Re: Social Network Metrics (reading for July 22 meeting)


This is an extensive topic and you can measure social networks (and  
their success) in many ways. Previous posts on this topic have  
discussed global popularity, and while this is useful for working out  
which are the world's biggest SNs, and the metrics mentioned (page  
views, unique visitors, time spent on service) generally make sense,  
it also makes sense for people running social networks to measure  
their own success by looking at the efficiency of the network. A  
social network about a niche topic such as beekeeping may be highly  
successful in its field, but not necessarily generate enormous page  
views in comparison with a more generalised site.

I've had a few discussions with people about taking a more economics- 
based view and looking at the "network effect" to measure SN  
efficiency/performance (see 
Network_effect for details, plus I recommend reading Beinhocker's book  
called "The Origin of Wealth" as a great introduction to network  
effect and evolutionary economics). This can result in a set of tools  
which allows administrators/developers to measure empirically the  
effect of service changes. For example, we could deploy the following  
useful metrics:

1. Absolute number of nodes in the network
This would most probably represent the number of unique members a  
social network has, and could of course be broken down into grades of  
member activity and other criteria. A node could also mean an  
individual piece of user- or dynamically-generated content. This  
metric would measure the absolute size of the network.

2. Average links per network node
When used to describe member behaviour, this would generally be used  
to tell you the average number of "friends" per member of your SN. The  
higher the average number of links per member, the greater the  
potential efficiency of the network.

3. Average link utilisation.
A key element in social network efficiency is building the maximum  
number of links between members and then setting up the network so  
that the greatest possible amount of content is shared across each of  
those links. For example, if member A sees one piece of member B's  
content a week, then you introduce an improvement to your service and  
member A now sees 10 pieces of member B's content a week, you can say  
your link utilisation has gone up. Of course, defining what exactly  
"content" is can be tricky, but this is generally a useful metric.

I think you can say that a "successful social network" has high  
average links per network node plus high link utilisation, and a  
"large successful social network" also has a high number of (active)  

So does it make sense adding similar metrics to our discussion or do  
we propose limiting discussion to absolute size?



On Jul 20, 2009, at 3:56 PM, Christine Perey wrote:

> Hello,
> I took an action item on July 15 to send some suggested reading to  
> the mailing list in support/preparation for July 22 meeting during  
> which the members are invited to discuss the topic of Social Network  
> metrics.
> Here is the position paper I submitted for the workshop
> I see them as quite different topics, however, there is overlap  
> between metrics and community segmentations.
> Here is some reading on the topic of segmentations:
> See thread which began with this message
> I look forward to a discussion on social network metrics.
> Regards,
> Christine
> Spime Wrangler
> mobile (Swiss): +41 79 436 68 69
> from US: +1 (617) 848 8159
> from anywhere (Skype): Christine_perey

Sam Critchley
VP, Products and Co-Founder
GyPSii HQ Amsterdam
Mobile: +31 6 28 233 133
Landline: +31 20 715 5915
Fax: +31 20 524 8896
Join the GyPSii mobile lifestyle
Download GyPSii to your phone
Latest GyPSii Place

Received on Monday, 20 July 2009 15:10:45 UTC