In an attempt to boost its search-ad business, Yahoo has begun a project that lets anyone build a customized search engine atop the Internet company's technology.
Essentially, BOSS is a bid to enable others' search innovation then share profits from the results. It's also the most significant example to date of Yahoo Open Strategy, the company's effort to expose its own technology for outside developers in an effort to become a more indispensible part of the Internet.
The BOSS API (application programming interface) to Yahoo's search is free to use, but BOSS partners that succeed will be required to show search ads, said Prabhakar Raghavan, chief strategist for Yahoo Search.
"We fully expect it to expand the footprint of Yahoo search advertising on the Web," Raghavan said. "There is no payment of any kind we expect from partners, but we do say in the terms of service up front that over time we will require them as they build and grow out to use our search advertising."
That's a strong statement, given Google's rapid ascent and strength in that very market. Even Yahoo, faced with intense shareholder pressure and a hostile takeover attempt by Microsoft, has tried to hitch itself to the Google star through a search-ad deal with its rival.
But BOSS is an interesting idea nonetheless. Yahoo hopes to attract both entrepreneurs and researchers--it has formal BOSS partnership with Stanford University, the Indian Institute of Technology in Bombay, the Massachusetts Institute of Technology, and other educational sites--and it's got more in-depth with some business partnerships the company plans to announce later.
Yahoo already has launched a program called SearchMonkey that lets programmers augment Yahoo's search results with richer displays of information--for example by adding starred reviews or addresses to restaurants listed in search results. That's a skin-deep change compared to what BOSS permits, though, where outside sites can completely alter the order of search results, filter out particular results, display results only of a particular variety, and combine the Yahoo data with internal data.
Yahoo offered some examples of what could be done with BOSS. One idea is a visual search presentation that shows miniature versions of the Web pages atop the textual results. Another, social search, could be used to spotlight results relevant to attributes drawn from a person's social network.
Setting up a competitive search requires prohibitively large financial resources, Raghavan said. Yahoo estimates roughly $300 million to cover expenses such as staff and the hardware to constantly index new Web pages, analyze the index, and handle queries.
Indeed, search start-up PowerSet, acquired last week by Microsoft, said the difficulty of building a full-scale search business led it to its acquisition by Microsoft last week. "Building a large-scale semantic search engine is expensive, requiring an engineering effort and computing resources beyond what most start-ups could ever imagine," said PowerSet product manager Mark Johnson in a blog posting.
With BOSS, Yahoo offers use of its own hardware in exchange for search-ad revenue. It might well be that no single company becomes as dominant as the top three search engines--Google, Yahoo, and Microsoft--but that collectively the smaller players offer more alternatives, Raghavan said.
Monetization of BOSS through search ads is "a few months away," he said.
Programmers may use the BOSS interface for free and, if they get to the stage where search ads are displayed, will share in the revenue. Yahoo doesn't reveal what fraction of search-ad revenue those outside partners would receive.
Yahoo is being careful with the BOSS branding. "You cannot put in any Yahoo attribution. This is not indicative of the Yahoo search product," said Bill Michaels, senior director of Yahoo's search platform.
It's not clear yet what the implications are for BOSS advertising. "Vertical" search sites that cater to a particular audience sometimes could in theory charge premium advertising rates, because they link advertisers to a targeted audience, but it's also possible that visitors aren't as desirable to advertisers than the broad cross section that visits general-purpose search sites.
"There are a bunch of unknowns," Raghavan said. "You might have more focused audiences, which could potentially be enhanced by proprietary data. But on the other hand, it's unclear when you fragment traffic what happens to the overall quality of traffic. You have factors that could weigh in either direction. It's premature to make a clear conclusion on the revenue per search metric."