Comscore’s Fishy Facebook Numbers

by alec on July 6, 2007

Comscore published some web site usage metrics for Facebook yesterday.  While the aggregate visitor growth of 89%, year over year, from 2006 to 2007 was impressive, the demographic information was puzzling, and just doesn't add up. 

Facebook.com Demographic Profile

Unique Visitors  (000)
May 2007 vs. May 2006
Total U.S. – Home/Work/University Locations
Source: comScore Media Metrix  

 

 Age Segment

Facebook.com

May-06

(000)

May-07

(000)

Percent Change

Total Audience

Unique Visitors (000)

14,069

26,649

89%

    Persons: 12-17

1,628

4,060

149%

    Persons: 18-24

5,674

7,843

38%

    Persons: 25-34

1,114

3,134

181%

    Persons: 35+

5,247

10,412

98%

 

Facebook themselves published some data in February at the Facebook Tech and Tasting in Silicon Valley. I have transcribed from photographs of their slides into a table, reproduced below, and compared it to the Comscore data.  Clearly something is amiss.  For example, Comscore's May numbers appear to show substantially lower growth rates in the under 18 and 25-30 markets than Facebook's own data would suggest, as well as a substantial drop in the core 18-24 demographic. 

Taking into account that Facebook is measuring users, and Comscore visitors, some of the discrepancies can be explained away by the school year.  For instance, the fact that there were 12 million users in February in the 18-24 demographic, but just 7.8 million reported visitors in May can be explained by noting that the college year is over at the end of April.  Kids are moving, finding summer jobs, and so on.

In the under 18 category, and the 25 to 34 category, apparently robust growth in January and February isn't reflected in visitor statistics in May. A possible explanation is that a percentage of users sign up for Facebook, and then don't use it much.  Again, very plausible.  It's certainly the case in our household, where two out of four Facebook users are frequent visitors, and the other two never visit.

But in the 35+ demographic how does one go from 780 thousand users signed up in February to 10.4 million unique visitors in May? Moreover, how is it that Comscore can report over 5 million visitors were age 35 and older in May 2006, when Facebook claims only 55,000 signed up in that demographic in July 2006. 

And finally, Comscore's numbers don't add up.  Literally!  They're reporting 26,649,000 unique visitors, but their four categories add up to 25,449,00. Who are the phantom 1.2 million unique ageless visitors?

Something doesn't smell right.  Anybody from Facebook or Comscore care to comment?

{ 8 comments… read them below or add one }

Alec July 6, 2007 at 10:07 am

Hmmm… one would think that Facebook would police that then. After all, they do know ages when you sign up.

Reply

Rational Beaver July 6, 2007 at 12:24 pm

Great post. I kinda hoped someone would call out those numbers.

About the young ones: A. You can visit Facebook without creating an account (you can't do much, but still). B. Facebook is not lie-proof. C. Young kids may sometimes use their older siblings accounts.

About the old ones: Comscore uses a variety of means to measure internet use including surveys and some installed software (a toolbar, I believe). The age problem arises when Dad logs in to his Comscore account (a 35+ person) and then Junior (13-18) comes along and, instead of dutifully logging into his own account, he just heads straight to Facebook (or MySpace, which has also been reported to be ridiculously old).

Personally, I think this sort of demographic info is completely unreliable no matter what Comscore may claim about their methods.

Reply

Anonymous July 6, 2007 at 1:45 pm

A source at comScore alluded to the missing 1.2 million visitors being ‘outside the qualifying age range’ — which I took to mean below the age of 12.

Aside from the whole concept being ridiculous (9 and 10 year olds on Facebook? why not go to Whyville or Club Penguin), I think there are some strict federal guidelines governing the Internet activity of children below the age of 13.

Reply

Alec July 6, 2007 at 4:39 pm

Thanks for the background Rational. I’m with you in thinking that the demographic info is unreliable, especially when collected via toolbar.

Reply

Darren July 7, 2007 at 1:23 am

how about its people that hit the front page and don't go any further?

Reply

Alec July 7, 2007 at 3:23 am

If they're counting hits on the front page, Darren, then the data is completely worthless. It's hard to believe that they would be that stupid, or that they believe that the people using their data are that stupid.

Reply

Colin Smillie July 8, 2007 at 2:57 pm

The comScore and the whole panel approach is suspect. There has been a number of requests for outside auditing on these reporting services, the last one I can recall was from the IAB. You can see the comScore response here:
http://www.comscore.com/press/release.asp?press=1

Reply

ME July 9, 2007 at 2:54 am

how about that web measuring firms are useless?

Reply

Leave a Comment

{ 3 trackbacks }

Previous post:

Next post:

Alec on LinkedIn Alec on Twitter Alec on Facebook Calliflower on Youtube RSS Feed Contact me