people don’t like attractive partners. Mostly it’s the fear of rejection that causes some to lower or strategically shade their expectations.”

Counting the Cost

This fear of rejection in the online setting is a very real factor that designers in a two-sided matching market need to take under consideration. “Being online doesn’t seem to change the dynamic of approaching an attractive person,” she says. “The fear of rejection is still very much connected to the amount of mental and emotional energy that people invest. It’s very costly to approach someone you are likely to fail to attract.”

Of course, people who are already attractive or have more self-confidence don’t struggle with these fears of rejection as much as others do. But for those who are designing an online dating platform it may be wise to reconsider using a rating system at all. “Amazon, Yelp and other one-sided matching platforms use ratings to drive sales.” says Bojd. “But a rating system in a two-sided matching market may actually hurt and not help. Especially if there is a potential fear of rejection.”

Bottom line: you can’t copy and paste a feature from other platforms. As Bojd warns, “The rating feature is ubiquitous in most one-sided matching platforms. But those ranking features might trigger different mechanisms and create unintended results in a two-sided matching market.”

Final Thoughts

“A lot of times when we try to estimate actual preferences in markets, we look at the stated preferences,” says Bojd. “But sometimes we need to disentangle strategic behaviors from the stated preferences to estimate the actual preferences. For examples outside the online dating world, let’s look at how schools process admissions. An average school may give admission offers to less-desirable students to play it safe and avoid waiting for the best students in the market to apply.”