User:Stu/comScore data on Wikimedia/February 2009

February 2009 data

edit

During the month of February 2009, comScore estimates that 301 million unique visitors (UVs) viewed our projects from a personal computer, which it estimates was a "reach" of 27.9% of the 1.08 billion worldwide PC-based web browser audience:

Worldwide unique visitors
Google Sites (includes YouTube) 817 million
Microsoft Sites 675 million
Yahoo! Sites 578 million
Wikimedia Foundation Sites 301 million
AOL LLC 277 million
FACEBOOK.COM 276 million
eBay 235 million
CBS Corporation (includes CNET) 186 million
Amazon Sites 183 million
Fox Interactive Media (includes Myspace) 175 million

Facebook posted some remarkable growth numbers in February, adding 40 million users. Assuming that momentum continues, it should pass us and move into the number 4 spot within a month or two.

Geographic breakdown

edit

comScore estimates our audience in different regions, and also estimates what percentages of the audience within each region visited one of our sites:

Unique visitors Reach in region
Worldwide 300.8 million 27.9%
Europe 117.8 million 38.5%
Asia Pacific 70.9 million 16.4%
--India 6.6 million 20.0%
--China 2.0 million 1.1%
North America 65.9 million 35.5%
--United States 54.1 million 33.1%
Latin America 27.4 million 33.9%
Middle East - Africa 18.8 million 24.8%

This month comScore made some changes to its methodology around the Middle East - Africa region, raising its estimate of overall users and also treating Turkey separately. The net impact for us was an increase in the estimate of our visitors and a decline in reach in that region.

Language breakdown

edit

comScore estimates visitors to the different language versions of Wikipedia and estimates the unique visitors worldwide:

Worldwide unique visitors
English Wikipedia 156.0 million
Japanese Wikipedia 26.3 million
Spanish Wikipedia 26.2 million
German Wikipedia 21.8 million
French Wikipedia 21.5 million
Italian Wikipedia 9.8 million
Portuguese Wikipedia 9.3 million
Russian Wikipedia 8.3 million
Arabic Wikipedia 6.8 million
Vietnamese Wikipedia 4.4 million
Chinese language wikipedias 3.8 million
Korean Wikipedia 1.1 million
Indian language wikipedias .4 million

This month comScore updated its estimate of the overall internet audience in the Middle East and Africa. As a result, this month's estimates for visitors to the Arabic Wikipedia are significantly higher than previously. This doesn't reflect any sharp spike in usage, but rather is a one-time correction by comScore to catch up with the rapid growth in internet usage in the region.

Demographic breakdown

edit

comScore's panelists report age and sex so it can generate detailed demographic estimates, including raw data and also an index which measures the extent to which a set of visitors to our sites is over or under-represented compared to visitors to all sites on the internet. comScore estimates our 301 million audience is made up of 167 million men (28.7% of men online) and 134 million women (26.9% of women online). We index slightly higher with men (103) than with women (97). Here's a breakdown of different age groups:

Worldwide unique visitors Reach in
age group
Index
Ages 15-24 79 million 27.4% 98
Ages 25-34 69 million 24.2% 87
Ages 35-44 63 million 27.0% 97
Ages 45-54 51 million 33.1% 119
Ages 55+ 40 million 32.6% 117

We index highest for older users (ages 45-54 and 55+) and lowest for those 25-34 years old. I dug into this issue, at first thinking it was driven by twenty-something preference for YouTube and Facebook. As far as I can tell, though, our comparatively weak performance in the 25-34 year old demographic is the result of our weakness in China where comScore believes there is a huge audience in that age range. For example, Tencent, Baidu and SINA all index above 120 for this demo while Facebook, Google overall, and Yahoo are in the 90s while both MySpace and YouTube are with us down in the 80s.

Project breakdown

edit

comScore estimates our audience by project:

Worldwide unique visitors
Wikipedia 298.5 million
Wiktionary 7.7 million
Wikimedia Commons 4.9 million
Wikibooks 3.3 million
Wikiquote 2.5 million
Wikisource 2.2 million
Wikinews .5 million
Wikiversity .4 million
Wikispecies .1 million
edit

comScore estimates the unique visitors to our sites from home and office users in China (excluding Taiwan and Hong Kong). In July of 2008, comScore estimated 238 thousand UVs of our sites in China. In August, the month of the 2008 Beijing Olympics, comScore estimates we had 1.3 million visitors in China, for September 2.6 million, for November 2.3 million, and for February 2.0 million (1.5 million UVs to one of the Chinese language wikipedias and 0.5 million to the English Wikipedia). By contrast, comScore estimates the Baidu Encyclopedia had 38 million visitors from within China in February. Given that comScore does not track internet usage from public locations (e.g. internet cafes), these estimates certainly undercount overall activity from China.

In India, comScore estimates 6.6 million unique visitors came to our sites. Of these, 6.5 million visited the English Wikipedia while just over 0.1 million visited one of the different Indian language wikipedias.

Source of traffic

edit

comScore also provides analysis of the site a user surfs just prior to visiting us. The percentage of these "entries" from Google and other search engines is often used as an indicator of reliance on the search engines for traffic. Other major sites like YouTube, eBay or Facebook typically see entries from Google at 10% to 15% of their traffic while we are typically over 50%. Here's a breakdown of the top 4 for us:

Entries % of total entries
Google Sites (includes YouTube) 1,247 million 57.4%
Yahoo! Sites 131 million 6.0%
Microsoft Sites 91 million 4.2%
Logon 23 million 1.1%

Participation estimates

edit

I wanted a sense of what percentage of our audience actively participates. comScore gives good data on unique visitors, and Erik Zachte and others compile counts of registered users who have made at least five edits in a month, which seems a reasonable threshold for active participation as it would eliminate some casual or accidental editors. With data coming from two different data sources it's a bit apples-and-oranges, but is still useful.

The table below shows the calculations for the biggest few Wikipedias. Due to the size of the English Wikipedia, editor compilations happen infrequently so the most recent data covers September of 2008. On the English Wikipedia only about .03% of the unique visitors actively edit. Put another way, that's less than one-third of one-tenth of one percent. If you include all users who made at least one edit, it's about triple that amount or just under .1%.

Sep '08 UVs from comScore Sep '08 editors with 5+ edits % of UVs
with 5+ edits
English Wikipedia 140,710,255 41,393 0.029%
Japanese Wikipedia 25,698,145 4,390 0.017%
Spanish Wikipedia 25,388,063 4,016 0.016%
German Wikipedia 20,435,314 7,144 0.035%
French Wikipedia 16,428,023 4,602 0.028%
Portugese Wikipedia 10,787,686 1,710 0.016%
Italian Wikipedia 8,637,544 3,208 0.037%
Russian Wikipedia 6,534,903 2,672 0.041%
Source: UV stats from comScore, editor stats for English Wikipedia from http://en.wikipedia.org/wiki/Wikipedia:Editing_frequency, stats for other wikipedias from http://stats.wikimedia.org/EN/TablesWikipediansEditsGt5.htm

Portal usage

edit

We worked with comScore to include estimates on usage of the Wikipedia portal at www.wikipedia.org. There's a wide range of usage across geographies. Here are the numbers for February:

Unique visitors overall Unique visitors to portal % of UVs
using portal
Worldwide 300.8 million 16.7 million 5.5%
Europe 117.8 million 4.3 million 3.6%
Asia Pacific 70.9 million 4.0 million 5.7%
North America 65.9 million 6.1 million 9.3%
Latin America 27.4 million 1.1 million 4.1%
Middle East - Africa 18.8 million 1.1 million 5.7%

Trend data

edit

I've put together a PDF of comScore's estimates of monthly unique visitors to Wikimedia Foundation Sites from Sep 2007 through Feb 2009. Contact me at stu wikimedia.org if you'd like a worksheet with the underlying data.