1 Comment

Summary:

StyleFeeder recently launched a personal shopper tool, which made me think it was a good time to sit down with Phil Jacob and his team to get a more in-depth look at the product and the technology. StyleFeeder has been around since the end of 2005. […]

StyleFeeder recently launched a personal shopper tool, which made me think it was a good time to sit down with Phil Jacob and his team to get a more in-depth look at the product and the technology.

StyleFeeder has been around since the end of 2005. Initially developed by Jacob, it was quickly bought by TopTenSources and lived as a little widget that you could add to your Web site or your blog. At the beginning of 2007, following a restructuring, it was spun out of TopTenSources as an independent entity, at which point Jacob got funding, brought in a business development team and more technical talent to build out the product.

A number of key features were added to the product during this build out, including a sidebar plugin (written in Flex) for FireFox and Internet Explorer, and a recommendation engine. The sidebar plugin is a dashboard to StyleFeeder, giving you access to all its functions. For example, it allows you to manage your StyleFeed and search for products while recommending products it thinks you will like, finding your StyleTwins, and managing your profile.

The recommendation engine is key to the system. When you first create an account on StyleFeeder, you are prompted to rate a number of product selections. This takes less than a minute and allows the system to gather information about your product preferences, which it, in turn, uses to make recommendations. Rating the choice further personalizes the system, and search results are personalized as well. The system can even pick out your “StyleTwins” — people with tastes similar to yours.

The rankings are computed on the fly, so they require some pretty heavy computation. But the StyleFeeder team feel that it would scale as more and more users joined the site.

The recommendation engine uses an algorithm called Maximum Margin Matrix Factorization. What makes this algorithm interesting is that it takes into account how people rank items, adjusting for those who tend to make extreme rankings compared with those who lean more toward the middle. You can learn more about it on the StyleFeeder tech blog.

Shopping is a social sport, so there is a social side to the system. You can chat with other people on the site via a built-in messaging system, shop online with friends, and keep tabs on what other people are adding to their StyleFeeds. This is, of course, pretty standard stuff for a social shopping site. To differentiate themselves from their competitors, StyleFeeder has been focusing on the personalized shopping search and its personal shopper tool application.

The company also recently created a Facebook application that allows people to add their StyleFeed to their Facebook page. Phil told me that the aim, more than anything, was to use Facebook as a marketing vehicle. It appears to be working; more than 10,000 users were added over the last week.

Currently there are about 125,000 products listed; users can add products to the system using a tool that sits in the browser navigation bar. The system will then prompt for additional product metadata, automatically filling the URL, the picture, and the description. There is some product duplication, such as with the iPhone, but there are plans to fix that.

StyleFeeder wasn’t willing to disclose member figures, but they did say they’ve been growing at the rate of 30 percent a month for the past eight months. And since Kaboodle was acquired by Hearst, the company has received strong interest from potential acquirers interested in both the technology and the team.

You’re subscribed! If you like, you can update your settings

  1. StyleFeeder Fed $2M – GigaOM Tuesday, January 29, 2008

    [...] Highland Capital Partners and Schooner Capital. The Cambridge, Mass.-based company, which recently morphed into a full-fledged site (and maybe business?) from a widget, offers recommendations about clothes, [...]

Comments have been disabled for this post