Background
Many internet dating sites have sprung up over the past fifteen years… such as eHarmony, Match.com, and Chemistry.com. While the most of these sites seek to aggregate profiles for users, eHarmony and Chemistry have created a different product model by integrating psychological profiling, proprietary algorithm development, and selective filtering to enhance the customer experience.
Both eHarmony and Chemistry.com require that customers complete a psychological profile before using the service. This information is then used to match customers based on their emotional, social, and intellectual compatibility. For example, an extreme extrovert and extreme introvert would likely not be a successful match, owing to one’s desire for social interaction while the other favors solitude. Furthermore, couples with substantial differences in IQ are prone to divorce.
Methods
Illusion of Scarcity: As mentioned in the Paradox of Choice by Barry Schwartz (and in keeping with the book’s tenets), eHarmony and Chemistry limit results to a finite number per day. This alleviates the judgment errors that plague competitors such as Match, or OKCupid, where most customers flock to the same users… overwhelming a few desirable individuals and largely ignoring the rest. The other undesirable side effect is that the abundance of opportunities actually reduces one’s likelihood of contacting anyone.
Selective Filtering: eHarmony rejects potential customers, if they are separated or already married, under the age of 20, have been married more than four times, or if the personality profile was filled out haphazardly. People are also rejected if they have anger management issues, a poor self image, consider themselves dishonest or are discontented with their lives. Contrarily, Chemistry.com has deliberately targeted such rejected users with some public relations success.
Customer Behavior: eHarmony and Chemistry both go beyond matching strictly based on the personality profiles. eHarmony has refined their matching process to “include crucial information about how users behave -- like time spent on the site or how long they take to respond to an e-mail about a match…[For example] Manhattanites are less willing than the general population to travel outside their immediate area to meet people, so eHarmony might not present matches beyond narrower geographical boundaries.” Chemistry’s Chief Scientist indicates that they’ve designed their process to maximize the neurological effects of dating, and the website indicates matching on interests as well.
Recommendation Systems: While much work has been done on recommendation systems in recent years, as one eHarmony executive said, alluding to the Netflix recommendation system, “It’s not just that you have to like the movie…The movie has to like you back.”
Strategic Benefits
Risk Avoidance: eHarmony screens out emotionally unstable individuals, as well as those who are not ready for committed relationships.
Branding: eHarmony boasts of having led to 271 marriages a day, and says they’re in the relationships business… not the dating business. Chemistry was spun-off from Match.com to compete directly with eHarmony’s business model owing to that model's success.
Networking and Scale: The branding benefits contribute directly to both scale and networking as a competitive advantage because of the virtuous cycle created by word of mouth advertising. The immediate benefit of scale is to improve geographic coverage and the density of coverage with more users. Scale also allows eHarmony to use experimentation on significant samples to improve their service (a capability that sparsely populated competitors can't duplicate). The networking benefits are more profound, because each additional user's value is equal to the number of connections they make possible. If there are 100 women and 100 men, there are 10,000 possible connections/couples. The addition of another woman creates 100 more possible connections/couples, thereby conferring a competitive advantage on larger populations.
Psychological Profiling and Cumulative Learning: Users are matched largely based on the well documented Five Factor Model, which classifies customers based on their openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism. These results are then used to couple similar users. This ensures that organized individuals are put together, extroverted people are matched, people that are adventurous are put together, etc. The cumulative learnings gleaned from years of match algorithm improvements also creates an advantage which can't be recreated without similar investments.
Commenary
What I find particularly thought provoking about this these services is that the algorithm itself is the product, whereas Netflix recommendations or Apple iTunes recommendations are ancillary to the service. I find it even more interesting, that the algorithm is used to drive a social dynamic... people signing up for the service, based on the belief that it leads to healthier relationships and marriage. This also feeds a virtuous cycle, whereby additional users increase the value of the service.
I am unsure about the Five Factor psychological model used by eHarmony and Chemistry, but I’ve heard that the results of Meyers-Briggs personality tests can vary based on temporary effects, such as weather and if the subject recently finished a meal. For example, people are inclined to respond they’re happy on sunny days and inclined to respond they’re contented when they have a full stomach. Accordingly, customers would be well advised to take the initial personality profile test after a gluttonous meal on a pristine day.
Shambora, Jessica. "eHarmony's Algorithm of Love." Fortune Tech. 23SEP2010.
http://eharmony-blog.com/104 "Your question answered: Why eHarmony rejected you." 12AUG2006
Schwartz, Barry. The Paradox of Choice: Why More Is Less. HarperCollins, New York, NY. 2004.
http://mashable.com/2011/07/13/match-com-equation/
