Promoted posts and native monetization can be a bad idea if not done well.
We all know it. Social networking is yet to find a revenue model that is sustainable. Facebook seemed to have cracked the code when its advertising started churning billions of dollars in revenues but with the increasing shift to mobile, it seems like the wheel was invented around the time that people stopped using the roads.
So Twitter and Facebook have been trying a new revenue model: Attention Grabbing!
Attention grabbing is really important on the internet. The internet thrives on overabundance. While traditional media’s hallmark was scarcity (newspapers came once a day with limited content, TV had a finite number of channels), the internet’s hallmark is overabundance. With overabundance comes a greater fight to gain consumer attention. With social networks, this problem has only gone worse as MORE people are producing MORE content MORE often on social networks than on the rest of the web.
Clearly, anything that helps a content producer (read a tweeter, a scrapper, a wall-poster) to get more attention will be perceived as valuable by the content producer. So they might be willing to pay for it! Bingo, you have a revenue opportunity!
The risk, however, is that attention grabbing, if not done right, decreases the signal-to-noise ratio on the social network.
Given the overabundance of content on the web, consumers are attracted to websites, networks and forums with a high signal-to-noise ratio. The proportion of content that is relevant to me should be much higher than content that is irrelevant or useless. When noise starts increasing on a network, engagement goes down. e.g. Orkut started losing popularity because the site was gradually filling up with more noise (“Will you make fraandship with me?”) than signal (relevant scraps from friends).
The best method of gaining consumer attention, obviously, is one that is ratified by the market itself. Something that people talk about or recommend has to be a good signal.
Google identified signal by linkbacks (the Page Rank algorithm). With social networks (and especially Facebook), the offline idea of elevating content which was most recommended by people was for the first time fully implemented online through the share and the like function. The entire Facebook News Feed algorithm is designed around elevating content that is most relevant to you based on how your friends interact with it and how you interact with what your friends post.
Attention grabbing (through promoted posts and the like), if not done right, threatens to increase the noise in the system. The effect may not play out at first but, over time, noise leads to lower engagement and angry users.
For attention-grabbing to be done right, the following need to be considered:
1. Power on a social network should be a function of influence and activity. Quora hasn’t made it a revenue model yet but it has a much more relevant approach to attention grabbing. Quora allows you to gain credits with usage and then it lets you use these credits to promote your questions or answers. The key, though, is that the power to promote is not determined monetarily but is based on your existing influence and activity on the network and is, hence, more organic. Of course, this goes for a toss if they ever start selling credits but the current architecture optimizes well for a social network.
2. Artificial attention grabbers (read paid promotion) should not come in the way of promoting good content. A promoted post takes up real estate that would otherwise have gone to an organically elevated post. Where does one draw the line on that?
3. All attention grabbing should be relevant, and better still, personalized. LinkedIn is fundamentally more about one-on-one communication than about conversations around posts. On LinkedIn, attention grabbing is more about getting access to someone who might not be a direct connection. And this is where the Freemium aspect of LinkedIn works well. The ability to write an InMail to anyone is something people want to pay for. At the same time, it is targeted and personalized rather than mass spam.
4. Consumers should clearly be able to distinguish between organically elevated content and paid promoted content. Facebook advertising, for example, shows out separately as advertising but sponsored stories (almost) do not. If there are no upper limits to the number of sponsored stories in the feed, noise may become a problem. The same problem applies to promoted posts.
5. Attention grabbing mechanisms should self-optimize. Google Adwords is an attention grabber on the search results page but because of the auction mechanism and because the advertiser knows which keywords are performing and which aren’t, the ads that are shown get optimized and become more relevant with time.
6. Noise is defined differently for different media. Twitter and Facebook are fundamentally different forms of media. Twitter allows following without two-sided communication. Hence, there is a culture of receiving a thread of updates from people without any conversation element to it. Facebook is more about conversations and is a lot more participatory. Hence, even if promotion of content works on Twitter, it may still be noise on Facebook.
7. Noise is about the real estate too. Twitter has a standard format for all tweets, 140 characters in text. Facebook, on the other hand, allows image and video preview making some posts occupy larger real estate than others. This is a critical difference because, increasingly, the people who want to be heard more (read brands, self-promoters) are attaching photos to their posts to grab attention. I’m willing to bet that this problem will only get worse with promoted posts. I’m not saying Twitter’s got it right (quite the opposite), I’m just saying it might be a bigger problem for Facebook.
8. Monitor the signal-to-noise ratio: One way Facebook can make this work is by having a threshold on how often promoted content appears in a News Feed and how they change that frequency threshold based on whether the user interacts with the content or not.
After all, if the user interacts with it, it may not be noise anymore.
What do you guys think?
A framework to understand online community moderation across editorial, algorithmic and social mechanisms.