New algorithm increases successful dating site matches by more than 25%
[Sept 10, 2022: Ashley Smith, Institute for Operations Research and the Management Sciences]
Accounting for user preferences and current user experience, a new algorithm can increase the number of matches by at least 27%. (CREDIT: Creative Commons)
New Study Key Takeaways:
User behavior shows that when a user has several matches, they are less likely to “like” other profiles.
Accounting for user preferences and current user experience, a new algorithm can increase the number of matches by at least 27%.
These findings can be applied to freelance or task-based work, ridesharing and travel accommodations.
Online dating is one of the top ways people meet in 2022. That said, matchmaking – online or offline – is never a perfect process. New research showcases a new algorithm to increase dating site matches by analyzing individual user preferences and whether the user currently has a lot of matches.
The study, “Improving Match Rates in Dating Markets through Assortment Optimization,” was conducted by Daniela Saban of Stanford University, Ignacio Rios of the University of Texas at Dallas and Fanyin Zheng of Columbia University.
Related Stories
In the past two decades, hundreds of dating services have emerged worldwide, making dating a $12-billion industry worldwide. Moreover, online dating platforms have become one of the most common channels for couples to meet: 39% of heterosexual couples and 65% of same-sex couples that met in the US in 2017 did so online.
A common feature across many dating platforms is that they display a limited number of potential partners’ profiles (or simply profiles) to each user on each day. Some platforms, like Tinder and Bumble, implement this by imposing swipe limits, others put in place a limit on the number of likes (e.g., Hinge), and still others explicitly limit the number of profiles displayed on each day (e.g., Coffee Meets Bagel).
As described on Bumble’s website, platforms do so to “help foster more genuine, quality connections for our users and encourage more intentional swiping.” As a result, one of the primary roles of dating platforms is to select the set of profiles—the assortment—to display to each user on each day, based on the preferences and characteristics of the users involved. This is the problem we study in this paper.
The aforementioned problem resembles the classic assortment optimization problem, where a retailer must decide the set of products to display in order to maximize the expected revenue obtained from a series of customers. However, distinctive features from the dating context make our problem particularly novel.
First, both users must mutually agree—by liking each other—to generate a “match,” which considerably affects the probability that a transaction occurs. Thus, platforms should consider the preferences and behavior of the users on both ends of a potential match when making assortment decisions.
Second, users interact often and repeatedly with the platform, with those living in the same geographical area being part of the same “market.” Importantly, users may interact sequentially; i.e., users need not see each other’s profile (henceforth, see each other) in the same period. Thus, platforms must carefully manage the timing of these interactions. Notice that some of these features are not exclusive to dating platforms, and may be relevant in other online platforms, including freelancing (e.g., UpWork), ride-sharing (e.g., Blablacar), and accommodation platforms (e.g., Airbnb).
The size and relevance of the dating market highlight the need to make these platforms more efficient.
The researchers used data from a dating site that only allows users to view a certain number of profiles per day, no matter how many times they log on. But how do you choose which profile to show to a user and when?
“Users become pickier if they currently have a lot of matches, so their probability of liking a new profile decreases,” says Saban, an associate professor of operations, information and technology in the Stanford Graduate School of Business. “By incorporating this user history and individual preferences, our new algorithm can increase the number of matches generated by at least 27%.”
“There’s a lot of emphasis on correctly understanding user preferences and matching that way, but that’s not all that should be considered. Our work shows there’s a lot of improvement that can be made to better understand how users’ decisions change based on their recent experience on the platform. Keeping this in mind, we can increase new matches by carefully “timing” when we show users some profiles that are more likely to end in a match. If a user has had many matches recently, then it is better to wait to show them profiles that are likely to generate a match, and in the meantime show other profiles that are less likely to end in a match,” says Rios, an assistant professor in the Naveen Jindal School of Management at UT Dallas.
The researchers emphasize that these findings are also relevant to other types of online matching platforms, including those for freelance or task-based work, ridesharing and travel accommodations.
For more good news stories check out our Good News section at The Brighter Side of News.
Note: Materials provided above by Institute for Operations Research and the Management Sciences. Content may be edited for style and length.
Like these kind of feel good stories? Get the Brighter Side of News' newsletter.
Comentários