Can You Judge a Person by His E-Mail Address?

March 10, 2011— -- Before you send your next e-mail to a potential boss or blind date, you might want to make sure that your e-mail address itself gives the right impression.

The recommendation engine Hunch.com, which asks users a set of questions to create individual "taste profiles," mined its database of answers and found that a single e-mail address can say a lot about the person behind it.

People with a Gmail account are likely to be thin, city-dwelling men between the ages 18 and 34. AOL users? Picture an overweight, suburban woman between 35 and 64.

"We were surprised at how much differentiation there was," said Kelly Ford, the company's vice president of marketing.

Considering how many tens of millions of people use the top four providers -- Gmail, AOL, Yahoo and Hotmail -- he said the company thought it was possible that they wouldn't unearth significant differences between the users.

But they were actually able to build distinct and detailed profiles for each provider's average user.

When it comes to Gmailers, for example, Hunch wasn't only able to discover that they're well-traveled, career-focused, tech-savvy agnostics. The site also found that they're partial to "salty snacks" and lounge around their apartments in T-shirts and jeans.

Yahoo users, on the other hand, tend to be overweight women ages 18 to 49, with high-school diplomas, children and a sweet tooth.

Hotmailers are more likely to be suburban women of average build, aged 18 to 34, according to Hunch. They're more likely childless, not religious and pessimistic.

The @AOL crew -- the oldest of the bunch -- tend to put family first, read magazines and live in the suburbs. They likely haven't traveled internationally, but are considered "optimistic extroverts."

So how did Hunch come up with such comprehensive profiles? By asking hundreds of thousands of Web users hundreds of quirky questions.

With the goal of personalizing each user's Internet experience, the website pairs traditional questions regarding users' location, age and marital status along with more obscure inquiries like, "When did you last blow a dandelion?" and "When did you last see the sunrise?"

The answers not only help the site match user profiles with preferences, they help the site's algorithms get smarter over time. Even though the questions are a bit offbeat, the site says they are very predictive of people's tastes. Once Hunch has a user's answers, it uses its growing database to make inferences and fill in the gaps.