Putting The Data Into Dating

What can we understand about human behaviour from dating website data?


What exactly it is that we find attractive about another person is often unknown to us. Many girls say they like bad boys. Which is why I shoplift so openly. That, and all the free Kinder Eggs. These same girls will shout at you when you pull their chair away though, so who knows if it’s even really true. The truth is that people often don’t know what they want in a partner. Data scientists are currently working hard to establish exactly what it is though.

Dating sites claim to be able to find you the perfect match using data science. With hundreds of millions of users, they are awash with information about people, and in order to work effectively they use some powerful predictive analytics algorithms to leverage this. Many, such as Match.com, make users take personalized surveys to gather all the information they can about their preferences. They then match people based on this.

The major complication is that people are extremely messy, and people aren’t necessarily attracted to who they think they are going to be. The data gathered from dating platforms is also notoriously unreliable. People often exaggerate their personal attributes on profiles so the data lacks accuracy. According to a Berkley study, 81% of online daters have at some point reported inaccurate information about their weight, height, or age.

In order to navigate both of these issues, dating sites are constantly tweaking their algorithms to better match people up. Plenty Of Fish, for example, uses contextual searches, taking words that may appear in someone’s likes, such as ‘snowboarding’, and linking them to others in the same field. So for snowboarding, they would be matched with people who liked all outdoor sports.

Tinder is different to Match.com and apps that require people to fill in forms. Tinder relies on profiles that are primarily focused on images, and there is relatively little information about the user. The app works by people swiping right if they want to engage with another person, and left if they don’t. If both swipe right, they can start talking. These swipes generate a massive amount of data, as there can be as many as one billion per day. From these, there is a lot we can learn about attraction. Buzzfeed ran a study of what caused Tinder users to pick others, with much of it unfortunately coming down to race. There have been a number of other insights though. Mentions of words like divorce and separated gets men 52% more messages, for one. Women also get 60% more attention if their photo is taken indoors, whereas men get 19% more attention if their photo is taken outside, which suggests something about men still having a traditional attitude to gender roles.

Understanding the data produced by dating sites is not only important to allowing them to develop their algorithms, it also helps us understand more about society and our gender’s attitudes to one another.


Read next:

Social TV: Cross-Channel Insights on the ShareThis Platform