March 08, 2013, 9:27 PM — Facebook's Graph Search could be a powerful tool for steering people toward products and services on the social networking site, and marketers are starting to wrestle with the impact it may have on their brands.
Graph Search allows users to perform searches that let them uncover things their friends have shared with them on Facebook. It can also help them discover things like popular restaurants or shops in their area, based on what their friends have liked or talked about.
Marketers hope the results will help them make more connections with potential customers and become more visible on the site. But there are still a lot of unknowns that need to be addressed, speakers said Friday during a panel discussion at South by Southwest Interactive in Austin, Texas.
One issue is that users can "like" things for a variety of reasons, such as simply being polite, and can check in at places just because they happen to be there, not necessarily because they like the place. That could mislead marketers and other Facebook users about what is truly popular.
Other metrics, therefore, should be incorporated into Graph Search's results, they said.
"If the results aren't helpful because the only people coming up are the people getting the most Likes, it's not going to be helpful for anything," said Paul DeJarnatt, vice president and search director at Starcom USA, a marketing and branding agency.
Others said it's hard to use Graph Search as a way to assess their level of engagement with fans or customers, because the service is still in beta and at the moment only a small fraction of users have access to it -- on the order of hundreds of thousands - and some marketing professionals haven't been able to try it yet either.
"The first thing is we haven't had it, and the fact that it's in beta makes it almost nonexistent," said Natanya Anderson, director of social media at Whole Foods. "We have nothing to play with now."
For Graph Search to be effective in steering people toward brands and places, it first has to gain credibility among consumers and deliver useful results, speakers agreed.
One way to make the results more credible and useful, they suggested, might be by applying a more sophisticated scoring algorithm so that weaker results based on only one or two metrics are not displayed so prominently.
"Social input and recommendation has its limits," said Darren McColl, vice president of global brand strategy at SapientNitro. "As much as I love Whole Foods, I don't need to know 15 of my friends also love Whole Foods," he said.
"I would really like a measure that would help us say, 'This is relevant to me,' and a 'Like' button is not that," agreed Whole Foods' Anderson.