YOU ARE HERE: LAT HomeCollections

Fashion Diary: Google's, an online personal stylist

Google has created an algorithm for personal style with its new fashion retail website. Here's how the search engine learned what clothes and accessories to suggest.

November 21, 2010|By Adam Tschorn, Los Angeles Times

When Google took the wraps off its foray into online fashion retail a few days ago, anyone who hadn't sat through one of the nearly hour-long Web demonstrations couldn't have been faulted for thinking was just another in an increasingly crowded field of "curated retail" sites. (At launch, the site focused on women's clothing and accessories only — with the goal of eventually expanding into menswear.)

It is — and it isn't. Aiming to serve up clothing and accessories based on personal preferences (the way Pandora suggests new music based on the music you like) is hardly new territory. Nor is enlisting fashion taste-makers (drafted by to help cut through the clutter (users can shop the preferences of designers such as Diane von Furstenburg and Derek Lam and celebrities such as Carey Mulligan and Elisabeth Moss).

Fashion Diary: In the Nov. 21 Image section's Fashion Diary column about Google's, the first name of Manjal Shah, a director of product management at Google, was misspelled as Manjul. —

But in fine-tuning the search processes they use to winnow the wardrobe choices, the folks at the Googleplex have done something noteworthy — they've essentially created a personal stylist algorithm.

Manjul Shah, a director of product management at Google, explains that as a result of trying to tackle some of the challenges peculiar to Internet clothes shopping, the computer program, in essence, "learned" to style.

"Shopping for clothes is about discovery and not just search," Shah said. "You go into a store looking for one thing and you might come out with nine other things too. We needed to find a way to create that 'bump into' experience."

One way to do this, Shah explained, was to "teach" the search engine to identify items of clothing and accessories by one of several basic style genres: classic, romantic, casual chic, edgy, street and boho.

This was accomplished by hiring 100 fashion-savvy folks to weigh in. "We had a hundred fashionistas and local fashion students tag 50,000 different items," he said. "[And] each item [was tagged] by two different people." This allows the program to more accurately suggest items that fit the user's personal preferences. For example: if you're an "edgy" and you search for a black dress, you won't, in theory, be offered an ankle-length boho sweater dress.

Fashion Diary: In the Nov. 21 Image section's Fashion Diary column about Google's site, the first name of Manjal Shah, a director of product management at Google, was misspelled as Manjul. —

Trying to duplicate another aspect of the bricks-and-mortar "bump into" experience resulted in creating a set of fashion guidelines for the search engine to follow.

"In addition to computer-learning PhDs and scientists, our team had people who went to fashion school and designers and a stylist who literally wrote fashion rules like 'heavily patterned dresses don't go with heavily patterned handbags,' and 'vertical stripes don't go with horizontal stripes.' She wrote hundreds of these rules to go with our 'complete the look' algorithm."

As if that's not HAL 9000 enough for you, on top of that, the site taps into Google Trends data so if a customer is shopping for dresses in the spring — when the word "floral" is trending high, for example –- searches will automatically seed more floral offerings into the mix. If boyfriend blazers or jeggings are hot topics, the offerings served up once a day or once a week by e-mail will reflect that.

So as Google gives us the ability to accessorize by algorithm, we're about to really find out if computers can acquire a sense of style.

And if they can't, you can always just shrug your shoulders and say your motherboard let you leave the house that way.

Los Angeles Times Articles