pat-sajak-twitter

Facebook is an advertiser’s dream.

On what other platform are users so ready and willing to provide their age, job, level of college education, marital status, and a host of other personal information that make advertisers lick their chops? Furthermore, Facebook’s sophisticated algorithm responds to user behavior such as “likes,” and then attempts to serve up ads accordingly. That’s really the end game of Facebook’s algorithmic tinkering — just look at what happened when Wired’s Mat Honan “liked” every single post in his feed for 48 hours. The social network became a wasteland of branded content. And the company only seems to be getting better at it, or at least that’s the perception from advertisers who keep buying its real estate — Last quarter, Facebook’s revenue grew by 61 percent, bolstered largely by the massive expansion of its mobile ad business.

But on Twitter, the advertising picture is far different. The company doesn’t ask for a whole lot of information up front so advertisers are more limited in how they target ads. That’s one reason Facebook’s ad revenue has been so impressive while Twitter is still working on getting profitable (it also doesn’t help that Facebook has four times as many users).

But a Nova Scotia-based company called Leadsift is launching a new product today that claims to bring the power of Facebook ad targeting to Twitter. It’s called the Audience Intent Miner, and it combs through millions of tweets in an attempt to determine the same kind of demographic information users willingly give Facebook. For example, if a user tweets all day about One Direction, Leadsift can make some reasonable assumptions about that person’s age, gender, and purchasing power to help advertisers launch paid campaigns on Twitter.

The other thing Leadsift does is unearth “intent to buy” in tweets. For example, the company could provide a list of accounts that tweeted about buying a car in the past week. Furthermore, it can divide those users in different stages of the buying process, from curiosity to research to those ready to pull the trigger.

“We extract over 100 attributes by modeling on a user,” Leadsift CEO Tukan Das told me over the phone. “Let’s say you check into an airport more than three times per month, we’ll automatically learn that you are a frequent flyer and a businessperson.”

While Leadsift is just now unveiling the product to the public, for the past six weeks it’s been testing the intent miner with clients belonging to Salesforce, Mindshare, and others — and Das claims the results are impressive. For “direct response” marketing (in other words, advertising where the user both finds and buys the product online), he says Leadsift helped drive “cost per action” or CPA down 46 percent. For advertisers like car companies, where the purchase is made offline, Das says time-spent by users who clicked on an ad increased five times over.

“Imagine you love a U2 song and you shared that, and you see an ad on your Twitter stream to download that song on iTunes,” Das says. “That’s how we see our algorithm helping our advertisers market on Twitter.”

It sounds like a compelling product for those companies looking to turn our social networks into giant canvases for ads. (Hey, we can complain about it all we want, but how else are these free services going to make money to keep the lights on?) But I wonder how accurate its “intent-based” ad targeting really is. For example, Google Ad Words constantly serves up ads for products I researched for a story, or other strange Internet ephemera I’ve searched for but have no interest in buying. It’s easy enough to ignore these banner ads — my brain has basically become hard-wired to do so. But once those ads start appearing in Twitter streams, they’re much more annoying and harder to ignore.

Of course, what if I’m wrong and it does work like gangbusters? That raises a whole other set of questions about what these algorithms can figure out about us. Even if we prefer a service like Twitter to Facebook because it requires less personal information upfront, advertisers (and governments) can still scrub our tweets to potentially learn just as much about us — and target us accordingly.

In any case, between the rise of more sophisticated ad targeting services like Leadsift, not to mention reports that Twitter may be experimenting with a Facebook-style algorithm, it’s beginning to feel like we’re exiting the golden age of social networks, when it was about conversations with other humans, and entering some strange corporatized new era, where the brands have taken over our favorite sites.

[illustration by Brad Jonas]