Stylefeeder gets $1m for social shopping

October 8, 2007

Launched in 2005, acquired by TopTenSources then recently spun out, Stylefeeder just raised $1 million from Highland Capital and Schooner Capital.  SF also raised money from RSS Investors.


Comparatively, ThisNext has raised $3.5mm to date and is looking to raise $7-10mm more, so on the surface it looks like they’re losing the “money game” — SF counters this by saying that they’ve got the hardest part out of the way…the technology.

And that’s what makes Stylefeeder interesting.  The site is actually based upon a product discovery engine (hence the “personal shopping engine” tagline), as opposed to relying on users to “suck in” products from the outside via bookmarklet or whatever.  It’s also adaptive — so, in other words, as you add products to your “stylefeed,” the engine learns what styles you prefer and weights the products you’re more probable to like at the top of the heap.

At the same time, Stylefeeder offers their own bookmarklet — I suppose by offering a number of product introductory mechanisms it enriches the holistic experience.

A couple weeks ago, Stylefeeder launched a personal shopping tool that plugs into Firefox and IE, offering access to the site’s functions and making it simpler to pair you up with your “StyleTwins” or sending a “Shopping SOS” in case you can’t possibly decide between that black or red BMW M6. (I suppose I should be flattered, one of my top StyleTwins is John Palfrey, founder of RSS Investors and a Clinical Professor of Law at Harvard Law School and Executive Director of the Berkman centre for Internet & Society).

Again, the key part about Stylefeeder is technology.  It’s fairly sophisticated stuff and enriches the process since product recommendations and StyleTwins are created on the fly and dependent upon what you as a user have added to your feed. And I’ve been impressed by the recommendation quality thus far.

Part of the technology behind the service is called Maximum Margin Matrix Factorization, an intense rating /recommendation algorithm that “predict(s) preferences based a user’s rating history.”

Stylefeeder is based in Boston.  You should check them out; there’s some great things happening there.  I can foresee an acquisition by a major mainstream CSE player.

Written by Scott Hurff