Wikimedia penetration

Wikimedia penetration by country (i.e. percent of internet users within the country who use Wikimedia projects)

One way to think about expanding reach is to increase the number of Internet users who use Wikimedia projects. By examining Wikipedia, the most popular project, it is clear that there is ample opportunity to increase the number of Wikipedia users simply by expanding its use amongst people already online. The following table provides estimates for the percentage of the online population using Wikipedia for all regions of the world and selected countries. This table shows that even amongst countries such as the United States, Germany, and Japan whose population speaks the languages of the three largest Wikipedias and serve as home to many of the most active Wikipedia community members, there is still some room for expansion. However, the greatest opportunities for growth are amongst Asian countries such as China, Korea, and India where below 10% of the online population is currently using Wikipedia. An estimate of the total global population that uses Wikipedia is between 17 and 25%.

Estimate of Internet users who use Wikipedia, by region and selected countries.^[1]
Region	Country	Estimated range of Internet users who use Wikipedia
Asia & Pacific		10-16%
	Australia	21-27%
	China	1%
	India	8- 20%
	Japan	26-38%
	South Korea	5-7%
	Taiwan	10-13%
Middle East & Africa		12-30%
	South Africa	14%
Europe		26-37%
	France	36-39% *
	Germany	25-42%
	Russian Federation	18%
	United Kingdom	28-37%
Latin America		15-30%
	Argentina	22-26%
	Brazil	9-22%
	Mexico	19-35%
North America
	Canada	45-50%
	United States	24-32%

Internet access speed

See also Regional bandwidth

Participation is more likely in countries with higher broadband adoption and higher broadband speeds. Dial-up internet users can't use Wikimedia projects a lot and are less likely to have enough (connection) time to contribute.

Akamai State of the Internet Report 2010Q1 figures, figure 24 (Average and average maximum connection speed, average megabytes downloaded per month by mobile provider).

Wikimedia penetration by language

The following table displays the top 11 languages by number of native speakers and the number of Wikipedia articles in that language. It also shows the number of language speakers per Wikipedia article. This is one possible way to measure penetration by language, with higher numbers meaning that there is less content available per speaker. It is clear from this table that some of the most widely spoken languages such as Arabic, and Hindi and Bengali have significantly smaller Wikipedias.

Number of native language speakers and number of Wikipedia articles. ^[2]
Language	Number of Speakers (millions)	Number of Articles (thousands)	Number of speakers/Number of articles
Chinese	1'213	255	4,757
Spanish	329	440	748
English	328	3'048	112
Arabic	221	99	2,232
Hindi	182	34	5,353
Bengali	181	20	9,050
Portuguese	178	496	359
Russian	144	440	327
Japanese	122	603	202
German	90	925	97
Javanese	84	108	778

Articles per number of speakers

The following chart provides another take on the penetration of Wikipedia by language.

The main resource to create Wikipedia articles are speakers of the language. Drivers to contribute might differ per culture or language.

In the table below are listed the Wikipedia in descending order of number of articles per thousand speakers of the language. This table might change your view on the reach and participation. Included are languages with 20 million speakers or more. Also included are Wikipedias with more than one hundred thousand articles. Please note that English is not within the top 10 of this list.

Warning: The number of speakers are rough estimates of vernacular language speakers (secondary language speakers have apparently been ignored). The effective source of data should be retrieved from an accurate and maintained source, such as the Unicode CLDR extra data, which also includes data about the literacy level (in order to estimate the number of readers of the language).

Language	Wikipedia	Speakers (million)	Articles (thousand)	Articles per speaker (thousandth)
Volapük	vo	000.00003	0,120	4,000,000.0
Esperanto	eo	002.00000	0,117	0,000,058.5
norsk bokmål	nb	005.00000	0,225	0,000,045.0
svenska	sv	009.00000	0,325	0,000,036.1
suomi	fi	006.00000	0,213	0,000,035.5
Nederlands	nl	020.00000	0,553	0,000,027.7
català	ca	007.00000	0,192	0,000,027.4
dansk	da	005.00000	0,114	0,000,022.8
slovenčina	sk	005.00000	0,109	0,000,021.8
polski	pl	050.00000	0,627	0,000,012.5
čeština	cs	012.00000	0,134	0,000,011.2
Deutsch	de	105.00000	0,943	0,000,009.0
italiano	it	100.00000	0,596	0,000,006.0
English	en	510.00000	3,000	0,000,005.9
română	ro	026.00000	0,130	0,000,005.0
日本語	ja	130.00000	0,609	0,000,004.7
українська	uk	044.00000	0,158	0,000,003.6
français	fr	270.00000	0,839	0,000,003.1
português	pt	180.00000	0,500	0,000,002.8
Türkçe	tr	060.00000	0,134	0,000,002.2
Bahasa Melayu	ms	023.00000	0,046	0,000,002.0
ไทย	th	030.00000	0,049	0,000,001.6
español	es	320.00000	0,504	0,000,001.6
русский	ru	280.00000	0,422	0,000,001.5
Tiếng Việt	vi	068.00000	0,096	0,000,001.4
한국어	ko	078.00000	0,108	0,000,001.4
srpskohrvatski / српскохрватски + српски / srpski + hrvatski + bosanski	sh + sr + hr + bs	021.00000	0,024	0,000,001.1
فارسی	fa	070.00000	0,066	0,000,000.9
azərbaycanca	az	031.00000	0,025	0,000,000.8
తెలుగు	te	070.00000	0,044	0,000,000.6
Bahasa Indonesia	id	180.00000	0,109	0,000,000.6
Sunda	su	027.00000	0,014	0,000,000.5
العربية	ar	250.00000	0,100	0,000,000.4
मराठी	mr	068.00000	0,024	0,000,000.4
മലയാളം	ml	035.00000	0,011	0,000,000.3
中文	zh	900.00000	0,270	0,000,000.3
தமிழ்	ta	066.00000	0,019	0,000,000.3
Jawa	jv	075.00000	0,020	0,000,000.3
ಕನ್ನಡ	kn	035.00000	0,007	0,000,000.2
贛語	gan	021.00000	0,004	0,000,000.2
ગુજરાતી	gu	046.00000	0,008	0,000,000.2
اردو	ur	060.00000	0,010	0,000,000.2
粵語	zh-yue	071.00000	0,011	0,000,000.2
閩南語 / Bân-lâm-gú	zh-min-nan	046.00000	0,006	0,000,000.1
বাংলা	bn	185.00000	0,020	0,000,000.1
हिन्दी	hi	360.00000	0,036	0,000,000.1
မြန်မာဘာသာ	my	032.00000	0,002	0,000,000.1
吴语	wuu	077.00000	0,004	0,000,000.1
پښتو	ps	023.00000	0,001	0,000,000.0
ਪੰਜਾਬੀ	pa	057.00000	0,002	0,000,000.0
ଓଡ଼ିଆ	or	032.00000	0,001	0,000,000.0
客家語 / Hak-kâ-ngî	hak	034.00000	0,001	0,000,000.0
湘語 (Xiang)	hsn	036.00000	0,000	0,000,000.0
मैथिली	mai	030.00000	0,000	0,000,000.0
भोजपुरी	bh	026.00000	0,000	0,000,000.0
Hausa	ha	024.00000	0,000	0,000,000.0

Sister projects global penetration

Wikipedia: 96 % of total requests

Number of contributors and page views, compared
Sister project	Number of contributors^[3]	Contributors over total (%)	Monthly page views (millions)	Page views over total (%)
Wikibooks	12,942	1.28	35.6	0.33
Commons	6557^[4]	0.65	173	1.63
Wikinews	2969	0.30	12	0.11
Wikipedia	962,555^[5]	95.44	10,175	95.71
Wikiquote	5963	0.59	33.5	0.31
Wikisource	4605	0.46	34.1	0.32
Wikispecies	633^[6]	0.06	8 ca.^[7]	0.08
Wikiversity	2329	0.23	6.6	0.06
Wiktionary	10,021	0.99	151	1.42
Total	1,008,574	99.7^[8]	10,631	99.97^[8]

With regard to share, Wikiversity and Wikibooks seem to have a number of contributors which is "four times" the number of page views, Wikinews "three times", Wikiquote "twice", Wiktionary "two thirds", Commons almost "one third" (but the figure is wrong), other projects quite balanced. What does this mean? If

the total number of edits / number of active contributor ratio is more or less the same in all projects, and
if the number of edits is an indicator of the total effort of the community more or less equal in all projects,

then we can consider that

we have some projects where the community does a better focused work (i.e. less work, more useful to readers), either
1. because the community is better directed/more able,
2. or because the whole project is more needed than another (i.e. maybe a dictionary is more needed than a wiki newspaper),
3. or because the project is more difficult and needs more effort to achieve some critical mass to be useful,
or we have an unequal distribituon of:
1. readers: some projects should be more advertised or linked from other projects/other web resources/other offline resources (e.g. schools);
2. editors: some projects fail to attract the needed editors, either because they're "wasted" on another projects which needs them less, or because they are not able to attract new wikimedia editors (i.e. editors who don't arrive from another project) for some reason (see Participation).

Note that the premises need to be verified:

a project can attract more very active editors and another can attract more less active editors (e.g. because maybe to collect quotations is less difficult than to write a book);
some projects may require more "external" preparative work (e.g. a public domain dictionary which is processed offline and then uploaded to Wiktionary by bots) or spread work on many edits (e.g. the uploading of a book full text on Wikisource requires an edit per page plus others for index, metadata etc.).

References

↑ Estimates for upper bound of range from ComScore for the month of December 2008. Estimate on lower bound of range produced by Bridgespan using both ComScore December 2008 figures and the International Telecommunication Union 2008 data on internet use ITU data. In cases where no range is given it is because the two sources converged on the same number. Note that ComScore data is produced though an opt-in panel of two million internet users around the globe and uses a range of statistical techniques to create an internally consistent portrait of the global internet audience. However it is important to note that they do not estimate users who access the Internet via Internet cafés, and people below age 15, therefore, their statistics will not be representative of certain populations. For information on ComScore as well as what countries are included in the regions see ComScore data on Wikimedia (*)Upper bound from ITU and ComScore data. Lower bound from ComScore data.
↑ Numbers from Data on language speakers from Ethnologue 2009 http://www.ethnologue.com/ethno_docs/distribution.asp?by=size#3, data on Wikipedia articles from wikistats http://stats.wikimedia.org/EN/Sitemap.htm
↑ From WikiStats, authors who edited at least 10 times since they arrived, total number for all subdomains (when appliable), e.g. Wikiquote. August 2009.
↑ This figure is wrong because WikiStats does not consider edits to categories and file description pages, and thus considers only 1/30 of the total edits (compared to e.g. 1/3 on Wikipedia).
↑ July 2009.
↑ June 2009.
↑ From WikiSpecial WikiStats figures (August 2009): 11844 views/hour*24 hours*30 days in September/1,000,000.
↑ ^8.0 ^8.1 Approximation error.

[1] Estimates for upper bound of range from ComScore for the month of December 2008. Estimate on lower bound of range produced by Bridgespan using both ComScore December 2008 figures and the International Telecommunication Union 2008 data on internet use ITU data. In cases where no range is given it is because the two sources converged on the same number. Note that ComScore data is produced though an opt-in panel of two million internet users around the globe and uses a range of statistical techniques to create an internally consistent portrait of the global internet audience. However it is important to note that they do not estimate users who access the Internet via Internet cafés, and people below age 15, therefore, their statistics will not be representative of certain populations. For information on ComScore as well as what countries are included in the regions see ComScore data on Wikimedia (*)Upper bound from ITU and ComScore data. Lower bound from ComScore data.

[2] Numbers from Data on language speakers from Ethnologue 2009 http://www.ethnologue.com/ethno_docs/distribution.asp?by=size#3, data on Wikipedia articles from wikistats http://stats.wikimedia.org/EN/Sitemap.htm

[3] From WikiStats, authors who edited at least 10 times since they arrived, total number for all subdomains (when appliable), e.g. Wikiquote. August 2009.

[4] This figure is wrong because WikiStats does not consider edits to categories and file description pages, and thus considers only 1/30 of the total edits (compared to e.g. 1/3 on Wikipedia).

[5] July 2009.

[6] June 2009.

[7] From WikiSpecial WikiStats figures (August 2009): 11844 views/hour*24 hours*30 days in September/1,000,000.

[app-8] 8.0 ^8.1 Approximation error.

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]