Facebook will soon embark on a pre-IPO "road show" that will trumpet its revenue potential by hyping the company's chief asset: the data it possesses on its 845 million users. But a crop of startups mining social data assert that they can go the social network one better by targeting ads across the web, not just inside a black box.
Their pitches revolve around the notion that ad targeting on Facebook, using demographic and even interest-level data, tends to be a blunt instrument because that data isn't paired with other information from the internet. And with private Facebook information off-limits to startups in the space, many look to vendors that power "share" on publisher websites as a source of data signals.
But the efficiency of these social startups is still a big question for marketers obsessed with tapping into interactions that occur within Facebook, but less so about those that happen within obscure blogging platforms.
"I can guarantee you that in every one of their business-plan meetings, someone says, "As long as Facebook doesn't start offering this, we're in business,' " said Craig Atkinson, chief digital officer at PHD.
Mr. Atkinson noted that many companies he's encountered have social-retargeting offerings, which let a company like Purina target ads to users who had mentioned one of its brands in an online dog-lovers forum, and a "social look-alike acquisition" feature, where the goal is to target prospects who look similar to your existing social fan base.
The latter in particular doesn't seem scalable, as Facebook isn't part of the mix, according to Mr. Atkinson. "Frequently we've had clients test these services, and their reaction is , "Let's do it at 10 times the scale,' " he said.
The industry of social-data mining is mature enough to have some veterans, one of which is 33Across, founded in 2007. It has 85 employees and has raised $13 million in funding. It recently acquired Tynt, a tool that lets publishers see when their content is copied and pasted anywhere on the web.
33Across claimed to have data on 1.25 billion web users that it can mine to find brand affinities based on social connections and interest similarities with users classified as "brand loyalists." Clients include AT&T, American Express, eBay and Macy's .
Tynt brought 1 billion of those users, and the balance came from sources like content-sharing and messaging platforms, Twitter apps and photo-sharing sites. The company can deliver ads with inventory from exchanges and real-time bidding platforms.
While 33Across buys from many sources, Meteor Solutions obtains its data from brands and publishers. Launched in 2009, it works with demand-side platforms to deliver ads to audience segments developed based on how content has been shared. Working with a telecom business, for example, Meteor would embed some code in a web page with a product video, designed to be engaging, that introduces a phone, and would then track how and where it's shared to inform the advertising.
"It's different from retargeting in the sense we're not looking at users who looked and didn't buy," said Meteor CEO Ben Straley. The Seattle-based company has 15 employees and is profitable, with a client roster including Microsoft, AT&T and NBC Universal, according to Mr. Straley. "What we're focused on is looking at the user who's taken a specific action and shared -- someone you should always be thinking about as part of your media plan," he said.
One of the companies most widely associated with social targeting has publicly disavowed that part of its business. Media6Degrees has $24 million in funding and clients such as American Express, Disney and AT&T.
It was founded in 2007 on the premise that people who act alike online are apt to have similar brand preferences, but the thinking originally was that the strongest signals should come from social media, according to CEO Tom Phillips. It initially bought data about MySpace users from apps operating on the platform (Facebook data wasn't available), but exchanges and real-time bidding platforms have since become the primary data source.
Media6Degrees officially recast its business from "social targeting" to "prospect targeting" last year.
"We saw that data we got from the broader web was richer in its signal than the social data," said Mr. Phillips, who gave the hypothetical example of a cruise-line client to show how the company now works. Its algorithm might say that visiting a travel-research site is a strong signal of brand intent, and then rate those users highly and bid on them in the exchanges.
Might Facebook one day launch an external ad network?
While such a product could be in the long-term road map, Facebook might be hesitant to proceed because of the trade-off involved, according to Wunderman Senior Digital Strategist Ben Bloom.