Can predictive analysis be used to find true love? Researchers at eHarmony have begun using such tools to build the algorithmic models that power the online matchmaker's relationship services.

When dating site eHarmony got its start in 2000, it hired social psychologists who interviewed married couples to figure out which characteristics--values, interests, personality traits--make for a successful relationship. Today the company has more than 11 million registered users and claims that, on average, more than 90 eHarmony members marry every day after being matched on its site. (No mention of how many matches end in divorce.)

The cyber-Cupid is shooting for higher success rates. That means identifying more variables to add to the operational models and more effectively measuring the variables it already has. "There have to be tons of things we haven't identified yet," says Steve Carter, eHarmony's senior director of research and development.

The software, from SPSS, will be used by eHarmony for scientific research, brand development, compatibility models, customer satisfaction and retention, and product research. One plan: Begin tracking couples before they get married to figure out which relationships last and which ones don't. The company should think about sharing those significant findings with the rest of the world.

Illustration by Steven Noble

Return to the story:
Businesses Mine Data To Predict What Happens Next

Continue to the sidebar:
Systems Management Tools Predict Then Adapt