In Search of ... Search

June 25, 2007 — -- In the current market, there is perhaps no more heavily invested technology category than search, especially in a market with a dominant player, a dominant business model and an absolutely unbeatable capital advantage. Dozens and dozens of companies are chasing Google for one reason: Tens of millions of people search the Internet and still have a hard time finding what they are looking for.

Indeed, the Top 5 sites in the world are search sites (according to Alexa rankings), if you'll grant that YouTube, currently ranked No. 5, is effectively video search. Each of these businesses is ad supported and each would claim that it targets ads based on the interests -- as expressed in search terms -- of its audience.

It's the same story for nearly every search startup in the market today, and that may well be the Achilles heel. But more on that later.

In defining search as a "market," one needs to separate the technology platform from the expression of that technology to recognize that many "markets" are affected by new search methodologies. We see a concentration of activity in three areas: infrastructure, vertical and social.

Technologies in Search of Problems

We define "infrastructure" to mean fundamental search technology typically commercialized as middleware, a software appliance, or delivered as a specific application. Companies playing in the infrastructure sector might best be thought of as "applied search" technologies, where their search methodologies are applied to specific data sets, typically in large enterprise data centers.

That definition casts a wide net over a set of startups developing core search technologies. The French startup Exalead, for example, provides a uniform search metaphor across data sets from desktop to the Web. Reveal Technology Inc. has just launched a peer-to-peer search application to find documents across workgroup computers. X1 Technologies, an Idealab company, uses fast indexing technology to enable find-as-you-type search of desktop content. Seed-funded System One blends collaboration with search to deliver an enterprise information discovery engine. No doubt that is why we are seeing a resurgence of natural language technologies applied to information search.

Despite decades of fundamental research, the ability for algorithms to handle intention and nuance of meaning remains elusive on a mass scale. The companies playing in the natural language space are doing so in all manner of ways and have one core belief in common: Computers should understand and interact with people at a higher level. Some companies to watch include Powerset, Hakia, HeadCase, Radar Networks, TextDigger and Snap.

Chris Shipley is a leading technology and product analyst. As cofounder and chairman of market intelligence firm Guidewire Group, she analyzes emerging technology companies around the world to identify market opportunities and accelerate innovative products to market. As the executive producer and host of the DEMO Conferences for IDG Executive Forums and Network World, Chris has helped technology companies launch more than 1,500 new products since 1996.No matter the technical approach, the vast majority of the search infrastructure companies apply their technology to limited domains (a desktop hard drive or an enterprise server farm). In fact, we have only recently discovered an as-yet-unannounced company working on a very early stage project to index the Internet in a more scalable manner than that of Google, Yahoo, MSN and other leaders. It's too soon to tell if this company will be the next Google, but as the "informationsphere" expands, the sheer volume of content will demand that search technologies will have to scale by using more efficient algorithms, not more commodity servers.

These examples demonstrate the diversity of applications for search technology. Indeed, as businesses and consumers generate -- and save -- increasing amounts of content, be it structured data or rich media, existing search methodologies are challenged to keep up.

Vertical Is Horizontal

Working on known data sets makes the search problem more manageable, and that's why many new search companies focus on vertical markets. At Guidewire Group, we think "vertical search" has -- oddly enough -- both a vertical and a horizontal axis.

Along the vertical axis are subject areas -- health care, consumer electronics, shopping, etc. These vertical silos enable tighter taxonomies that act on specific data sets to deliver better results. In the last year, we've seen search and categorization technologies applied to consumer electronics (Retrevo), health care (Kosmix) and company information (ZoomInfo), among other vertical content areas.

Along the horizontal axis are data types -- photos, video, music, 3-D objects, and the like. These horizontal slices enable engineers to target specific attributes of the data type to hone search results. Sites such as YouTube and Flickr merge rich media sharing with searching, a necessity as their user-generated content databases become massive. We expect to see significant activity in vertical search that cuts across the Websphere.

BiggerBoat provides aggregated search for a range of media types. Blinkx and Dabble index video, Krugle programming code, Qloud music discovery, Boorah restaurant reviews, and Europe-based Tablefinder restaurants and restaurant reservations.

Smart algorithms, focused on clear taxonomies and specific data types are improving search results. Yet smart people can offer more intelligent processing than a thousand servers when the collective knowledge of the community is leveraged against an information problem. That's the basis of "social search," which might be more aptly labeled social information sharing. Social search approaches layer human insight over basic search methodologies to provide context, texture and ranking to information.

Chris Shipley is a leading technology and product analyst. As cofounder and chairman of market intelligence firm Guidewire Group, she analyzes emerging technology companies around the world to identify market opportunities and accelerate innovative products to market. As the executive producer and host of the DEMO Conferences for IDG Executive Forums and Network World, Chris has helped technology companies launch more than 1,500 new products since 1996.

Social search addresses the challenge of finding rather than searching. It is, effectively, about narrowing search results from the thousands or even millions of possibilities that a Google search might offer, for example, to the few really useful and vetted results recommended by fellow travelers. These social approaches also deliver serendipitous results that can't be attained from specific search terms applied to literal algorithms.

Some players are approaching social or contextual search from a vertical perspective, like Tourist Republic. This Europe-focused site simplifies the search for travel resources by aggregating specific content on this topic and then allowing users to share their own information with each other and rate the content. This adjusts how the content appears in future searches.

Vozavi uses crawling and semantic technology along with input from shopping "experts" to provide search for shoppers seeking product reviews. Yoono's social and contextual search engine enables users to find relevant sites that are based on their own previous online history as well as the collective set of preferred sites from others in the system who have also indexed their bookmarks. Aggregate Knowledge is a service that uses powerful analytics to crunch large data sets and provide personalized content recommendations on each page viewed by site visitors based on their preferences. Netherlands-based bliin takes GPS technology and integrates it with social networks and localized search, creating a system that gives people recommendations and information based not only on where they are but also on who else in their social network has been there.

Can the Ad Model Hold Up?

As more startups chase the search market, business models and market assumptions will be severely tested. Most search-focused startups are chasing the advertising market, an approach that seems logical enough. If one assumes that an individual's search terms are an explicit declaration of interest and intention, then ads can be surgically targeted to address those declarations. More highly targeted ads are more valuable, the theory goes, and that makes the search site that much more valuable to advertisers.

That theory, however, may find itself at odds with Web advertisers' continuing attraction to mass-market traffic. On the one hand, Web sites herald big traffic numbers to support handsome ad rates. Mass audience equals advertising reach equals efficient advertising spends. This is the old thinking that drives mass media advertising in print, broadcast, and online.

Chris Shipley is a leading technology and product analyst. As cofounder and chairman of market intelligence firm Guidewire Group, she analyzes emerging technology companies around the world to identify market opportunities and accelerate innovative products to market. As the executive producer and host of the DEMO Conferences for IDG Executive Forums and Network World, Chris has helped technology companies launch more than 1,500 new products since 1996.

On the other hand, Web advertisers also talk about traffic profiles and champion the "long-tail" theory to place a high value on a tightly targeted audience. Indeed, advances in search technology, local content filtering, and algorithmic and collaborative discovery mechanisms are all about fine-tuning the user experience and delivering to a specific individual exactly the right information at exactly the right moment. Most marketers would agree that this precise, motivated Market of One, where conversion is not just probable but likely is the most valuable ad spend. Today's technologies identify and deliver that Market of One, yet few sites have figured out how to calculate and collect on that value efficiently.

Meanwhile, we see a troubling reliance by early startups on a few ad networks, most notably Google's AdSense. While these ad networks provide a readily accessible revenue stream for search startups, they can also create a dependence on another company's business model that we believe builds uncontrollable risk into a startup's business plan. Call us old-fashioned, but we like businesses that ask real customers to pay real money for the value they receive. For this reason, the search infrastructure companies may have an advantage over Web-based search companies, particularly if the latter cannot monetize their value independently.

Chris Shipley is a leading technology and product analyst. As cofounder and chairman of market intelligence firm Guidewire Group, she analyzes emerging technology companies around the world to identify market opportunities and accelerate innovative products to market. As the executive producer and host of the DEMO Conferences for IDG Executive Forums and Network World, Chris has helped technology companies launch more than 1,500 new products since 1996.