Peter Kafka

Recent Posts by Peter Kafka

Twitter Tries Cranking Up the Money Machine: More Precise Targeting = More Ad Dollars

Someone smart who doesn’t work at Twitter tells me that the company expects to do $350 million in revenue this year.

Here’s one of the ways they could get there: A new slug of ad dollars, generated by an overhaul of its ad platform.

We’ll go into details below, but here’s the summary:

*Twitter is effectively providing advertisers access to all of its 140 million users, via targeted “Promoted Tweet” ads. Prior to this, the only way you could shove a marketing message into someone’s stream was if they already followed your company, or if they followed someone who looked a lot like your company. Now those restrictions are gone.

*Twitter is also making it cheaper, at least potentially, to reach those 140 million users. Twitter sells its ads via an auction model, and used to require advertisers to bid at 50 cents per “engagement.” Now they’ve dropped the minimum to a penny.

Big picture: If advertisers like Twitter’s platform — and we don’t really know yet, because it’s so new — they now have many more ways to use it. Best-case scenario for Twitter is that this gets them closer to creating that elusive equivalent to Google’s AdWords, the most effective ad machine in history.

On to the details:

Lower prices: Straightforward, no? But for the record, Twitter product guy Kevin Weil wants us to know that this doesn’t mean Twitter will be flooded with bargain-basement ads for teeth whiteners. Twitter’s “promoted” product suite is only supposed to show ads that users are responding to, so simply winning an auction isn’t enough to get into the stream. Twitter’s preferred way to describe the move is that the lower prices will “drive greater ROI.” But an equally valid way to describe the move would be: A bunch of people who couldn’t afford Twitter ads now can.

“Enhanced interest targeting”: This is what’s going to allow advertisers to hawk their stuff more broadly on Twitter, and it’s what Twitter was referring to earlier this summer, when we got our hands on their ad pitch deck. The gist: Now advertisers can try targeting some 350 “interest categories,” as well as users who share “similar interests” with a specific Twitter account.

That allows more precision for pitches, but it also opens up a lot more territory.

Here’s how it would work: In the past, Starbucks could only send its “Promoted Tweets” to the 2.8 million people who were already following its own Twitter account, or to users that Twitter determined were “similar,” based on its own black-box algorithm.

But now Starbucks could pitch, say, technology fans, to tell them about its new Square tie-up. Or it could target people who resemble the people who follow Twitter and Square founder Jack Dorsey.

Twitter guesses which “interests” you have, or if you’re similar to the users who follow a given account, based primarily on the stuff you’re already following, and the stuff you retweet. Weil won’t go into detail about how it arrives at those conclusions, but notes that it’s already using similar logic when it makes “who to follow” suggestions on your home page, or when it tells you about particular stories on your “discover” page.

One important note: Twitter still isn’t responding to anything you actually type on Twitter yourself. That is, if you ask your Twitter pals for recommendations for a new backpack, Twitter can’t figure that out and serve up Tweets from backpack makers.

That makes it less effective than Google’s ad machine, which has users’ “intent” nailed. But, then again, no one has a machine like Google. And given that most Twitter users rarely or never Tweet at all, Twitter’s never going to work that way, anyway.

But figuring out what kind of stuff you like, based on the people and things you pay attention to, is one of the core tenets of social media marketing. Which no one has really cracked yet. If Twitter can do it, it’s a big deal.


Latest Video

View all videos »

Search »

First the NSA came for, well, jeez pretty much everybody’s data at this point, and I said nothing because wait how does this joke work

— Parker Higgins via Twitter