Twitter's Hidden Revenue Stream: How The Social Media Firm Plans To Turn Every Tweet We Post And Everything We Reveal About Ourselves Into Big Profits

When Twitter Inc. recently filed for its initial public offering with the U.S. Securities and Exchange Commission, a lot of space in its Form S-1 was dedicated to demonstrating the company’s successful advertising platform, which isn’t surprising given that advertising accounted for roughly 85 percent of its $316.9 million in revenue in 2012.

Twitter kept the reporting about the other 15 percent of its sales, or $47.5 million, discreetly tucked away within the S-1. That money is generated by the licensing of its massive quantity of data to analytics companies -- a side business that, although not as sexy or profitable as advertising, has the potential to grow into a huge revenue stream for Twitter.

Since the first tweet was sent on March 21, 2006, by Twitter co-founder Jack Dorsey, also the co-founder and CEO of the privately held Square Inc., Twitter users have posted more than 300 billion messages of 140 or fewer characters. Those tweets have been about literally everything, and each one contains valuable information about the user’s interests, physical location and when the tweet was sent.

Think about it: Millions of people publicly announce their personal interests in succinct detail, frequently with links to a diverse array of other people with their own specific interests, including their own followings and followers and a record of when and where the pronouncements were made. And it’s all presented in brief and often cross-referenced summaries.

Access to that kind of data was once a fantasy for advertisers and market researchers. The downside is that the rapid-fire pace of Twitter makes trying to ingest that information like drinking from a fire hose. To turn the data stream into something useful, an entire industry dedicated to monitoring and analyzing social media was born. The privately held International Data Corp. told the Wall Street Journal that it estimates the so-called big-data industry is growing seven times faster than the overall information-technology sector, and could be worth as much as $16.9 billion in two years. To put this figure in perspective, it is almost $1 billion more than the blockbuster Facebook Inc. (NASDAQ:FB) IPO generated last year.

There are dozens of social-listening companies, but only four have been labeled by Twitter as “official data resellers.” These four companies -- San Francisco-based Topsy Labs Inc., Boulder, Colo.-based Gnip Inc., the Tokyo-based NTT Data Corp. (TYO:9613) and the London-based DataSift unit of MediaSift Ltd. -- have direct access to every single tweet that has ever come through the data fire hose and have developed unique algorithms to turn that information into usable data.

Topsy developed an algorithm that automatically compiles and indexes every tweet, similar to the way Google indexes websites. The tweets can be organized by text, pictures or videos, links, Twitter accounts mentioned, hashtags, time of tweet and user information such a the Twitter handle, location, what accounts that user follows and who follows that user. Topsy also has data linguistic algorithms to detect sentiment toward a product.

“Topsy sells that information to other companies for the ability to search through Twitter for anything,” a Topsy executive said. “[Brands can] sort by volume, share of voice and the exposure that a tweet or topic gets.”

So if a movie studio wanted to gauge public reaction to an announcement that Ben Affleck would star in the next Batman film, it could use a firm such as Topsy to see how many people on Twitter are tweeting about Batman or Ben Affleck. It can see where those people are located, what sorts of accounts they follow and who follows them. Most important, the studio could see whether the tweets tend to be positive or negative and then adjust its marketing strategy accordingly. These social-listening companies work with newsrooms to help them understand the real-time impacts of news events, show politicians which issues are most important to make their campaigns more effective, and identify market trends and important events for financial institutions to get a leg up on trading. Even nonprofit organizations such as the Red Cross are using analysis of the Twitter stream to more effectively deliver aid. The companies can also pull data from Facebook, its Instagram subsidiary and the Tumblr unit of Yahoo Inc. (NASDAQ:YHOO), but Twitter still leads the way when it comes to real-time information.

Television networks are likewise heavily invested in analyzing the Twitter data stream. Once worried that social media took fans’ attention away from commercials, networks now encourage fans to chat about the shows using official hashtags. A network such as CBS can then use an analytics service to see how many people are tweeting about “How I Met Your Mother” versus “2 Broke Girls,” and whether those viewers are saying positive or negative things, while uncovering information about their ages, locations and interests. Nielsen ratings could never provide that sort of detail about demographics, and it helps CBS adjust its advertising as it delivers targeted commercial messages across an array of platforms, including Twitter.

In its IPO filing, Twitter acknowledged the usefulness of the data, albeit in understated terms. “We provide our data partners with the tools and data to find the right signal for the right audience,” the company noted. “Our users tweet to express their thoughts and opinions about what is happening around them, creating data that can be analyzed to identify trends and other valuable insights.”

In the first six months of 2013, data licensing generated $32.2 million for Twitter, a 53 percent increase compared with the same period in 2012. Although Twitter said the company expects data-licensing revenue to decrease as a percentage of its total revenue over time, that will be more the result of ballooning advertising revenue than from Twitter shrinking this portion of the business.

So why isn’t Twitter making more of a show of it with its IPO? There are many potential reasons. Data reselling isn’t as publicly alluring, doesn’t make nearly as much money as the advertising department makes and involves a chain of sales rather than just direct revenue. It could also be a deliberate public-relations move by Twitter in the wake of Edward Snowden’s leaks and growing concerns about digital privacy.

“You go down a slippery slope the moment someone talks about ‘my data being used,’ even,” said Arvind Malhotra, a professor of strategy and entrepreneurship at the University of North Carolina who studies how brands successfully leverage social media.

Malhotra indicated he sees the two revenue streams of Twitter as more interrelated than separate. Analytics are essential to good advertising, and the data-licensing business is what’s happening behind the scenes to create a strong advertising platform. By keeping it in the background, even in its IPO filing, Twitter can highlight the sexier, more profitable revenue stream to investors.

“It’s the plumbing, a means to an end,” Malhotra said. “The goal is to accurately target messages to people. [Twitter] avoids the negative backlash and highlights the larger revenue-producing product.”

In fact, data licensing has several pitfalls beyond the potential for a public backlash. The Wall Street Journal reported that 75 percent of the information shared on Twitter comes from Americans and that 30 percent of the data comes from people under the age of 30, which means the demographics are limited. Twitter also has a problem with spammers, bots, fake Twitter accounts and hackers, all of which contribute their share of incorrect information. These misrepresents the number of users on Twitter and disrupt the social-listening algorithms designed to compute demographics, the popularity of a message and user sentiment. Such pitfalls may be another reason Twitter is attempting to play its data-licensing hand close to its vest.

Still, as long as Twitter can continue to grow its user base, there is no question the data those users generate will continue to become more valuable, and there will be no shortage of businesses looking to condense the data fire hose into the equivalent of 140-characters-or-less of extremely useful marketing information.