Online Influence : Aiming Big with the Big Shots

Just a quick update to present you a fascinating study I read a few days ago on online influence.

There are some very common prejudices when it comes to online communication strategies. One of them is that in order to reach a large audience, one should identify the websites which have the largest traffic and use them as potential relays to disseminate and broadcast information.

Most people involved in online communication and conversational marketing know that this is completely erroneous: if you are seeking to disseminate information in an effective way on the Internet, you need to identify first relevant communities and relays that will eventually support your marketing campaign and the message you are ultimately trying to convey.

Information dissemination obeys to specific rules, and scientific studies are slowly emerging and highlighting these rules.

Jure Leskovec

Jaegon Yang and Jure Leskovec, two scholars at Stanford University, identify in one of their papers the mechanics of information dissemination online. Yang and Leskovec used for their research a huge corpus, consisting of 580 million tweets and 170 million articles (whether from blogs or news sites). Yang and Leskovec identified the 1000 most recurrent phrases in their corpus and monitored the way these phrases were disseminated on the Internet, taking into account 5 different kinds of relays :

–          Professional blog (P) ;

–          Newspapers (N) ;

–          News agency (A) ;

–          Television (T) ;

–          Blog aggregator (B).

To sum it up quickly, Yang and Leskovec distinguish 6 main models of information dissemination online :

The models of information dissemination online

(the Y-axis indicates the actual volume of traffic)

This chart tells us two things :

1. The information conveyed through mass media (clusters C1, C2 and C3) is seen by a lot of people, but on a very short time span. In less than 20 hours, the information has almost vanished from the Internet since almost no Internet user shared it or published it (on their blog, board, website, etc).

2. The information first conveyed by blogs (clusters C4, C5 and C6) is initially seen by fewer people (when compared to C1, C2 and C3), but is overall viewed by a larger audience for a longer time, thanks to the fact that Internet users shared it and/or republished it.

So, there you have it, QED, the proof that the structure of social networks (friends, collegues, anonymous who share common interests) has far more influence than the actual size of the audience exposed to the information. A network with limited audience, but consisting of relevant, dense and strong connections is likely to be more receptive to recommendations emanating from one of the members of the network, and the information which has been recommended is likely to be disseminated more effectively.

I strongly encourage you to check the complete study, which is available online at : http://cs.stanford.edu/people/jure/pubs/memeshapes-wsdm11.pdf