The acquisition highlights two realities in today's web search marketplace - the shortcomings of keyword indexing and the considerable cost (and therefore challenge) any new entrant faces in gaining user share.

Important Details: On July 1st, Microsoft's Live Search team announced that it had acquired the natural language search engine Powerset. Terms of the deal were not disclosed, though various sources speculated that the price tag was in the range of $100 million range.

Powerset was founded in 2005 and received a round of funding in November 2006 of $12.5 million, which at the time valued the company at around $30 million. Powerset signed an exclusive license and collaboration agreement with Xerox's PARC in 2007 (see Insights 15 February 2007, Powerset: PARC Gives Credence To Natural Language Search) which enabled Powerset to use PARC's research in natural language technology in exchange for Powerset providing PARC with research funding, royalties and an equity stake. It is unclear how Microsoft's acquisition will change the terms of that deal.

Microsoft intends to leave the current operations of Powerset intact at its San Francisco locale. According to a posting on its blog, the staff at Powerset will continue to work on and operate its recently launched public beta site that provides search across Wikipedia articles and Freebase. However, it will also be focused on integrating its technology into Microsoft's Live Search offering.

Implications: From the start, Powerset's founders were not shy about touting natural language as the next great leap forward for web search. At first glance, the estimated price tag of $100 million may seem to indicate that their expectations for this approach lost its zeal - especially when you consider recent valuations around unmonetized (though well trafficked) social network sites. Those who doubt that natural language search will ever displace keyword indexing and page ranking algorithms as the de facto standard for web search (i.e. Google) have been quick to seize on this announcement as proof of the concept's limited use. But such an interpretation of this acquisition would be wrong.

For starters, Microsoft is paying $100 million for a company that is less than three years old, has no revenue, and a relatively small user base. Microsoft therefore clearly believes that Powerset is doing something right in terms of the technology it has developed. This price tag would be a hefty finders fee for the acquisition of a small but extremely well regarded pool of engineering talent.

Second, Powerset's decision not to go it alone probably has more to do with the lead Google has amassed in web search than the validity of its underlying approach. In today's market, displacing the current leader will take more than a great idea or great technology. It will also require a massive investment in computational power. Powerset's decision to launch its initial search service only across Wikipedia reflected in part its need to maintain lower computational cost. But by limiting its search base to this content pool, Powerset limited its application and user uptake. Given that this deal came just over a month after the site's launch, it appears that Powerset's expectations for user traffic did not materialize despite the site's well publicized beginnings.

Powerset's self-imposed limitation of content may also suggest that its search technology is not yet robust enough to function as a search engine across the open web, but that does not mean the approach is flawed. There are other engines out there using natural language processing in similar ways and some of them, like Hakia, are able to index the open web. One of these other search engines may yet succeed were Powerset did not.

So why Microsoft did buy Powerset? Regardless of whether its technology was ready for public trials or not, Powerset's work on natural language processing, augmented by Xerox PARC's considerable developments in the field, means that Microsoft is getting more than its money's worth in R&D alone. Outsell also believes that natural language search will play an increasingly important role in augmenting the performance of the general web search engines in the coming years - there is considerable room eof improvement in the relevance of today's search results.

Microsoft clearly believes this as well, and by incorporating Powerset's technology into Live Search it will be looking for improvements to gain market share. This technology will help Microsoft do that in two ways. First, it allows the meaning of complex queries to be deciphered in ways that keyword search approaches can not. This has benefits to ensuring a set of results that addresses the user's real question. Second, it allows the meaning of content to be understood more precisely than keyword indexing allows. This in turn means search results can be returned on relevance alone rather than using the system of ranking based on linkage which is more akin to identifying an individual's potential success via a high school popularity contest rather than a true and thorough examination of an individual's talent.

In light of recent events regarding Yahoo!, Microsoft clearly has the money and the intent to spend its way to success in online search. By funneling some of that towards building out the natural language technology acquired from Powerset, it may just succeed in creating a platform that is superior enough to gain share. The team at Powerset gets to keep working on the dream without the immediate worries of cash and time running out.

Related Insights

• Apple's Newsstand - A Shelf Publishers May Want to Avoid
• Netflix: How Big Data Can Lead To Bad Decisions
• A River Runs Through It...Amazon Starts to Lend Books Directly
• Hello Color My Old Friend - How A Korean Bookseller Just Outdid Apple and Amazon
• The Financial Times Sets the Pace in Digital Again