HeyStaks makes searching social

Search engines don’t seem to have evolved much since the dawn of Google. If Google was the sports car of search engines when it launched, it’s now a rather rusty vintage sport car.

Enter HeyStaks, a new startup launching today which wants to make search more efficient by making it social. HeyStaks revolves around the notion of a Stak, which synthesizes the best shared search results of a group of users on a particular subject.

Granted, social search is already a crowded field, with a host of startups trying to solve the problem. And Google itself has said it will be putting a “social layer” on its products to combat the rise of Facebook, though it’s not yet clear what that will look like.

But if you look at the amount of searching that gets done, it seems reasonable that a startup like HeyStaks might slice off a piece of the market. Search is still a time-consuming and often frustrating business, especially if you are searching for information on a more esoteric or specialized subject. According to search engine marketing firm iProspect, a typical knowledge worker spends 16 hours a month searching for information and 50 percent of all those searches fail. HeyStaks’ founders claim that 1 in 4 searches are repeats of your own past queries and 2 out of 3 of searches have already been executed by someone else in your social network.

In HeyStaks, each user has a default Stak of their own searches and can start or join other Staks. A Stak might cover startup advice or travel in San Francisco. As a user, you choose your preferred Staks and who to collaborate with to get better search results. You might have one friend who has great insight into the design scene but knows nothing about high-tech startups.

HeyStaks provides a browser plugin (currently Firefox only) and a mobile application. Once you have installed the plugin, relevant HeyStaks community search results start appearing in your Google, Yahoo and Bing search results.

The results to the left are from a search for “hard rock reviews”. The searcher is a mountain biker who is looking for reviews of Hard Rock mountain biking gear. Google thinks that the user is looking for the Hard Rock hotel or hard rock music, so its default results are not relevant. Since this user is a member of a mountain biking Stak, HeyStaks anticipates that the user is more likely to be looking for mountain biking results. So it makes suggestions accordingly.

HeyStaks is based on technology developed by a group of researchers in search, data mining and personalization. The company’s CTO is Barry Smyth, a prominent expert in personalization whose previous startup, ChangingWorlds, was sold to Amdocs for $60 million. Smyth told me that around the time Google launched there was a rival, and now now long-forgotten, search engine called DirectHit which used page popularity (the number of times a page has been selected by searchers) to rank pages. Like Google, it had the potential to deliver demonstrably better search results than existing search engines but it turned out to be easier to “game” and so Google won out.

Intrigued by this idea, Smyth’s research group embarked on various research projects to improve upon a simple, popularity-based approach to result-ranking. The group, including researchers Maurice Coyle and Peter Briggs, developed and patented a number of core technologies in this area.

The HeyStaks social ranking and relevance engine takes 10 different types of user behavior into consideration. These include behaviors like how often a user selects a page and whether they tag it, share it or post it on their social networks. Results to date suggest that HeyStaks recommendations can be up to 50 percent more relevant that the vanilla search engine results.

HeyStaks CEO, Jonathon Dillon, was previously a VP at Yahoo. He says Yahoo tried to do something similar with social search after the acquisition of Delicious 5 years ago but failed because social graphs were still too immature.

The success of HeyStaks depends on how willing users are to share their search results. The founders say, that in the private beta group of 500 users, 70 percent of users shared 70 percent of Staks with 3-4 people. A typical beta user got community recommendations for about 1 in 4 searches.

Anything that improves search results is relevant to advertisers. HeyStaks’ business model will, at least partly, be based on advertising, but since the advertising model is not yet launched, the company is unwilling to discuss it in detail. Another obvious application is knowledge sharing on company intranets.

HeyStaks was founded in 2008, has 9 employees and is based in Dublin and San Francisco. The company has seed funding of $1.4 million.