Teams Work: Social Search Gets Results

Page 3 of 3


Separating Signals From Noise
While Chris Anderson’s legendary Long Tail concept points out there’s a lot of mileage--and money--in hit and miss content, following this formula for success forces companies to over stock their sites and, consequently, make it increasingly difficult for people to find what they want within an acceptable click-distance.

“Too many business websites are stumbling because their reliance on first-generation search technology and static interaction design has become a huge liability,” notes Dan Keldsen, senior analyst at Delphi Group. On paper, social search goes a long way toward uncovering the gems of content buried deep in destinations including catalogs, websites, and blogs, but it is designed to mine the expanse of the internet, not single websites.

But what if individual companies and publishers had the tools to pick up on how people navigate their websites to find what they want? More importantly, what if that same software could somehow identify and apply these patterns to the visitor population at large, thus tackling the problem of how to help users accomplish what they need to on the site? Surely, software that could intuit what people want the moment they land on a site, and then methodically guide them to what they want in just two clicks, would pay dividends. In fact, it already does.

Baynote, a provider of community-guided software and services, has harnessed what it calls “the wisdom of invisible crowds” (people who are anonymous and unknown to each other) to tap into what people really want, observes Jack Jia, Baynote’s CEO. In his view, tapping into the wisdom of invisible crowds is a far better indication of customer behavior than tracking the visible crowd--the one percent of hyperactive contributors who share reviews, write blogs, and fill out surveys. “Generally speaking, the majority of people in this group are opinionated, have too much time on their hands, or have some other hidden agenda,” Jia explains. “In any case, we can’t always trust the visible crowd; we have to tap the invisible one.”

To do this, Baynote has developed patent-pending content guidance technology that is transparent and intuitive. “Visitors are not burdened with the need to perform explicit actions, such as completing surveys, tagging, or engaging in other time-intensive and distracting processes,” Jia says. “Community wisdom is captured implicitly.”

This path-breaking approach borrows heavily from the insights outlined in the book Emergence: The Connected Lives of Ants, Brains, Cities, and Software by Steven Johnson. Ants, Johnson writes, are not intelligent as single insects. However, they develop a kind of collective intelligence, which he calls “emergent intelligence,” when they are interconnected in complex colonies. Put simply, a feedback loop between individual ants allows them to organize themselves and adapt to their environment.

Individual searchers are like single ants. “A website visitor is a lonely, invisible explorer thrust into a complex information landscape without any community support,” Jia explains. An ant colony is powerful because the members are connected. One ant that stumbles on food leaves a chemical trail for other ants to follow. When the food disappears, so does the trail.

In the case of the internet, a single searcher, like an ant, can happen upon an excellent result. But there is no way for the searcher to leave a trail for other searchers to follow. The result: Each searcher must endure the same process to find the content they want. The rising dissatisfaction with this trial-and-error method is captured in a study recently cited by The Wall Street Journal. It noted that 83% of visitors do not find the information they are looking for when using a search engine. These dissatisfied visitors either abandon the site, resulting in lost sales and ad revenue, or resort to service calls and emails that cost a company hundreds of dollars per hour.

“Baynote’s approach is to be that chemical trail that will lead searchers to what they want,” Jia says. A code embedded in the destination website tracks user behavior and signals the Baynote server to shuffle website content to match the right content to the right people. “We detect user intent through some twenty behavioral characteristics that in aggregate can tell us what a person wants--and therefore what like-minded people would likely want.” Baynote then drops a trail for the user to follow. “If they do, then we know we’re on the right track and can begin to promote the content the individual is most likely to appreciate.”

The impact is the same as if Baynote had produced a customized site for every single individual over thousands of profiles. But in reality it has harnessed implicit behavioral intelligence to help people find content they want on the first try. While people-powered search is unlikely to dethrone machine-based search from the likes of Google anytime soon, its innate ability to tap into individual or group-based knowledge, as well as harness the hidden intelligence of invisible crowds, make it a worthy complement to horizontal search. Indeed, there are markets in which pure algorithmic search may best be supplanted by social search, and it appears that some companies will need to harness the collective power of searchers or risk being left out of the wisdom of the in-crowds.


Sidebar: Eurekster! We Found It!

With search becoming the de facto interface to all digital content, it makes sense to give publishers tools to capture and leverage the knowledge and expertise of the user-community. Such an approach would provide people with the results they care about most and ultimately pave the way to a read/write search engine, allowing everyone a voice in deciding which results to display and in what order.

This vision of collaborative social search lies at the heart of Eurekster’s recent upgrade to its social search service. The provider of community-driven search has combined search algorithms with wiki-style collaboration into a what they call a “swicki,” empowering site owners and publishers to create a customized search engine that reflects the ideas and passions of the community at-large.

At first the company did this implicitly, using click-stream analysis to track user behavior and feed its insights directly into the “buzzcloud” (a display of the recently searched key words and terms weighted by popularity). But the newest version of this people-powered search tool walks the walk, encouraging people to contribute to their search experience directly by writing their own answers into the search results. What’s more, the community can vote on the relevance of answers.

It’s a people-driven popularity contest that can expose the best results and delivers the best experiences for the group on the whole, notes Steve Marder, Eurekster’s CEO. “It’s about serving the community and creating a search engine that is highly focused on the passions of the members.” To date, the company counts more than 95,000 swickis implemented across sites by 20,000 unique publishers, including Forbes, Techcrunch, and Pittsburgh Live.

And such tools don’t just potentially increase the value delivered to each member with each search; it makes the whole site more appealing for potential advertisers. To this end, the company recently took the wraps off a swicki ads service, allowing advertisers to buy ads on multiple swickis or make category-based buys across a similar group of swickis.

Looking ahead, Marder fully expects hundreds of thousands of social search engines to break on the scene. “Publishers must engage with their communities, and meet the needs of the communities they serve. Providing the community a means to participate in their search experiences strengthens those ties and builds loyalty.”


Companies Featured in this Article

Page 3 of 3