Facebook Unveils Search Tool That Could Supply Intent Data for Ad Targeting

'Graph Search' Places Queries Could Also Challenge Local Stalwarts Yelp and Foursquare

By Published on .

Facebook lifted the veil this morning on its search tool, now in beta, which enables users to mine their social connections for specific pieces of information and could be a competitor to the likes of Yelp as well as a rich new source of data for ad targeting.

While the social network has long been a treasure trove of information for people looking to peer into their friends' lives, that information hasn't been easily searchable. For example, someone who meets an attractive girl at a friend's party would historically have had to go through the more time-consuming process of looking at the host's list of friends and searching for the corresponding name and photo. Now the girl could be more easily located by typing out a query for "single friends" of the host in a search box in the upper-left-hand corner that will appear on their Facebook pages.

Facebook graph search
Facebook graph search

In addition to benefits for the user in helping them navigate their social graph, Facebook might finally get a source of intent data for its ad targeting. Hypothetically, Facebook could add users who search for "maternity wear" to its advertising segment of expecting mothers, according to Chris Copeland, CEO of Group M Next.

"The more you get people expressing intent [in search], not only do you give them relevancy at that moment, but you can also serve them up later on Facebook," he said.

When announcing what Facebook is calling its "graph search," which currently just focuses on searches for people, pictures, interests (such as music and movies) and places, CEO Mark Zuckerberg underscored that the product is different from web search without mentioning Google by name. (To illustrate a hypothetical web search, the presentation actually used a fictitious search engine, SearchIT.com.)

Instead of taking users to links, "graph search is designed to take a precise query and give you the answer," he said.

Model queries ranged from the whimsical to practical searches for restaurants, a potential threat to tech companies such as Foursquare and Yelp whose ad models depend on location-based data.

Queries for people could be fairly granular, like "Friends in Palo Alto, CA who like 'Game of Thrones.'" Users will also be able to search for photos with specific friends, photos of friends taken in national parks (for example), and all photos they've "liked."

A hypothetical interest-related query was "Music liked by people who like Mitt Romney."

But it's with places queries that the threat to Facebook's competitors is most apparent. Some hypothetical queries that were highlighted include "dentists liked by my friends" and "restaurants in Palo Alto liked by my friends from India."

"Once the mobile component exists, I think it's very natural to want to query your friends first for a place to eat," Mr. Copeland said. "So I think that really will challenge the Foursquares and the Yelps over time."

He noted that those companies have the current advantage of richer location-based data sets, but that Facebook -- which will use signals such as likes and check-ins for places queries, according to its director of engineering Lars Rasmussen -- could potentially close that gap by urging its users to start rating restaurants and hotels. During his presentation, Mr. Zuckerberg acknowledged that Facebook is already making an effort to do this for restaurants to enrich its data.

"There's so many people on Facebook that you can get that signal very quickly," he said.

The products will be rolled out to a miniscule subset of Facebook users numbering in the hundreds or thousands today. Mr. Zuckerberg said that adding search functionality for mobile devices and expanding beyond English speakers were on the road map, as well as opening up an API for developers. He equivocated on advertising but certainly gave a nod toward the likelihood of it.

"This could potentially be a business over time, but for now we've really focused on building this user experience," he said.

The scope of Facebook's ambitious to index all posts and open-graph activities (which could include actions taken like listening to songs on Spotify or Rdio) hints at a major big-data project and also helps put the company's privacy overhaul last month in context. Facebook implemented a privacy button on users' toolbars to help them quickly identify what content linked to them is visible, and to whom.

Mr. Zuckerberg described graph search as "privacy aware" and emphasized that all content that can be discovered in search would otherwise be viewable to the user who's making the query. However, some degree of backlash seems inevitable.

"Facebook has certainly put all the right capability in for people to control their privacy, but what I don't think people have anticipated is that things you posted five or three years ago that may be embarrassing or not reflective of who you are right now can be turned up in search," said Debra Aho Williamson, eMarketer's principal analyst. "We'll probably see a mass unliking and untagging event as this new search feature rolls out."

Despite Mr. Zuckerberg's characterization of the product as being immensely different from traditional web search, Bing is a partner for it. Its results will be surfaced for queries with results that aren't currently indexed under the Facebook focus on places, people, interests and photos. (The example Mr. Zuckerberg gave was "Rihanna's newest album.")

For Bing -- which last year rolled out its own social-search product using Facebook data, which apparently got little traction -- this could be an opportunity to gain some market share.

"I think it actually may be an opportunity for Bing if this catches on with the Facebook power user -- the people who live on Facebook and almost never leave," said Kevin Lee, CEO of the search-marketing firm Didit.

Most Popular
In this article: