EBay has announced it is buying New York startup Hunch, a recommendation engine created by Chris Dixon and Caterina Fake, to help improve its recommendation services. The purchase price hasn’t been announced, but venture capital investor and former TechCrunch editor Michael Arrington has said it was about $80 million.

EBay said it will use Hunch’s “taste graph” technology to provide its users with non-obvious recommendations for items based on their unique tastes. The company said it will also apply Hunch’s technology to other areas such as search, advertising and marketing, in order to better surface product information based on its customers’ tastes.

“We are engaging consumers in innovative ways and attracting top technologists to shape the future of commerce,” said Mark Carges, Chief Technology Officer and Senior Vice President, Global Products, Marketplaces. “With Hunch, we’re adding new capabilities to personalizing the shopping experience on eBay to the individual relevant tastes and interests of our customers. We expect Hunch’s technologies to benefit eBay shoppers as they browse and buy, and to bring sellers on eBay new ways to connect the right products with the right customers.”

Hunch will remain in New York with co-founders Chris Dixon, Tom Pinckney and Matt Gattis staying on board with the company. Fake, who also co-founded Flickr, stepped away from the company earlier this year.

Dixon said in a blog post that he and the team decided to sell because of the opportunity to apply Hunch’s Taste Graph to one of the top e-commerce leaders. He said Hunch will continue to operate a standalone site and will be providing eBay with predictive merchandising, interpreting unstructured data and creating merchant insights. Hunch had reportedly turned down previous offers for as much as $60 million from suitors including Google .
I think the purchase could improve eBay in a number of ways. For one thing, it could help the retailer better compete with Amazon, and it keeps Hunch out of the hands of its rival, which has been well-regarded for its ability to recommend products to users based on their purchase history and preferences. EBay launched its own recommendations last year to help suggest items based on past searches, but it clearly has a long way to go in matching Amazon, which moved ahead of eBay in sales last year. Earlier this year, Om wrote that Amazon should buy Hunch because of its ability to create an interest graph that can be tied into social commerce. And as retailers like Amazon and eBay build out almost limitless inventories, matching recommendations to these products becomes even more important. Here’s what Om wrote in April:

“Amazon should buy Hunch. It could use the decision engine to help customers sift through the ever-expanding array of offerings and make purchasing decisions. That little kernel of an idea still looms large in my thinking, especially as I wonder what the future of media and e-commerce looks like….Interest graph, for me, is the underpinning of a new kind of e-commerce experience. Think of it as a new kind of social commerce experience that goes beyond the notion of group shopping (Gilt Groupe, Groupon), shopping communities and recommendation engines,”

Hunch could allow eBay to create some very smart recommendations based on more than just searches. I think past search or purchase history is limiting, and sometimes an inaccurate predictor for recommendations because a lot of gifting happens on these sites, which can skew the suggestions. Hunch uses what it knows about a person’s likes and interests along with answers to personal questions to help understand a user’s tastes, then makes very intelligent recommendations based on those answers and their correlation to other data. Hunch originally began a consumer-facing service but changed course last year to license its technology to retailers who participated in its Hunch Partner Platform.

I think this could also help with eBay’s self-described goal to become a technology-driven company. The company is in the midst of a big push to become much more than just a seller of goods. You could see that at eBay’s X.commerce Innovate conference, where it showed off its X.commerce platform for retailers and developers. EBay has been on a buying binge recently, buying a number of companies that have helped it aspire to be a sort of plumbing or infrastructure for e-commerce. It has bought GSI, Magento, WHERE, Milo and RedLaser, much of which is being put into the new X.commerce platform and also its bid to enable payments in-store. Hunch is a very intelligent company that uses a lot of machine learning and data mining to create recommendations. And it has a smart team led by the well-known Dixon, who can also help build out eBay’s presence in New York

Adding Hunch could be another tool for X.commerce customers, who could plug smart recommendations into their business operations. EBay is really trying to build a commerce operating system for businesses, and it’s putting all these new assets together into one big resource to help enable sales for retailers and developers. Increasingly, retailers need better ways to engage consumers and personalize service to the tastes of their users. By providing a tool like Hunch to X.commerce customers, it can make the platform that much more attractive.

I think the move makes sense for eBay and follows through on CEO John Donahoe’s attempts to remake the online seller into a technology company. Donahoe said at the X.commerce conference that eBay is not done buying up companies and it looks like it will continue to add more pieces to its growing collection of e-commerce assets.

Here are a couple of video interviews we did earlier with Dixon:



Watch this video for free on GigaOM





Watch this video for free on GigaOM



Related research and analysis from GigaOM Pro:Subscriber content. Sign up for a free trial.
Connected world: the consumer technology revolutionThe future of mobile: a segment analysis by GigaOM ProFlash analysis: Collaborative consumption – a first look at the new web-sharing economy

Wikets, a soon-to-be-released iPhone application that recommends places and products to friends, announced a $1.5 million seed round from Andreessen-Horowitz and Battery Ventures today. But will yet another recommendations product survive when reviews and opinions already litter the Internet?

I imagine these recommendation apps to be like little bodies treading water in the Internet ocean. Apps like Where exist in rafts, websites such as Yelp are the ships that float on their own, and companies that provide exit strategies or funding are the lifeboats searching the scene for people to pick up.

Not every treading body drowns. Wickets is hoping a classic approach to recommendations will help it edge out competitors, openly acknowledging that the space is crowded.

“We are doing it the old fashioned way. Our goal was to come up with a way users can put their best recommendations at their friends’ fingertips,” said Wickets chief executive and co-founder Andy Park in an interview with VentureBeat.

Wickets works by allowing you to follow other Wikets members and see their recommendations in a stream. Recommendations are made by searching a particular place or company, finding the product and adding your two cents right on its page. You can save recommendations in wishlists and define a wishlist by topic. Commenting is also enabled on the various reviews.

Based on how many people “re-recommend” what you’ve already recommended or add your review to their wishlist, you get points, which can result in gift cards. The company is not releasing the names of all of its retail partners, but does say the first gift cards awarded could be from Amazon, iTunes or other large scale retailers.

Park says that the app solves two problems: having to find a product and send an email or text regarding the product, as well as time spent waiting for a recommendation when requested from friends. The app solves these issues by attaching a review directly to the product or place page and, in theory, your friends’ recommendations will already be available on the app, so you won’t have to query the person directly.

This requires that people actually use the app, however. The app can’t reach its full functionality until a majority of your own friends are using it. So how will Wikets get people to use the app? The company believes Facebook and Twitter connect, a prompt to find friends, and a natural desire to meet people will bring in the required amount of users.

By “natural desire to meet people,” Park means that because you can see friends-of-friends’ reviews, you will want to get to know those people as well. For example, if you’re with someone who mentions a restaurant your friend reviewed, you can offer to invite that friend along since you know they like the place. It’s a little intangible, however, and hard to measure if downloads will directly correlate to people meeting each other.

User acquisition is a big problem for app companies today, which have turned to gamification, or a rewards system such as Wiket’s as a way to lure people. A number of crowdsourced products such as CrowdTwist, BunchBall and Needle are all doing this. But as any gamer knows, getting rewarded is great, but its entertainment value will die if the means by which you get rewarded is boring.

Parks did, however, say that he is saving specific strategies for the app’s launch.

The growth of a company like this is very much up in the air, especially as funding becomes less and less of an indicator that a company is worthwhile.

Matt McCall of Chicago venture firm New World Ventures told VentureBeat, “There are just way too many companies getting funded.”

For now, however, Wikets has been pulled out of the ocean to develop its app with capital from investors Battery Ventures and Andreessen-Horowitz. It is using the funding to launch the application, which went into full development after the company closed on the round in May.

Prior to the round, Park knew Marc Andreessen and Ben Horowitz when his former company BladeLogic was a competitor of their Opsware, both optimization focused data centers. BladeLogic went public in 2007 and was sold to BMC Software in 2008. The teams grew mutual respect over the years and are collaborating this time around.

Wikets plans to launch the iPhone app in October, as well as announce its strategy and partners further. For now, the seed round will keep the project afloat.

[Photo courtesy of Andrew Doran/Shutterstock]

Filed under: deals, mobile, VentureBeat

The New York Times released Thursday a finished version of the Recommendations platform it quietly introduced in beta in late January.
Available at nytimes.com/recommendations and on the “Recommended For You” tab on article pages, the tool is designed to help logged-in readers “see through the news fog,” as NYT lead technology reporter Nick Bilton put it. It serves up recommended stories based upon the kinds of articles visitors have read.

“We wanted to make the site more engaging, to expose content to our readers on a more customized, personalized basis — and not customized in the way you select your topics like a My Yahoo or iGoogle, but more of a passive personalization,” Marc Frons, CTO of The New York Times, explains. “We created an algorithm that exposes users to content they may not have seen otherwise,” he adds.
The algorithm is one of the most sophisticated we’ve seen on a news site. When serving up recommendations, it calculates a number of factors, including recency (visitors who tend to gravitate toward breaking news should see recommendations for more timely topics), sections, topics and keywords.
A New Model for Curation

Recommendations is part of a broader exploration of new curation and aggregation methods for Times readers. For time immemorial, the editors of the Times have determined what appears on the cover of the paper and, beginning in 2006, what appears on the front page of nytimes.com.
Now, the front page of nytimes.com shows recommendations from one’s Facebook network alongside stories chosen by Times editors. Visitors can easily navigate to the “Most Popular” tab to surface stories that have proven most popular among bloggers and readers.
In the future, we wouldn’t be surprised to see the front-page content of nytimes.com divided into three sections: one for stories recommended by human editors, another with stories recommended by one’s social network and a third that delivers stories chosen by the site’s internal recommendations engine.
“The challenge is to balance recommendations that are editorially driven, based on what editors think is important, with recommendations from the social sphere and algorithmic recommendations, based on what you’ve read and who you are, your own likes and dislikes,” Frons says. “It’s something we’re constantly looking at and experimenting with.”
It’s a tough challenge, especially in light of the Times‘s emphasis — and subsequent reputation — for editor-driven curation. So far, the Times has proven to be open-minded and progressive without overwhelming readers, for which we commend them.
More About: media, new york times, Recommendations
For more Media coverage:Follow Mashable Media on TwitterBecome a Fan on FacebookSubscribe to the Media channelDownload our free apps for Android, Mac, iPhone and iPad