Here’s a little Christmas present for all the media-buffs who read our blog – a colourful diagram from a simulation I ran of the distribution mechanism of viral marketing. The colour reflects the number of viewers the content ends up with, with Red being the most and Blue being the least.
(Click for larger version)
Explanation:
The x axis measures the quality of the viral content – i.e. the chance that someone will pass the content on to a friend when they see it.
The y axis measures a slightly abstract property of graphs. In this situation, given I could pass a viral to you, what is the probability that you could pass a viral to me?
The two extremes are Facebook and Bloggers:
On facebook there are only really “undirected” (two-way) connections – If you’re my friend then I’m your friend, and we both have the same ability to pass the content to the other person.
The Blogosphere is different – If I write a blog post (like this one) you can comment, but the vast majority of readers will not, so the connections are virtually all one-way (“directed”). I am far more likely to pass content to the readers of this blog than a random reader is to pass content back to me.
It turns out that this difference can be really important in the amount of traffic you can get from seeding a viral marketing campaign, but (in my model) there is a significant change in what this difference will mean for your campaign depending on the quality of the content.
Looking at the main plot again, it’s clear that there is a bifurcation in the results (commonly known as the “tipping-point”) half way along the x-axis (It’s only half-way because I have chosen the range on the x-axis to place it in the centre).
This is the point where many people will say the content has “gone viral” – although very little content ever reaches this level. It is close to the point where you expect every person that sees your content to pass it on to exactly one friend (see “Jamie Oliver starts to get viral marketing“.
This is often chosen as the point of “viral”, since it is a distinct point that can be predicted in models – even though it cannot possibly be measured in reality (perhaps that is also part of the reason…).
But notice how the behaviour changes at that point – on the left (where the vast majority of content lies) your content will do better if you focus on Blogger-like seeding, with one-way arrows. On the right your content may do better with facebook-style seeding.
More information:
This simulation was run assuming a population of 60,000,000 (approximately the size of the UK internet population), and with an initial seeding pool of 1,000 viewers. The colours shown are the predicted number of unique viewers of the content. For More information, contact Rubber Republic.
Tim
[...] (oh, BTW, My definition of “viral” was brought here) [...]
[...] (oh, BTW, My definition of “viral” was brought here) [...]
[...] they managed to create a fully connected social graph (which I’ve previously shown to be the most desirable form of social graph to “go [...]
[...] over a year ago, I ran some basic simulations of content “going viral” across various social [...]