What Google Can't Do

Search comes up short delivering video, pics, even text. Meet the startups trying to fix that.

Ever try typing your grandfather’s name into Google images? Unless he’s famous, you’ll probably find yourself staring at photos of people you don’t know.

But if Sweden’s Polar Rose lives up to its promise, it will soon become easy to find stills of family members and friends on the Net. Polar Rose, which emerged from stealth mode in December, uses 3D and face-recognition technology that it claims can single out Grandpa from other family members in a photo posted on a web site, even if he’s not directly facing the camera.

Before long, the technology could be incorporated in mobiles and combined with other services. Recognize someone at a conference? Just snap a photo, send it in, and within minutes receive details on the person’s identity.

If pundits are right, search will soon do a much better job finding multimedia and going mobile. It will also offer more accurate probes of local information, health information, gaming, music, and other areas of interest to consumers, as well as get better at finding the best deals in travel and real estate. No one is seriously challenging Google’s dominance. But Google, or even Yahoo and MSN, can’t do it all—so nimble startups across the world are seizing opportunities in emerging areas of search.

Take Russia's Yandex, which is expected to go public on the Nasdaq in 2007. It owns 60 percent of local-language desktop search in Russia, compared to 8 percent for Google, according to LiveInternet. Like Google, Yandex is branching out into mobile search. Its search engine was one of two just incorporated in the Russian version of Nokia’s N80 Internet-edition smart phone.

Irked that all the top engines were U.S. companies, French President Jacques Chirac announced plans last April to spend $2.5 billion on several technology projects—including $100 million for one dubbed Quaero to develop multimedia search engine technologies—declaring that Europe could not afford to cede the search space to the Americans.

Quaero, explains Francois Bourdoncle, CEO of French project member Exalead, is not trying to build a Google rival from scratch but is focusing on developing multimedia search technologies, an area U.S. giants are only just beginning to tackle seriously. Exalead and other private companies participating in the Quaero project (Latin for “I search”) will build on Quaero’s research to enhance their own commercial products including the already existing exalead.com Internet search engine.

<...>

The race is also on to improve local search, as well as local-language search. Google may be the world search leader, providing search in over 100 languages, but engines in Russia, China, the Middle East, and India claim, usually accurately, that their localized search engines can do better.

Yandex, for example, is already Russia's largest search portal, far surpassing Google in popularity. Profitable since 2002, it raised $5 million in venture capital in 2000 and has since received backing from different institutional funds and private investors, including Baring Vostok Capital Partners. Its CEO is Arkady Volozh, the ex-CEO and co-founder of CompTek International, Russia’s largest network and telecom equipment distributor.

Yandex, which now claims 23 million users per month, rose to prominence on the strength of its own search technology. Now it’s a full portal with its own mail service (powered by unique anti-spam technology), news clustering and aggregation service, blog search, free web hosting, research, shopping, and many other services—including its own payment system. And when it comes to Wi-Fi, Mr. Volozh points out that Google followed Yandex: it has long run one of the largest Wi-Fi networks of free hot spots in Russia. He declined to respond to questions concerning the company’s rumored upcoming listing on the Nasdaq.

Local-language search engines like Yandex fulfill an obvious need. Joerg Ueberla, a venture capitalist with Wellington Partners in Munich, points to a French study that compared search results for Google, Yahoo, MSN, Exalead, Voila, and Dir.com in the French language. They found that search engine entries overlapped in the best case in only 25 percent of the results they produced. “With results like these, this can’t be the end of search,” says Mr. Ueberla, who recently made an investment in a German search company which is still in stealth mode.

Another German search engine company, called Seekport, with offices in Europe, the Middle East, and India, is convinced localized search engines can do better. Next year, it plans to launch Arabic and Indian search engines.

The Arabic engine, called Sawafi, is being developed in Saudi Arabia to accommodate local spelling and cultural differences. Do a Google in English on the Koran and one of the top responses is the “Skeptic’s Annotated Quran,” a result many locals would find offensive and unrepresentative, says Stefan Karzauninkat, an executive at Seekport Internet Technologies in Hamburg. Certainly, Sawafi is on to a growing market. The size of the Arab world Internet advertising market is set to increase from $15.7 million in 2006 to $146.3 million by 2010, according to Madar Research.

In China, Baidu commands more than 60 percent of the Internet search market, according to CNNIC’s China Search Engine Market Survey, and in December it announced that it intends to enter the Japanese search market in 2007.

Baidu, in fact, was the inspiration for the founders of Guruji, a search engine launched in October to focus on serving up solid local information in India. Having raised $7 million from Sequoia Capital India, it is concentrating on local business listings and local articles, reviews, and blogs, and is among a clutch of other local-language search engines that have also launched in India recently. But none will find it easy to fight Google, which already offers search in five Indian languages, and is ranked the country’s No. 1 search engine.

Still, a quick search for yoga classes in Bangalore shows the need for sites like Guruji: On one recent search, Google and

Yahoo turned up a mix of articles and pointed to other web sites about yoga in Bangalore, whereas Guruji gave a straight listing of yoga institutes and instructors along with their addresses and phone numbers.

Hellen Omwando, consumer markets analyst with Forrester Research in Amsterdam, predicts there will be more collaboration between the large players and local engines. The small players need the traffic and clout, and the big ones need local know-how. “Generally, local markets are poorly served by the large search engines,” Ms. Omwando says. “Google has room for improvement in this area, as well as on language capabilities.”

Another problem with Big Search is that it isn’t optimized for deep crawls in vertical areas, an issue Silicon Valley-based Kosmix addresses. The 2006 Red Herring 100 winner helps consumers get information about health, video games, finance, travel, U.S. politics, and cars. The idea is that looking for answers in specific search categories avoids retrieving the huge number of irrelevant responses that pile up at Google. Specialty sites like the U.K.’s Zoomf, which focuses on properties for sale, do the crawling across different agents for surfers looking for the house with the balcony on the right street in the required price bracket.

Travel metasearch sites offer more complete searches for airfares, hotels, car rentals, and vacation packages. Sprice, the soon-to-be-launched brand name for Travel Meta Search, formed by the merger of Singapore’s Fare.net and France’s Coelis, is one example. The company hopes to fill a void in the fragmented online travel sector in Europe and Asia, and recently raised $10.2 million from Paris-based Sofinnova Partners and San Francisco-based Walden International.

A profusion of ticket agencies and hotel booking sites in China created a need for a site to allow comparison shopping—leading to the birth of two metasearch engine sites, Go1000 and Qunar.

While opportunities abound in expanding areas of search, the chance of a startup getting global traction over the giants is still slim, of course—unless their search engines produce such stellar results that people immediately notice after a few trial searches, says Wellington Partner’s Mr. Ueberla.

But if they are able to spew out better results, that just changes everything—a fact Google and the other giants would be foolish to forget, especially since they all got their start that way.

“A switch costs next to zero,” Mr. Ueberla says, “and is only a click away.”