Loomia: a product-recommendation Web service

At the San Francisco New Tech Meetup, I saw a presentation from Loomia.com--a company that's working to give even the smallest sites a competitive edge by providing a recommendation engine as a Web service. With Loomia, just about anyone can add Amazon- or Netflix-style recommendations to their site just by pasting a bit of JavaScript into the HTML code.

Loomia provides two different kinds of recommendations. The first will place a listing of similar products or stories on your pages. (See BetterPropaganda.com for an example.) It's a "see more like this" function for any site that has a catalog of items on it.

The second recommendation system is a more interesting personalized list that shows you what items on a site you might like given your behavior on that site. The site tracks your clicks and also offers a voting system. Results come from correlating this data with items or stories that similar users clicked often and/or rated highly. I tried LearnOutLoud.com, a Loomia-powered site that offers downloadable audio books and videos, and found the results to be pretty good. After rating just 4 books, the system made 16 recommendations, unwittingly suggesting 6 books that I have already read and enjoyed. After removing those 6 by rating them and selecting a few other books for a total of 15 ratings, LearnOutLoud.com recommended 5 books already on my reading list.

Loomia's Web service could be great for smaller shopping sites looking to improve their profitability by displaying items users are more likely to purchase. It should also help sites keep customers coming back, as recommendations become more accurate (and therefore more valuable) the more time a customer spends browsing the site.