A recent study on changes in eCPMs among differently-sized online publishers shows the poorest performance among its large-volume sample (read: social networks). Jason Calacanis drew a quick but clear conclusion: “social networks are great for traffic but horrible for advertising.”
It’s not difficult to understand why. The largest social networks, at the most basic level, group users on a contextless basis: simply that they know each other. If you look at your own Facebook friends list, you’ll see a few friends with whom you have lots of interests in common, and many more that you know through work, school or some other broad social vein. Most of your interaction with the latter is simple, fun banter and games online, and similarly fun banter and games offline. And the market for local online advertising is just starting to heat up.
You might have friends in which you have specific interests in common, whether it be cycling, international travel or chess, but there are themed social networks for just about everything nowadays. And, naturally, they have far stronger vertical contextual strength with which to demand higher CPMs (something Chris Anderson recently affirmed).
So what can large, general social networks do?
1. Funnel users into vertical common-interest groups. Facebook’s unrestricted policy towards group creation has led to thousands of groups with memberships in the hundreds of thousands to those with a handful of members. Each has a comment interest that provides greater contextual richness for targeted advertising and higher CPMs.
2. Assess topical interest and buying process stage from behavior. Facebook already does this through SocialAds–allowing advertisers to target topical interest–although more rich targeting is probably possible if a more granular assessment is made of where the user might be in a particular product/service’s buying process.
3. Track browsing behavior beyond the site’s walls. If the social network can not assess a user’s topical interests itself, it can work with services like Revenue Science with broad enough reach across the browseable Web to know a user’s interests and buying intent.
4. Provide advertisers with a platform to interact with current and potential customers. Service-intensive and high-ticket purchases could benefit from the human touch and personalized attention from the likes currently provided by Get Satisfaction and others.
5. Harvest recommendations from users and collate on searchable pages. Allow users to ask for recommendations, and recommenders to allow their suggestions to be public (i.e. not restricted to their friends). A large social networking site could collate all publicly-available recommendations on heavily-searched items into context-rich pages. (the Berkeley Parents Network–possibly inadvertently, considering the lack of ads–has done it here)
They key for social networks is to carve, parse and segment its user base according to interests through group and application functionality, behavior analysis, and better use of crowdsourced content. But we expect plenty of innovation to come as traffic continues to flood into social media.