Story from Information Week
The company hopes bolstering its semantic search capabilities will help it better compete with Google for ad dollars and happier customers.At media event in San Francisco on Tuesday, Yahoo Search executives insisted that Internet users don't want to search.
"Nobody really wants to search," declared Prabhakar Raghavan, head of Yahoo Labs and Yahoo Search Strategy. "People want to run their lives."
The stated theme for the press briefing was "The End of the 10 Blue Links," a title that reflects an evolution of Yahoo's search strategy beyond document retrieval. As described by Raghavan, Yahoo is directing its search efforts toward assessing user intent.
When a user types "Star Trek," Raghavan said, he doesn't want 10 million documents, he wants actors and show times.
Yahoo's bid to redefine search as a matter of intent rather than index size can be seen as an admission that it can't match Google's index.
In August 2005, index size mattered to Yahoo. "[O]ur index now provides access to over 20 billion items," Yahoo's Tim Meyer boasted at the time. A month later, Google's Marissa Mayer answered Yahoo's challenge, stating that Google's index was three times larger than anyone else's. And in July 2008, Google said that its index had reached 1 trillion unique URLs.
So it's perhaps understandable why Yahoo might want to reframe the debate. Given its lack of success challenging Google directly -- Google's April search share in the United States reached 64.2%, a 0.5 point gain, while Yahoo's search share fell to 20.4%, a 0.1 point decline, according to ComScore -- Yahoo wants to change the game.
And to some extent, the game has changed. Everyone is focused on mobile applications now. And in mobile search, user intent is easier to determine because location data often provides a clue about what users want. A search for "Star Trek" from a mobile device is more likely to reflect a desire to find a nearby theater and purchase tickets than it would be from a desktop PC, for example. To understand user intent better, Yahoo is developing its semantic search capabilities. Raghavan describes the process as building a Web of objects in the real world and understanding how they relate to each other. In other words, Yahoo is adding structure to its data to make inferences about user intent more accurate and to define relationships between search terms and things in the world.
Rather, its partners are doing that through initiatives like SearchMonkey. SearchMonkey allows developers to share structured data with Yahoo, enabling the display of search results enhanced with related information. Thus, a Yahoo SearchMonkey search result for a restaurant might include not only a link to the restaurant's site, but star ratings, reviews, map links, and related data, all in one search listing.
This is similar to what Google does with its OneBox search results enhancements, but Yahoo is relying on its partners to feed it with structured data.
The gambit appears to be working, too. Since October, the amount of RDFa structured data available through SearchMonkey has increased by 413%. Yahoo reportsthat 70 million SearchMonkey-enhanced search results are viewed by users every day, in 23 markets around the world. And these search results deliver more traffic to site owners, too. Some site owners say they've seen a 15% increase in click-through rates, according to Yahoo.
Yahoo's BOSS (Build your Own Search Service), the company's open search platform, has also been growing. Having recently been upgraded to allow developers to access SearchMonkey structured data, the BOSS API now handles three times as many queries as it did six months ago. With daily query volume in the 30 million to 35 million range, according to Raghavan, BOSS alone is just shy of the estimated 40 million queries handled by Microsoft (NSDQ: MSFT)'s Windows Live Search.
Looking ahead, Yahoo is testing new search enhancements based on its improved ability to understand user intent. One of these, for example, draws on Yahoo's data about locations to suggest related queries. Larry Cornett, VP of consumer products for Yahoo Search, demonstrated how a future search for "Paris" might produce a rich set of pictures and links of other Parisian landmarks like the Musée du Louvre and the Eiffel Tower, data not normally brought up for that keyword.
Yahoo's focus on user intent could lead to happier users, if Yahoo Search can guess user intent accurately. It could also help Yahoo make more money from advertising.
"If we can divine the user's intent, that's obviously of great interest to advertisers," said Raghavan.
