Social data is the nectar all brands want to drink, but tapping into the source can be a costly and arduous undertaking.
Consider Facebook and Twitter, the suppliers with the most scale to offer. They have drastically different approaches when it comes to meting out access to the millions of conversations occurring daily on their platforms. And in Twitter's case, the approach seems subject to constant change.
Twitter's "firehose" of tweets is already an important revenue stream for the company, and it takes a cut from sanctioned resellers that furnish raw data to enterprise customers. But it's also been looking to restrict the firehose access of existing partners. Facebook, meanwhile, has nothing resembling a firehose and keeps the majority of conversations taking place on its pages under wraps. Brands that want to know what's being said about them can use listening tools to tap into public posts that haven't been hidden by privacy settings, but no more.
"My standard canned response to clients is , "I can only see what Facebook lets me see, and that depends on each individual user's setting and what their API feels like giving me at the time,'" said GolinHarris Director of Insights Eric Swayne.
The social network has no agreements in place with data resellers, so in theory an individual who knows how to code can get just as much out of Facebook's data conduit—its Graph API—as an enterprise-level service. (API stands for application-programming interface; it's a set of rules that enables third parties to interact with platforms and services.) In practice, of course, the infrastructure that social-listening companies have built up makes them better equipped to handle the available data.
There's a broad consensus among marketers that Facebook furnishes rich data insights on a one-off basis to high-spending media partners. (Facebook declined to comment.) If and when the social network decides to make enhanced data insights into a product that advertisers can pay for—offering a view into how many mentions of a brand are trending across the network, for example—it could have a robust new revenue stream.
"It would be a pretty valuable tool, one that people would be willing to pay for," said Tim Fogarty, lead strategist for the social-media agency M80.
Twitter, on the other hand, is already realizing the value of its 400 million daily tweets. It has three authorized resellers—Gnip, DataSift and Topsy—which can provision companies that need Twitter data to fuel a service, like Gnip customer Klout, or listening platforms and business-intelligence firms.
As Twitter data have become more sought after, the company has moved to clamp down on access. Last fall it tried to cut out the social-analytics firm PeopleBrowsr, but the company won a temporary restraining order and then earlier this month prevailed over Twitter's attempt to have the case adjudicated in federal court.
An order by the U.S. District Court for the District of Northern California moving the proceedings back to state court reads: "The fact that the removal shortly followed the state court's issuance of a [temporary restraining order] suggests that Twitter's decision to move this case was born out of a desire to find a more sympathetic forum."
PeopleBrowsr claimed it had been spending upward of $1 million annually on the firehose and that pulls from Gnip or DataSift couldn't substitute for the full access its business was modeled on.
Twitter-data pulls from Gnip or DataSift can get expensive, but some agencies have little choice but to buy in. WCG Director of Analytics Chuck Hemann said his team is tasked with building custom platforms using social data, and pulling Twitter results from the Radian6 API now yields only links to tweets. Twitter is making the full body of tweets that are needed for sophisticated analytical work available exclusively through the resellers.
"It's definitely an expensive proposition," he said. "[But] it's becoming incumbent on agencies or corporate partners to develop relationships with a DataSift or a Gnip to get content in any meaningful form."
