Search's Social Exploration

Siri on a TVIn the latest version of Mars meets Venus, real-time media and social data are helping marketers discover the new Web 

Ask a search engine a question through keywords, and it returns well-organized information by interconnecting pieces of content stored across the Web. No easy task for companies building the search engine, such as Google and Microsoft. Aside from finding the correct information, there are ethical and privacy concerns at stake.

The engines might do a reasonably good job today at serving related content. But they will becoming increasingly sophisticated as more social signals support search algorithms and as alternative ways of searching, such as voice and even gesture-based queries, emerge.

And using the spoken word to search is a “tricky” challenge, says Jack Manzel, product management director of search at Google, due to the ambiguity of language. Human conversation is built on content. Voice search requires the understanding of natural language and elements of speech not critical to textual queries.

Google may be the first word in search, with Google sites taking 66.8 percent market share in July. That translates to 11.8 billion U.S. explicit core searches, up 3 percent for the month, according to comScore. But it isn’t the only company: Microsoft followed with 15.7 percent, up 0.1 percentage point, with 2.8 billion searches. Yahoo sites took 13.0 percent market share, with 2.3 billion. Ask Network accounted for 3.1 percent of explicit core searches, up 0.1 percentage points, with 548 million.

And Google was an early leader, developing voice search and speech recognition tools in-house, and using the shuttered goog-411 service (goog-411, shut down in 2010, was the initial version of Google’s voice recognition technology). The company combined its speech recognition expertise, understanding of language, and the Knowledge Graph, allowing Voice Search to better interpret questions and sometimes speak the answers back as full sentences. Voice Search sits on top of Google’s Knowledge Graph, which ties together information about relevant things, and search index turning voice search into text and text into speech. 

Then there’s Google Semantic Search, which the company calls the Google Knowledge Graph, a system that puts questions into context. From each search, the technology gains intelligence, based on answers to billions of queries. The technology will organize content, and pull in related topics from across the Web. Universal search will enable images, videos, maps, email from Gmail, and more to serve up on the same Web page.

People searching for information won’t need to open different Web pages to find airline flight times, prices, hotels, and meeting times. In the future, people who forget the time of a dinner reservation can opt to search online.  
While a push search model will slowly emerge based on location, Gmail content, historic search queries, and personal calendar information, Manzel says it’s not likely to happen any time soon without doing a physical search.
 
Search’s Push Model

Microsoft’s vision of search’s future differs a bit. Redmond, Wash., researchers Daniel Liebling, Paul Bennett, and Ryen White provide a glimpse into the future of search and the process of query-free anticipatory Web search. The paper, “Anticipatory Search: Using Context to Initiate Search,” published in August, identifies possible search content to determine ranking and recommendations.

Employing a log-based approach, the engineers compare methods for predicting potentially searched content based on pre-query context and behavioral signals. The research then predicts the url a user will click in a set of search results. Rather than waiting to receive the user’s query to generate the results, Microsoft uses the pre-search browse url combined with existing click-through data and/or the browsed page title to predict the needed content.

Users frequently click URLs that appear in the top results generated by this method, explain the paper’s authors. Since these searches and clicks represent a real need for information, the research suggests that prior information did not completely meet the need of the user.

Web managers could find or expand on the information across sites by using the urls to identify content their sites do not serve up. Search engines could leverage the pre-search context to present relevant urls without requiring that the user issue a query. In practice, large advertising and search companies already cover much of the browsed Web via contextual ad matching.

“People will no longer say, ‘I’m going to search for something online,’ similar to the way they once said, ‘I’m going online,’” says Stefan Weitz, director at Microsoft Bing.

As engines shift emphasis from keywords to natural language, content will serve up without the act of typing words into a search box. The Web originally consisted of linked documents and anchor text that sent signals to search engines through connected nouns. Real-time media and social data create the new Web. Unlike ads, which pull in demographic data, Weitz says search relies on actions.

Until now, people were removed from the equation; search relied on a mathematical algorithm. Weitz says the ability to store personal data in Internet-connected devices and online in Facebook, Foursquare, Google+, Yelp and Twitter changes the Web. “It’s no longer about a search engine optimization professional gaming the system to make sure his client’s restaurant appears at the top of the listing,” he says.

The “technical seo trick” taps collective wisdom through an information supply chain that connects trusted friends who, by the way, check into the nearby German Pub four times in two weeks.

The industry has begun to see search technology analyze context through natural language processing, connecting to calendars, location and other data on connected devices across desktops, mobile and television to access data. Voice search will become a tool, but Weitz calls today’s technology “a parlor trick,” because of ambient noise and homonyms. Voice will become one of many methodsm including vision and gesturem to find content.

The future will see a system smart enough to search on the consumer’s behalf, creating a push rather than pull model, and helping people do rather than find things. “I’m at the office at about 6 p.m. in downtown Seattle, and my calendar knows I have dinner at 7 p.m. on the east side,” Weitz says. “The virtual assistant sends an alert that says ‘Hey, Stefan, I just did a search on your behalf for Seattle traffic. You should leave in 10 minutes to get there on time.’”

Search will integrate with calendars from across devices, such as the Microsoft Kinect, connected to a 55-inch smart tv in a family room of a home that also allows consumers to verbally request a dinner reservation. Hundreds of thousands of servers support searches. Weitz says parts of each Bing search query can span up to 10,000 servers, and get consolidated before the information serves up within 30 or 40 milliseconds based on any word on the planet.

The engines pull data sources from across the Web, but Adam Riff, senior vice president at digital marketing agency MediaWhiz, agrees the industry will need cloud storage and faster bandwidth speeds to support connections.

Stephen Scarr, ceo of Info.com and Zenya, predicts that references for location, food, and prices pulled from a variety of data will cross-reference the request with booking information at restaurants to find, for example, a table for four at 7.p.m. The device will usher in the “death of the log-in,” by recognizing the user by voice and biometrics, and by tailoring results based on personal information stored across the Internet.

People become the “Infomediary,” Scarr says, pointing to a concept from John Hagel iii and Marc Singer’s 1999 book “Net Worth” (McKinsey and Co.). The theory suggests a string of connected content building an information supply chain accessible to search engines and applications that serve content based on dates, times and location.

Since consumer trust builds this information supply chain, Scarr believes search will rely more on the human psyche, and less on emerging innovative software and hardware. He says that to achieve a more relevant search experience, consumers must share lifestyle interests and need to assist search engines in determining syntax and location.

For example, a search engine query for “hp reviews” could mean Hewlett Packard, but boating enthusiasts would more likely want reviews on engine horsepower (hp). Geographic location compounds complexity. In England, Scarr says, users searching for “hp” might want hp for Hire Purchase information, or even hp Sauce, named after the Houses of Parliament.
Because trust plays an important part in such searches, Scarr doesn’t believe the industry will see change until Internet users feel comfortable sharing and updating personal information.

Although Google has hit privacy speed bumps before, the overall trend clearly points to personally identifiable signals, according to Will Critchlow, cofounder of seo agency Distilled. Social signals and authorship markup are tied to individuals; even search results are becoming personalized. “We have known for years links that weren’t placed by a person are not the ones Google really wants to count,” he says. 

Recent algorithm updates move the industry closer to Google’s objective, but Critchlow believes that personal connections will become Google’s key to advanced search.

How Will SEO Change?

seo professionals will likely need to evolve and adapt to changes in the next five years. Search engines Bing, Yahoo and Google will make it more difficult for search engine marketing specialists to control the placement of content in natural queries, but some believe links will still play a part in optimizing Web sites. The major search engines continue to shift from link development toward content marketing, creating on a natural flow of links and connections. Building quality content will generate links and prompt search-marketing agencies, like San Diego-based Covario, to move from link development toward content development and syndication.

Andrew Devine, senior seo strategist at Covario, says the industry’s position on link development needs to evolve, by privileging quality content rather than keyword stuffing to rank higher in the search engines. “Ten years ago you could pay for links or submit the site to questionable Web directories, resources with little value today, to see a strong impact on ranking, but now it’s becoming more difficult,” he says. “Google is getting better at determining the quality of content linking to your site.”

As search engines place increasing emphasis on social signals, WordStream founder and cto Larry Kim believes the value of seo, though still significant, will diminish, becoming less productive and more expensive. Social search means increasingly personalized search results. Google favors popular matter in a person’s network, as well as short-term search results generated by recent likes. Anything that Google does to make seo a more volatile lead or sales channel reduces the latter’s business value. More signals in the ranking formula means the coveted top organic search listing is harder and more time-consuming to attain.

Companies will likely need a full-time social media marketing professional to monitor and respond to chatter, grow the Facebook Fan base and Google+ and Twitter followers, as well as promote and share content on LinkedIn and Pinterest. Such a hire, in addition to the blogger, the link builder and the seo manager, will lead to a rise in overal costs, Kim estimates.

In the past, link-building tasks might have translated into the generation of directory submissions, page-rank sculpting, article marketing, and anchor-text optimization. Kim says that because many of these link-building tactics have recently fallen out of favor (and may even incur over-optimization penalties from Google), link-building has become very similar to Web marketing.

In the next five years people will find little distinction between general Web marketing and seo, according to Kim. “Think back to the late 1990s when the Internet was new and there was a big difference between ‘conventional’ and ‘Internet’ marketing,” he says. “Similar to the way marketing now includes the Web, I expect it will eventually just encompass elements of content creation, keyword research, social media, pr, and even online advertising, by default.”

Tags: search
Recommend (1) Print RSS