Once again, we’re looking at how the world is embracing new TLDs – country by country. Last time, I compared each nation’s nTLD footprint with its online population, calculating a ratio of overrepresentation / underrepresentation. Another way to view this is “domain user density”, which I define as the number of internet users per registration. Here’s what it looks like:
In case your screen is too squished to view all the captions, here is the list of countries from left to right: China, Armenia, Panama, Netherlands, Switzerland, USA, Austria, Australia, Germany, Lithuania, UK, Hong Kong, Canada, Norway, Cambodia, New Zealand, France, Ukraine, Russia, Indonesia, Ireland, Denmark, Singapore, Sweden, Belgium, UAE, Japan, South Korea, Israel, Czech Republic, Italy, Spain, Romania, Turkey, Greece, Poland, Hungary, Vietnam, Chile, Thailand, Pakistan, Brazil, Malaysia, South Africa, India, and Mexico.
Why these 46 countries? My study is based on nTLDStats, which lists 49 nations with the most nTLD domain registrations (in absolute terms, not concentration). From these I have excluded 3 outliers: Gibraltar, the Cayman Islands, and the Cocos Islands – tiny places that are home to large domain industry companies, giving them an oversized footprint. For instance, the Cocos Islands have fewer than 600 residents; but more than 19,000 nTLD domains list this territory in whois contact info. Why? Probably because of .CC.
As you can see, in China 1 nTLD domain has been registered for every 57 internet users. Meanwhile, in Mexico the ratio is 1 nTLD domain per 3172 people online. Note: nTLD volume within each country is underreported because 27.3% of domains are masked by whois privacy. Consequently, real user density is lower than what I’m reporting. If you assume whois privacy is used equally by all nationalities, then you can shave off a quarter. But these numbers are changing all the time, as new nTLDs are released, more domains are registered, and old nTLD domains expire and are deleted.
Exact numbers aren’t the important story here. The real question is this: Why do nations differ? Why are China, Switzerland, and the USA more interested in buying nTLD domains than Poland, Hungary, and Vietnam? Why the steep jump for Chile, Thailand, and Pakistan? Why the even steeper jump for Brazil, Malaysia, South Africa, and India? And why is Mexico – 35th nation in terms of nTLD registrations overall – so much less interested in nTLDs per online citizen than anybody else?
We’re in the dark here. I have a dozen theories. And 2 partial answers: Wealth and Internet Access.
Wealthy nations register more nTLD domains. The correlation is weak (-0.411) – meaning the dots don’t hug a straight line. But there’s a discernible pattern nonetheless. Really, it’s as if we’re looking at 2 different pictures glued together! Below $30k, we have a messy, haphazard situation. Some countries devour nTLDs; so their nTLD user density is quite low. China is a relatively poor country, with a gross domestic product (GDP) averaging $14.4k per person; yet China registers more nTLD domains than anybody else. Armenia ($8.4k) has registered 1 nTLD per 71 internet users – more than the USA! And the poorest country shown here, Cambodia ($3.5k per person), ranks 14th in terms of nTLD registrations. Its nTLD user density of 217 places Cambodia ahead of much wealthier nations like France and Russia.
Thus, below $30k, national wealth seems unrelated to nTLD registration rates. Above that cutoff, however, we see no such chaos. Wealthy countries register nTLD domains at a predictably higher rate per person. All of them (24 out of 46) have an nTLD user density below 500. Let’s state this precisely:
User Density | >$30k | <$30k |
---|---|---|
Min | 85.9 | 56.9 |
Q1 | 164.9 | 233.1 |
Median | 257.5 | 554.3 |
Q3 | 339.2 | 1095.4 |
Max | 442.6 | 3172.3 |
Mean | 248.5 | 807.3 |
Std Dev | 101.4 | 762.5 |
Apart from national wealth, we should also look at online access. To some degree, the percentage of internet users within a given country is tied to GDP. As you can see, these 2 variables track quite closely (with a correlation coefficient of 0.798). Still, they do diverge; and the more we know, the better. For instance, 2 countries differing greatly in terms of GDP per capita –Singapore ($85.4k) and the Czech Republic ($32.8k) – both have 82% internet access.
So what does internet access tell us about nTLD registration rates? Déjà vu. Once again, there’s a cutoff point. Where 73% or more of a country’s citizens are internet users, the nTLD user density is consistently low – again, below 500; and, again, this applies to 24 countries. As before, below this cutoff, national nTLD patterns are all over the place. Take a look:
What you ought to be thinking right now is this: 24 nations worth more than $30k per capita … and 24 nations with greater than 73% internet access … Are we perhaps looking at the same 24 countries in both cases? Almost. 23 of them are identical, while Italy and Russia each belong to only 1 list. This means 25 countries belong to a sort of elite group. They’re developed nations worth $30k or more per capita and/or having greater than 73% of citizens online. Knowing that, we can guarantee these nations have an nTLD user density below 500 – i.e. at least 1 nTLD domain is currently registered for each 500 internet users.
That’s interesting, right? But what I’m about to tell you is even more striking. Wealth and internet access explain most cases of high nTLD registration rates – not all. As you can see, there are 9 countries that fall short of both standards … yet have nTLD user density below 500. Who are these “registrant nations” that bucked the trend and ambushed us, registering more nTLD domains per person than we have any right to expect? Say hello:
Country | User Density | % Online | GDP Per Capita | Near Cutoff |
---|---|---|---|---|
China | 56.9 | 50.3 | $14.5k | 33% |
Armenia | 71.0 | 58.3 | $8.4k | 22% |
Panama | 75.9 | 51.2 | $22.2k | 52% |
Lithuania | 159.0 | 71.4 | $27.7k | 90% |
Cambodia | 216.8 | 19.0 | $3.5k | 3% |
Ukraine | 233.1 | 49.3 | $7.9k | 18% |
Indonesia | 277.9 | 22.0 | $11.1k | 11% |
Romania | 467.4 | 55.8 | $21.4k | 54% |
Turkey | 472.6 | 53.7 | $19.6k | 48% |
Some of these nations are quite near our cutoffs of $30k and 73% web access. Lithuania barely falls short – which is why I say it’s 90% near the cutoff. But Cambodia is far, far away in both respects. It measures only 3% of those standards. Why, then, are so many nTLD domains registered in Cambodia?
At this point, aside from Lithuania, the other 8 nations all remain a mystery. Even China, about whom so much has been written, doesn’t fit the mold. That is to say, national wealth and internet access aren’t the full story. Clearly other factors are in play. What are they? On the flip side of this coin, look at the following 12 nations, where fewer than 1 nTLD domain is registered per 500 internet users:
Country | User Density | % Online | GDP Per Capita | Near Cutoff | Above 500 | Discrepancy |
---|---|---|---|---|---|---|
Malaysia | 1725.2 | 71.1 | $27.0k | 87% | 345% | 3.0 |
Hungary | 603.5 | 72.8 | $25.8k | 86% | 121% | 1.0 |
Poland | 575.4 | 68.0 | $26.3k | 82% | 115% | 0.9 |
Greece | 533.2 | 66.8 | $26.6k | 81% | 107% | 0.9 |
Chile | 977.0 | 64.3 | $22.4k | 66% | 195% | 1.3 |
Mexico | 3172.3 | 57.4 | $17.3k | 45% | 634% | 2.9 |
Brazil | 1606.0 | 59.1 | $15.4k | 42% | 321% | 1.3 |
South Africa | 1725.9 | 51.9 | $13.2k | 31% | 345% | 1.1 |
Thailand | 1056.8 | 39.3 | $16.3k | 29% | 211% | 0.6 |
Vietnam | 604.8 | 52.8 | $6.0k | 15% | 121% | 0.2 |
India | 1784.3 | 28.8 | $6.1k | 8% | 357% | 0.3 |
Pakistan | 1095.4 | 18.0 | $5.0k | 4% | 219% | 0.1 |
Those nations that sit near our cutoff ($30k and 73% web access) ought to have an nTLD user density not too far above 500. Multiplying these 2 percentages together gives us a measure of the discrepancy. The farther this discrepancy is from 1, the less we know about what’s going on.
Above 1.0 indicates the country is registering fewer nTLDs per person than we’d expect, given its wealth and web access. Malaysia and Mexico apparently take no interest in nTLDs. Conversely, a “discrepancy” below 1.0 indicates the country is registering more nTLDs per person than we’d anticipate based on our development metrics (GDP per capita and online access).
It turns out, India, Pakistan, and Vietnam – despite registering few nTLDs per person – are actually over-performing. Once we make allowances for wealth and internet penetration, we find these countries are punching well above their weight. This squares with my experience in the domain industry. Indian and Pakistani domainers are quite active; and I was surprised, at first, to see such a small nTLD footprint for India in particular.
Having uncovered the reason for this lag in nTLD registrations in India so far, I can make a sort of pronouncement: This bodes well for the nTLD program. Internet penetration in India is expected to grow at triple the global rate. GDP will increase too. If web access doubles, reaching just 57.6%, that will mean 735 million internet users. And if India’s nTLD user density is reduced to 500 – in keeping with other developed countries – India will have 14.7 million nTLD domains … compared to 2.06 million today. As of today, there are 27.6 million nTLD domains registered worldwide. Thus, ongoing growth in India might conceivably add another 50% or so to the total. That’s a big “if”, but India will soon overtake China as planet earth’s biggest population. Obviously, India has big potential.
Many other nations could affect nTLD registration volume. Remember, I’m only looking at countries that appear among the top 50 at nTLDStats.com. Combined, those countries have 2.5 billion internet users; but another 0.7 billion live in other countries, where internet access is advancing by leaps and bounds. At present, some countries show a startling lack of interest in nTLD domains; and that might change. In addition to Mexico and Malaysia, consider which nations don’t appear at all at nTLDStats. Nigeria ranks 7th worldwide with 86.4 million internet users – more than Germany, France, or the UK. Nigeria also has 47.4% web access – better than India’s. And its GDP per capita is exactly the same as that of India and Vietnam. Those 2 nations are #9 and #19, respectively, in terms of nTLD registration volume. So why not Nigeria?
Other nations that didn’t rank on nTLDStats but which already have more than 10 million internet users include (in descending order by online population): the Philippines, Iran, Egypt, Argentina, Colombia, Bangladesh, Saudi Arabia, Kenya, Morocco, Venezuela, Taiwan, Algeria, Kazakhstan, Peru, Uzbekistan, Myanmar, Ethiopia, and Sudan.
These countries aren’t as far behind as you might think. Even a nation like the United States, which is only mediocre according to the metrics we’ve been discussing, does pretty well for itself. Depending on the source, the USA ranks 10th, 11th, or 13th by per capita GDP ($56.1k); and we come in 49th by internet access (74.6%) – which is about even with Kazakhstan, Hungary, Russia, or Lithuania … far behind Canada, Australia, the UK, Japan, Korea, and (of course) western Europe.
But let’s return to the mystery. Why don’t Mexico and Malaysia buy nTLDs? And why do Cambodia, China, Indonesia, Ukraine, and Armenia register so many? What factors are we missing?
Footnote: I should mention that I have filtered the data for this study. Some nTLDs describe cities, regions, or ethnicities. Predictably, a country will register more GEO nTLDs when those describe its own geography. But the distribution of these GEO nTLDs is far from uniform. Some countries have several; many have none. If we are to compare nTLD registration rates among different nations, looking for factors that underlie them all, then we must remove the unequal contribution of GEO nTLDs. So that’s what I did; I excluded GEO nTLD domains from my counts. Those GEO nTLDs will be the topic of my next article.
Sergey says
I think we need to take into account which nTLDs are preferred from country to country. According to ngtld stats and common sense, poorer countries tend to have much higher, say, n(registered .top)/n(total registered) ratio than well-developed ones. That’s because .top were selling at penny each, and people from less-developed countries are unlikely to afford registration of large amount of more expensive nTLDs, so they bulk-register cheapest ones, preferring quantity over quality. For Armenia the above ratio is 0.75 whereas for the USA it is 0.17. I think it makes sens to make breakdown by nTLD, say, perform analysis of only .xyz, then set of cheapest nTLDs etc.
Joseph Peterson says
@Sergey,
Absolutely. This is an article series, and that’s where I’m headed next. I thought it best to start by looking at each country’s overall nTLD volume. Next, I’ll be “zooming in” to look at specific TLD distribution.
Mark Thorpe says
Most nTLD’s are speculative and a bad Investment IMO.
You can produce all the data you want, but a bad Investment, is a bad Investment, no matter how it is presented. Again, just my opinion.
Stick with .Com .Net .Org .Co .Web people, forget about all the rest other than maybe .Club.
Happy Domaining!
Joseph Peterson says
@Mark Thorpe,
I don’t disagree. Some domains in some nTLDs make great brand names. A fraction of those (depending on price) might be good investments for resellers. Plenty of nTLDs aren’t very promising overall. And many good domains, though they might be put to use by end users, might be bad options for us middle men.
Keep in mind, DNW readers aren’t only domain investors. Part of this audience includes people at the level of registries, registrars, and so forth.
Also, I’m interested in domain names whether or not the domain passed through a domainers’ hands or turned a profit. My own role in this industry includes advising registrants on name selection, quite apart from domaining and domainers. And I like to know what registries themselves are up to – which requires examining who is registering their stuff.
So if I write articles about nTLD registration patterns, it doesn’t mean I’m endorsing the nTLDs I’m looking at. It just means I’m looking at what’s going on in the domain industry – pure and simple.
Mark Thorpe says
Okay, but I just look at what makes the most sense and what will make people like myself money.
I also like to look out for the small guy that does not know a lot about domain names and give my opinion on topics.
I don’t just post on blogs for my own purposes.
Plain and Simple.
Joseph Peterson says
Keep sharing your perspective, Mark. People thinking aloud and listening to one another is a very good thing.
Mark Thorpe says
I will.
I also like to keep people honest.
Cheers
Sergey says
Great article series Joseph, keep it going!
Rubens Kuhl says
Factoring the number of ICANN-Accredited RAA 2013 Registrars in each country might also shed a light. Brazil has 0 RAA 2013 registrars, so new gTLDs have a hard time selling in the country… and, after yesterday’s termination of one of the oldest registrars, there is now only one ICANN-Accredited Registrar in a 200 mi population country. internic.net still lists 3 but two of them have already been given termination notices.
Joseph Peterson says
@Rubens Kuhl,
I hadn’t thought of that. Thanks for the insight. One of my reasons for squeezing out odd cases – countries that don’t fit the norm (like Brazil or Mexico or Cambodia or even China) – is that I hope knowledgeable people like you will step forward with theories.
Rubens Kuhl says
Mexico and Brazil may also share some similarities here. Both have strong ccTLD registries that sell direct or thru a whole-owned subsidiary; the Mexico ccTLD registrar sell legacy gTLDs and some ccTLD, but the only gTLD they sell is .lat, which is operated by them. So even though the largest gTLD registrar in Mexico, Neubox, sells some new gTLDs like .xyz and Radix TLDs, overall their market share in domain selling is small.
Joseph Peterson says
@Rubens Kuhl,
I think you’re right.
168 says
I am aware of India investors using Panama to circumvent a kind of excise tax related to domains.
Maybe the China market is using Cambodia and Indonesia in the same way.
Similar reasons US companies use Cayman Islands. Ireland etc.
I agree with Sergey and you here,
“I think we need to take into account which nTLDs are preferred from country to country”.
And would like to add, Cultural language- as in China it’s not just about cheap reg promos but meaning.
In the English world the New “G”s gaining are ubiquitous terms.
Spot on with this observation. Especially with the mobile access available today.
“Internet penetration in India is expected to grow at triple the global rate. GDP will increase too. If web access doubles, reaching just 57.6%, that will mean 735 million internet users. And if India’s nTLD user density is reduced to 500 – in keeping with other developed countries – India will have 14.7 million nTLD domains … compared to 2.06 million today. As of today, there are 27.6 million nTLD domains registered worldwide. Thus, ongoing growth in India might conceivably add another 50% or so to the total. That’s a big “if”, but India will soon overtake China as planet earth’s biggest population. Obviously, India has big potential.”
Many thanks for the effort.
Cheers
Joseph Peterson says
@168,
That’s interesting. I knew Panama was used to hide assets, and figured it might see an outsized number of domain registrations either for that reason or because of whois privacy services.
But I didn’t realize, until you mentioned it, that Cambodia and Indonesia might fulfill a similar role for China. Emphasis on “might” because I don’t really know whether that’s the case, let alone how many nTLD domain registrations in those countries are, in fact, due to skirting taxes. Still, an enticing hypothesis.