Facebook Rate Cards, and Social Networking CPMs

Nate Elliott | August 09, 2007, 07:44 PM

Valleywag has been posting some Facebook rate cards recently. While Valleywag is interested in the fact that Facebook has apparently doubled its sponsorship prices in four months; I find the CPM analysis endlessly more interesting.

The problem with social networks' revenue models has always been that although they generate a massive amount of inventory, that inventory sells for very low prices. So even though MySpace generates many more pageviews per month than Yahoo, Yahoo generates about 10 times more revenue. This is mostly because Yahoo's CPMs are multiples higher than MySpace's (even if Ross Levinsohn's ancient estimate has changed a bit). The vast majority of social networking inventory gets sent away to Adsense and other ad networks as remnant inventory, while a tiny fraction is sold by the social networks directly as banners or as part of sponsorships. Social networks’ ability to grow their revenues depends on two key factors: raising the CPMs of all that remnant inventory (optimization and behavioral targeting are crucial here), and selling more and more of their inventory directly to advertisers for relatively high prices (sponsorships are crucial here).

The part of the Facebook rate card excerpted by Valleywag doesn't list CPMs for individual placements bought directly from Facebook, like homepage sponsored stories, 'next step' text links, and skyscrapers. But it does say that rate card price for a sponsored group is $100,000/month, with a minimum three month buy, and that advertisers get 4 million skyscrapers and 12 million text links are part of the deal.

When an advertiser looks at that $100,000 a month, they think: this is the price I need to pay to engage young users. But regardless of whether the advertiser actually manages to engage users, Facebook knows that’s not what the advertiser’s really paying for. Facebook (and the other social networks) can’t guarantee how many friends or members or pageviews sponsored pages will generate, so it’s hard for them to price a sponsorship based on these pages. Instead of priced based on the sponsored pages, then, they effectively give away the sponsored pages for free and price the deals based on the banners (and other ad units) that drive users to those sponsored pages. The advertiser who buys a sponsorship may be buying ‘engagement,’ but they’re paying for banners – and so are helping the social networks reach their goal of selling more inventory directly. (Even better, they’re paying for banners that drive users further into the network rather than over to the advertiser's site, increasing the networks' already-huge time spent and pageview numbers.) This is nothing new, by the way: we used this model at the DoubleClick Network ten years ago to sell sponsorships on Dilbert.com and a dozen other sites.

When looking at the Facebook rate card, I'm going to assume the skyscrapers attached to their sponsorships are worth 3 times more than the text links. (Not because I think it's true, but because the math is easier this way and I'm already late to the biergarten.) Using that assumption, half the $100,000 per month goes to the 4 million skyscrapers, and half to the 12 million text links. That gives us a $12.50 CPM on the skyscrapers, and a $4.17 CPM on the text links. Blended together, it’s a $6.25 CPM. Those prices, even as theoretical rate card prices that’ll get negotiated down by clients, look a hell of a lot better than Levinsohn’s old $.10 CPM. If they manage to sell enough of those sponsorships, the networks might just live up to the insane expectations being placed on them.

(Niki has some interesting analysis of the other side of this equation -- pricing on the ad network/remnant side of things -- here.)

[Update: Card points to an interview with Fox Interactive's Chief Revenue Officer saying that demographic ad targeting is (finally) in beta on MySpace. I've never understood why this was so hard. Haven't more than 100 million people told MySpace, in clearly defined database fields, their age, gender, geographic location, university, job title and employer, whether they smoke or drink, and a dozen other things? C'mon, Yahoo's been targeting from their databases for a decade. In any case, it looks like advertisers will be able to target demographics soon, and that'll be a big help driving up prices on the ad network/mass reach side of the social networks' business.]