almost 9 years on 2008-06-01


this morning i had a wonderful conversation which got me thinking about the future of "social advertising".  it led me down the path of considering how one would theoretically build a scalable ad network on twitter, as an example of a highly open social platform with which it is particularly easy to interface.  what became clear is that the framework for constructing social advertising networks looks a lot like the framework for constructing a traditional ad network, the only difference is some of the hooks you are selling against...  thus, the four steps i propose below are not revolutionary or even all that new, but what is interesting is that they are newly possible (and even easy to deploy) in the context of social content, and many companies are already doing different pieces of the overall equation.

step #1:  acquire inventory from publishers =  (enable people to opt in to having tweets from advertisers inserted into the feeds in return for fees)

every ad network needs inventory.  in the traditional web, this meant previews and ad-units next to content.  in the context of twitter, this would mean the right to occasionally insert a tweet into a user's feed.  to make this functional, all one would need do is allow users to enter their twitter name and password (giving the ad network the ability to publish tweets on their behalf) and then have each user indicate an amazon fps account, paypal, etc to which they would like to receive advertising revenue.  it would be highly appealing to add a few toggles to allow users to  define how their feed can me monetized (specifying the frequency of promotional tweets they want, hours of the day they are willing to push ads, whether or not they are ok promoting no adult content, etc).  to my knowledge, no one is actually doing this yet, but there are services like twittertise which allow individuals or companies to schedule add content to run on a single feed.

step #2:  value and categorize your inventory = (pull core stats and samples of twitter feeds to quantify the reach of each feed, impact, etc)

ad networks need to understand the nature and value of their inventory to be able to match the inventory with advertisers.  this is where twitter gets really fun.  using the twitter api you can first get a sense of how much reach each user's feed has.  this can be accomplished in a rudimentary way by pulling the number of followers each feed has.  more sophisticated models of valuing reach would include measuring things like how many followers each of the user's followers has (the potential secondary impact), the propensity of those followers to re-tweet, or repeat information pushed out through the feed, etc.  you could even screen for location based tweets to screen for local demographics.

in terms of understanding the nature of the users (essentially, the content of the feed on which you have inventory) you only need look at word frequency on the feed.  ideally you would also look at the content which other followers who subscribe to the feed to understand the ultimate audience directly.    there are many many companies brushing against these these concepts, though no one is explicitly valuing a tweet to my knowledge.  summize (which was acquired by twitter) had a big piece of the equation in the form of a twitter search engine.  there are countless services out there that measure the reach of individuals on twitter (twittergrader being a questionable example) and create tag-clouds based on frequently used words.  the upshot, people are very close to explicitly attaching a dollar value to a tweet on a given feed.

step #3:  sell inventory to advertisers, quantify impact, and refine your model = (sell 140 character ads with links to offers, quantify click through rates and sales, help companies optimize tweets)

finally, ad networks need to sell their inventory to relevant advertisers, and help those advertisers refine their targeting and messaging to increase roi.  more and more advertisers are experimenting with twitter and how they can leverage the platform.  they are interested, they just need inventory and help figuring out what messages work.  to do this effectively with twitter you simply need a system of measuring the impact of your tweets in a way which you can meaningfully feed back into a self-refining model of how you are valuing feeds.  no one is after this yet, but a company like which gives some analytics on click through rates on links inserted into communications platforms like twitter are on the right path.  in the end, it is just about refining a model of matching types of feeds, with types of followers, with the exact content, time of day, context, etc to get the desired response.

tada.  a twitter ad network - abstractable to any social content

several caveats...

a.  brands are still scared of how to insert themselves into the conversation - is it inauthentic to push marketing tweets on someone feed?  not sure.  i am sure there is a learning process here, and probably the tweets need a 'sponsored' disclaimer - but the clear fact is that more and more of the consumed content in the world is 'user generated'  so advertisers need to figure out how to get their messages out on those terms.  interestingly, pownce, which recently decided to shut down - had from day one advocated the concept that on their free version advertisers could insert messages into people's feeds

b.  very few twitter users have enough valuable followers to make any money from allowing advertising on their feeds.  this seems absolutely true... only a tiny tiny number of individuals could even theoretically 'twitter for a living' - just as almost no one successfully blogs for a living.  that said, the more you know about your audience the less it is just about scale.  when, all of a sudden, you can actually quantify the value of reaching the specific people who follow your feed, your cpm might be off the chart even with very low absolute numbers of followers - so long as you have the right followers.

c.  twitter will just shut you down.  maybe.

d.  50 other concerns i haven't thought of.  poke away.  i am interested in what others think of this direction.  for me, this is more of a framework for thinking about what will happen to social media platforms, and the core nature of information... but maybe someone is actually building this right now and has some more insight.

note:  i have just been directed to a company called 'magpie' ( by js -- it does something close to this idea (how cool... investigating further)

original swl blogposts and letters 2007-2010