Analyzing new top level domain registrations by country (Part 3: the Geos)

This is the third part of Joseph Peterson’s analysis of new TLD registrations by geography.

In 2 previous articles (here and here), we scrutinized nTLD registration volume by country, asking which nations are most / least engaged overall with the 1000+ new domain endings released since 2014.  But that was only a bird’s-eye view – meaning we only looked at an aggregate of all nTLDs.  Now it’s time to “zoom in” and examine which countries register which extensions.  After all, to answer why they’re buying, we must look at what they’re buying.

This installment focuses on “GEO” nTLDs – suffixes that correspond to cities (.BERLIN, .NYC, .SYDNEY, .ISTANBUL); regions (.OKINAWA, .WALES, .CORSICA); or nationalities / ethnicities (.IRISH, .SCOT, .LAT, .KIWI, .CYMRU).  For any study of national nTLD habits, such GEOs are an obvious place to start.

Among those domains that are available to register, many seem to be performing well:

Country	TLD	Rank in Country	% of Country	% of TLD
Germany	.BERLIN	1	9.3	96.7
Germany	.BAYERN	3	5	98.5
Germany	.KOELN	5	3.9	97.8
Germany	.HAMBURG	6	3.9	97.7
Germany	.COLOGNE	25	0.8	90.3
Germany	.SAARLAND	35	0.6	94.9
Japan	.TOKYO	6	4.5	32.6
Japan	.NAGOYA	22	0.4	33.8
Japan	.YOKOHAMA	28	0.3	29.4
Japan	.OKINAWA	37	0.2	23.3
Japan	.KYOTO	44	0.2	99.8
Japan	.OSAKA	53	0.1	60.7
Belgium	.VLAANDEREN	1	13.2	91.3
Belgium	.BRUSSELS	2	13.1	86.2
Belgium	.GENT	5	6.5	86.6
Belgium	.PARIS	51	0.3	0.5
Belgium	.AMSTERDAM	82	0.2	0.3
Austria	.WIEN	1	19.6	90.4
Austria	.TIROL	3	6.4	90
Austria	.BERLIN	55	0.3	0.4
Austria	.BAYERN	82	0.2	0.5
France	.PARIS	4	7.2	86.9
France	.BZH	7	2.3	95.4
France	.ALSACE	20	0.7	93.2
France	.CORSICA	49	0.3	95.6
South Africa	.CAPETOWN	1	15.8	88.1
South Africa	.JOBURG	3	11.6	87.5
South Africa	.DURBAN	4	8.1	84.9
South Africa	.LONDON	41	0.4	0.2
UK	.LONDON	1	11	76.9
UK	.WALES	6	2.6	84
UK	.SCOT	7	2.4	86.8
UK	.CYMRU	11	1.4	85.4
Armenia	.рус (.XN–P1ACF)	11	0.3	0.1
Armenia	.москва (.XN–80ADXHKS)	17	0.1	0.1
Armenia	.MOSCOW	22	0.1	0.1
Australia	.MELBOURNE	3	5.4	84.3
Australia	.SYDNEY	4	5.2	85.7
Australia	.KIWI	58	0.3	0.2
Russia	.рус (.XN–P1ACF)	3	9.8	57.5
Russia	.MOSCOW	4	3.5	69.1
Russia	.москва (.XN–80ADXHKS)	5	3.3	76.6
Switzerland	.BERLIN	34	0.5	0.8
Switzerland	.PARIS	50	0.4	1.5
Switzerland	.LONDON	65	0.3	0.5
Ukraine	.рус (.XN–P1ACF)	10	0.9	1.1
Ukraine	.MOSCOW	33	0.2	0.6
Ukraine	.москва (.XN–80ADXHKS)	34	0.1	0.6
USA	.NYC	6	2.6	97.3
USA	.MIAMI	38	0.4	84.5
USA	.VEGAS	43	0.3	71.5
Hong Kong	.CYMRU	52	0.3	1.3
Hong Kong	.WALES	79	0.2	0.4
Ireland	.IRISH	2	7.8	51.7
Ireland	.LONDON	23	0.7	0.2
Lithuania	.MOSCOW	44	0.1	0.1
Lithuania	.TOKYO	100	0.02	0.01
Malaysia	.VEGAS	81	0.1	0.1
Malaysia	.RYUKYU	92	0.1	2.6
Netherlands	.AMSTERDAM	2	12	97.1
Netherlands	.VLAANDEREN	75	0.2	5.3
Romania	.BERLIN	76	0.1	0.04
Romania	.LONDON	89	0.1	0.03
Spain	.BARCELONA	2	8.5	91.3
Spain	.LAT	99	0.2	5.5
Brazil	.RIO	6	1.9	100
Cambodia	.TOKYO	77	0.01	0.0
Canada	.QUEBEC	3	4.5	85.1
Cayman Islands	.LONDON	99	0.02	0.02
Chile	.LAT	15	0.5	2.1
Cocos Islands	.MOSCOW	15	0.01	0.0
Denmark	.LONDON	92	0.2	0.06
Gibraltar	.LONDON	38	0.04	0.03
Greece	.LONDON	86	0.2	0.05
Hungary	.WIEN	29	0.4	0.4
Israel	.VEGAS	100	0.2	0.2
Mexico	.LAT	4	5.7	54.6
New Zealand	.KIWI	1	91.8	99.2
Singapore	.VEGAS	100	0.2	0.2
Thailand	.PARIS	93	0.1	0.1
Turkey	.ISTANBUL	3	13.4	82.1
UAE	.LONDON	73	0.2	0.1

nTLDStats, on whose data this study is based, lists the top 49 nations by nTLD volume and, for each of those countries, its top 100 nTLDs (except for the Cocos Islands, which lists only 19).  Out of 389 distinct nTLDs arising in those national lists, 45 are GEOs.  Hence, if an nTLD ranks in the top 100 for any of those top countries, there is an 11.6% chance it’s a GEO.

However, only 87 of those 4819 chart positions (1.8%) are occupied by GEOs.  Why the discrepancy?  Because GEO nTLDs are concentrated in fewer countries.  On average, a non-GEO will appear on 14 country lists (4732 / 344), whereas a GEO is confined to just 2 country rankings (87 / 45).

We’d expect this.  For instance, .CLUB and .XYZ have a large footprint worldwide, whereas .KOELN sees negligible participation outside Germany where Köln is located.  In fact, there are 45 nTLDs so widespread that they appear in the top 100 lists for at least 40 nations; and 4 of those nTLDs rank in every country.  GEOs, on the other hand, behave very differently:

TLD	Nations	Domains	% of TLD	Participants
.LONDON	10	44,966	77.9	UK + 1.2% = Switzerland, South Africa, Ireland, UAE, Denmark, Greece, Romania, Gibraltar, Cayman Islands
.MOSCOW	5	16,466	69.8	Russia + 0.8% = Ukraine, Armenia, Lithuania, Cocos Islands
.BERLIN	4	57,370	97.9	Germany + 1.2% = Switzerland, Austria, Romania
.PARIS	4	19,659	89.1	France + 2.1% = Switzerland, Belgium, Thailand
.VEGAS	4	8960	72.2	USA + 0.5% = Israel, Singapore, Malaysia
.TOKYO	3	16,220	32.6	Japan + 0.01% = Lithuania, Cambodia
.москва (.XN–80ADXHKS)	3	15,601	77.3	Russia + 0.7% = Ukraine, Armenia
.рус (.XN–P1ACF)	3	46,485	58.6	Russia + 1.2% = Ukraine, Armenia
.LAT	3	1562	62.1	Mexico + 7.6% = Spain, Chile
.AMSTERDAM	2	25,165	97.4	Netherlands + 0.3% = Belgium
.BAYERN	2	30,491	99	Germany + 0.5% = Austria
.CYMRU	2	5709	86.7	UK + 1.3% = Hong Kong
.KIWI	2	200,882	99.5	New Zealand + 0.2% = Australia
.VLAANDEREN	2	6432	96.6	Belgium + 5.3% = Netherlands
.WALES	2	10,390	84.4	UK + 0.4% = Hong Kong
.WIEN	2	13,597	90.7	Austria + 0.4% = Hungary

As you can see, only 16 GEOs rank in more than 1 country.  .LONDON, which has charted for 10 different countries, at first seems to have the broadest appeal.  But we don’t get the full story by counting only cases where an nTLD has charted.  A given extension might be widely distributed outside the top 49 countries.  Or local competition may push the TLD off the top 100 list, even if that country constitutes a large slice of the TLD’s pie.  Also, domains with whois privacy are excluded.  For a more exact measurement of dispersion, look at “% of TLD” above.

Some nTLDs are concentrated in a single country (and use transparent whois): .RIO (100%), .KYOTO (99.8%), .KOELN (97.8%), .HAMBURG (97.7%), and .NYC (97.3%).  Note: With residency requirements for .NYC, a broad country profile is impossible.  Other GEOs are much more dispersed (or else use whois privacy more frequently).  For instance, visible home-country registrations for .ISTANBUL and .MELBOURNE constitute only 82.1% and 84.3%, respectively.  Even more extreme, .IRISH is found with Ireland listed in whois records only 51.7% of the time.  For .IRISH, the remainder would belong mainly to the UK, I imagine – even though the suffix doesn’t rank among the top 100 there.

Visibility varies.  Compare .BERLIN (where 97.9% of registrations can be seen within 4 countries) to .LONDON (where 22.1% of registrations remain unaccounted for despite 10 national rankings).  This suggests that .LONDON may have very scattered, global ownership.  Likewise, .MOSCOW (69.8%), .рус (58.6%), and .LAT (62.1%) are only partially visible in country charts.  The first 2 are perhaps widely registered in former Soviet states – nations not among the most active 49.

It turns out, secondary countries contribute few registrations, amounting to 1% or more of total volume for only 7 GEOs.  So, even though .LONDON ranks for 9 countries outside the UK, those other participants comprise only 1.2% of the total.  Looking at the percentage of visible domains owned by minority nations, the least centralized GEOs are .LONDON (1.6%), .рус (2.1%), .PARIS (2.4%), .VLAANDEREN (5.8%), and .LAT (13.9%).  Unlike city GEOs, .LAT is intended for a broad international audience of latinos.  To some extent we do see diverse ownership, though Mexico alone accounts for more than half of .LAT domains.

It’s interesting to observe which minority nations participate in GEOs that don’t apply principally to themselves.  Denmark buys .LONDON as a UK neighbor.  Switzerland buys .PARIS and .BERLIN because both French and German are Swiss languages.  Israel registers .VEGAS, thanks to cultural ties with the USA.  Belgium and the Netherlands reciprocate, sharing .VLAANDEREN and .AMSTERDAM across a shared border.  Ukraine and Armenia, as Russian satellites, each echo 3 Russian GEOs.  Oddly enough, Hong Kong shows uncanny interest in .CYMRU and WALES – perhaps due to Welsh expats living in this former British colony.

My next article will assess how successful these GEOs have been.  So far, we’ve been measuring how concentrated they are within narrow national markets.  In case this has bored you, let me explain why laying such groundwork will matter later on.  GEOs are an extreme case of national bias, and they give us a benchmark to measure other nTLDs against – to judge to what extent various non-GEOs (like .VIP or .FUTBOL or .XYZ or .HAUS) are skewed toward particular registrant countries.  Also, the footprint of GEOs within national borders has implications for non-GEO nTLDs meant for niche communities – e.g. .LGBT and.ARCHI.

Now let me leave you with a mystery:

Something very peculiar is going on with Japanese GEOs, which are mostly missing from these 49 national charts.  Very little of the ice berg is visible: .OSAKA (60.7%), .TOKYO (33.6%), .NAGOYA (33.8%), .OKINAWA (23.3%), .YOKOHAMA (29.4%).  Most striking of all is .RYUKYU at just 2.6%.  Why so low?

Japan itself is among the top 49 countries being tracked.  And the Ryukyu Islands (which include Okinawa) belong to Japan.  We’d therefore expect most .RYUKYU registrations to occur in Japan.  Yet .RYUKYU isn’t found among Japan’s top 100 nTLDs.  Why?  This 1 is easy to understand.  The 2.6% under consideration is actually due to a top-100 ranking for Malaysia.  Undoubtedly more .RYUKYU domains (in absolute numbers) have been registered in Japan – just not enough to break into Japan’s top 100.

What, though, can explain the low numbers for the other 5 Japanese GEOs?  Japan is shown as the registrant only 1/3 of the time in 4 of those cases!  Here are a few possibilities: (1) higher whois privacy usage; (2) an owner habit of listing Japanese provinces rather than Japan proper in whois fields; (3) an inability on the part of nTLDStats to recognize designations for Japan; (4) overwhelming buyer interest in Japanese nTLDs outside rather than inside Japan.  What do you think?  Perhaps most puzzling of all, .KYOTO, another Japanese GEO, doesn’t misbehave.  99.8% of .KYOTO domains do show Japan in whois.

Footnote: In defining GEO nTLDs, my dividing line is somewhat arbitrary.  Notably, keyword language is not a factor.  Thus, despite being a German word, .IMMOBILIEN is excluded.  Likewise, .GMBH (which is a German acronym) doesn’t count.  From an American perspective, those are obviously German.  Why set them aside?  Answer: For the sake of consistency.  If those German suffixes are to count as GEOs, then so must the matching English versions, .PROPERTY and .LLC.  Almost every nTLD is a keyword in SOME language – and foreign, depending on where you’re located.  To consider them all GEOs would be pointless.  Absolutely, language and geography do interact.  But I’ll postpone language until a later article.