Thursday, January 23, 2014

reports of Facebook's death greatly exaggerated

So, The Guardian reports on a study from Princeton University that claims Facebook is set to lose 80% of its users by 2017.

Indeed, according to the study, Facebooks's impending doom (sic) comes from 'comparing its growth curve to that of an infectious disease'.

While the twittersphere has erupted there's a couple of things to note about the article that might temper any death notions for the moment.

Firstly consider this quote 'John Cannarella and Joshua Spechler, from the US university's mechanical and aerospace engineering department, have based their prediction on the number of times Facebook is typed into Google as a search term.'

So things are already starting to smell a bit bogus.

Particularly when the report admits that 'The 870 million people using Facebook via their smartphones each month could explain the drop in Google searches – those looking to log on are no longer doing so by typing the word Facebook into Google.'

So that's 870 million from what Facebook report as a 1.2 billion monthly active user base, ie roughly 75%ish of users discounted from the off.

Secondly, read into the body of the article the claim is slightly less sensational 'Princeton forecast says it will lose 80% of its peak user base within the next three years'.

So, even if that prediction were to be true, dropping science on it a few rules come into play…

Retention double jeopardy:
All brands drop some users and the loss is proportionate to their market share (so big brands (i.e. Facebook) will certainly lose more but it’s a smaller proportion of their base.

Pareto law redux 60/20:
A bit more than half of Facebook's usage will come from the top 20% of users (peak user base) – the rest come from the bottom 80% therefore a drop of 80% peak user base is not a massive number, and will be balanced by….

Natural monopoly law:
Brands with more market share will continue to attract a greater proportion of light users.

And finally…

The law of buyer moderation:
Heavy users use less in the period after they were classified as heavy users and thus the flip may also probably be true of light users, they will use a bit more (i.e. regression to the mean) and some will become heavy users.

So, will Facebook be over in 2017?

Based on the data we have today, and barring any unexpected or random events in the near future that we can't possibly predict.



Facebook's own data scientists have published their own humorous rebuttal which includes this observation:

'In keeping with the scientific principle "correlation equals causation," our research unequivocally demonstrated that Princeton may be in danger of disappearing entirely.'

thanks Phil Sheard and Ciarán Norris for the heads up.

blog comments powered by Disqus