My former co-worker and friend, Andy wrote a great post yesterday regarding the monetization of social media. I highly recommend reading it. Click Here
The second post or article I read yesterday was Steve Rubel's "What Will Replace the Almighty Page View?" Steve breaks down what I've been thinking about for a while.
People have asked me before where I see Internet advertising going and I've told everyone I think the traditional model of cost per impressions will be challenged, by CPMPT (cost per impression, per time). The user behavior on social networking sites can be classified as high page view consumption - kids are clicking between pages.
Interesting content on a page - long time spent on the page - potentially long exposure of an ad unit.
Uninteresting content on a page - short time spent on the page - short exposure of an ad unit.
This combined with dynamic pages like Yahoo's mail product, you've blown up the traditional pageview.
From a direct response perspective it's easy to know if the right people are getting the right length of exposure to an ad. If response rates are inline with cost, the advertiser keeps spending. Layer on branding advertiser demand and you'll get a quick picture of the market price for the inventory. Of course measuring the branding impact takes longer and more effort.
Shall we get crazy for a minute?
Take Yahoo! Mail - They could set rules to update banners based purely on time. They could set rules to update the banners based on "events." Or they could...
Yahoo! can identify you, you logged in. They might even know your age. Either way they can profile your usage behavior of email. Do you quickly consume email? Are you a slow reader? (yes of course these are inferences from how quickly you consume events) Assuming different behavior affects your consumption and exposure of the ad placements.
Could they adjust their refresh rules by different consumption pattern segments of users? If they are running their own CPA campaigns they'd be able to adjust to the right level of exposure to maximize yield on time spent. (this breaks the maximizing RPM focus, notice I didn't say page view or event) Or they could experiment at what leads to the highest rate demanded by direct response advertisers. My gut feel is that brand impact is somewhat linked to direct response performance (assuming similar types of creatives - no punching monkeys vs traditional branding messages).
Back to reality -
CPMPT - Cost per thousand per time - Will the market move towards buying on time the banner is shown? Currently I don't know of an ad server that can handle that model. You'd probably have to target on some kind of average, as there would be breakage. Or would you have to identify the people who spent a lot of time on your site and target them for people buying long exposures, and save networks and direct response buyers for the short-exposure visitors.
Jumping further ahead, if these short exposure people jump from property to property who's best able to monetize them? My bet is the sophisticated networks. They'll adjust exposure across properties against each user.