It's against Twitter's Terms of Service for outside parties to sell followers on the popular messaging service, but that's not just because Twitter wants all the revenues for themselves. Most "buy followers" services are completely untargeted and negatively impact the user experience on the network. This new feature may work out very well for all parties involved, including Twitter users. Early reaction on Twitter, however, is not positive.
According to Kafka, as well as to parties with knowledge of the arrangement that we've spoken to separately, recommendations will only be served up when they are relevant to a user. Exposure to Twitter users who are following sports-related accounts will be available for purchase by sports-related brands on Twitter, for example.
It's unclear how that relevance will be determined. Kafka says it will use the same "who to follow" algorithm that Twitter uses today to power free, and widely enjoyed, recommendations. But how will placement be bid on by advertisers? Is there some semantic analysis of user profiles being done in the background that will facilitate keyword auctions?
How granular will the options for placement be? It would be great if this was a Twitter advertising option that had a lower price of entry. Reports this morning confirmed that a sponsored trend starts at about $100,000. Could these recommendations target a smaller number of people at a lower price? How many city-level book-lovers would your local bookstore like to be followed by, if the price was right?
We hear that the placements will be bid on, and that the price per follower will be impacted by the success of the campaign.
Such an ad product would be very interesting for a number of reasons:
- Advertisers would be buying not placement for their commercial messages, but an opportunity for users to opt-in to subscribing to messages.
- The advertising would not be for one-off creative messages, but for a stream of syndicated communication.
- Users could unsubscribe at any time, so an advertiser might pay for eyeballs - but if they just push marketing and explicitly promotional content all the time, without any other value added, those eyeballs might leave in a hurry.
This newest form of advertising seems likely to prove interactive as well.
You might object that this is like the old Suggested Users List, just with paid placement - but targeted sponsored recommendations, alongside algorithmic long-tail recommendations of users in general, seems like a good combination to me.