[CPWG] Reminder: ACTION/CPWG by Today 23:59 UTC: Comment on Metrics

[John McCormac]

On 24/09/2020 20:32, mail at christopherwilkinson.eu CW wrote:
> Thankyou, Evin:  In general, my earlier input has been taken into account.
> However, under. CCT Recommendations, the fifth bullet is far too weak:
> 
>  >. Make a decision about competition objectives, and therefore, 
> required metrics (¿?)
> 
> Well, No. Competition is already ICANN's responsibility, and was the 
> primary
> 
> justification for the CCT Review in the first place.
> 
> 1.  This should be the first bullet.
> 
> 2.  Delete 'Take a decision'. I trust that the decision has already been 
> taken.
> 
> 3.  Proposed competition metrics:
> 
> -   Evolution of market shares:  Legacy, ccTLDs, new 2012 entrants, by 
> region and by year 1998-2019.

Apologies for the late reply. Ironically, I was busy working on domain 
name and usage statistics and missed the meeting.

With point 3 above, a lot of the data on market shares is highly 
fragmented due to ccTLD registries publishing little or partial 
registration statistics until quite recently. The 2000s round of new 
gTLDs had requirements that the registry reports would include 
statistics on country level registrations and a few of these old reports 
still exist. The requirement was dropped along the way and there is no 
registrations by country breakdown in current registry reports.

Some of the new gTLDs (2012 round) have developed their own geography of 
registrations and a few of the NGTs have become pseudo-ccTLDs. There has 
also been a major problem with demand.

The demand for the 2012 NGTs might have been there prior to 2008 but 
after large-scale Domain Tasting ended, much of that demand effectively 
disappeared. Many NGTs are struggling to get a few thousand new 
registrations per month. Renewal rates in some of the NGTs are not good. 
This has lead to some registries indulging in discounting to drive 
registrations. Discounting is like crack cocaine for registries in that 
it is highly addictive, difficult to stop using and quite destructive. 
Once a gTLD starts to use massive discounting, it destroys the 
credibility of the gTLD, drives out good registrations and kills 
development.

On a country basis, the legacy gTLDs are in trouble. Outside .COM, they 
are surviving on the renewal of brand protection registrations. The 
market in countries with a strong ccTLD is typically over 80% 
.ccTLD/.COM and even .COM is struggling against the local ccTLDs. It 
maintains its market position because it is still the top global TLD and 
it has market awareness. But ccTLD growth has generally outstripped gTLD 
growth (legacy and new) in most markets. This is a market shift. With 
new markets, people are looking to sell globally. With mature markets 
people sell more locally and there's nothing quite as local as the local 
ccTLD.

The regional issue is a major problem for ICANN and or the gTLDs. The 
number of registrars in the continent of Africa is extremely low. The 
number of registrars in Latin America is also low but it has one of the 
strongest ccTLDs (.BR) in North and South America. ICANN's 
registry/registrar model was adequate for the late 1990s but it is 
completely broken for the 2020s.

The number of US registrars, according to the May 2020 ICANN reports is 
1,931. The second place goes to China with 63. The majority of ICANN 
registars are drop catchers. They exist only to register deleting domain 
names and are not retail registrars. (The table for the breakdowns by 
country/COM/NET/ORG/BIZ/INFO/MOBI/ASIA are on the front page of 
HosterStats.)

This is what the top 10 countries by gTLDs looks like:

Place - Country - Registrars - AMBIONIC - COM - NET - ORG - BIZ - INFO - 
- MOBI -ASIA

1	United States	1,931	115,344,679	96,254,119	7,980,423	7,120,955	780,752 
2,870,507	256,022	81,901
2	China	63	16,164,053	14,737,181	1,089,888	169,985	35,989	36,189	49,594 
45,227
3	Germany	20	10,702,418	7,876,179	1,135,195	871,438	165,140	604,423 
29,441	20,602
4	Canada	17	10,351,574	8,634,615	802,159	664,688	83,773	147,261	13,615	5,463
5	India	60	4,553,671	3,832,946	309,943	287,343	49,457	58,859	6,815	8,308
6	Japan	15	3,929,366	2,922,285	516,577	125,096	160,395	185,236	8,749	11,028
7	France	16	3,508,658	2,649,808	360,933	293,906	64,387	122,975	11,386	5,263
8	United Kingdom	21	2,634,699	2,052,407	234,152	192,976	44,067	99,922 
6,827	4,348
9	Hong Kong	53	1,732,305	1,605,868	65,757	1,987	56,910	682	77	1,024
10	Spain	13	1,482,705	1,189,114	133,519	99,738	13,342	41,188	4,882	922

> -  Indicators of industry concentration and its evolution: by 
> Registries, by Registrars, by RSPs, by portfolio investors.

On the domain name registrations side of the industry, it is massively 
concentrated. One of the portfolio and reseller statistics reports that 
I was working on shows that Godaddy's main registrars and hosters 
portfolio has 25.22% of the .COM market. It also has smaller portfolios 
which focus on PPC, brand protection and ccTLD markets. Endurance 
International Group is in second place with 5.44%. The market 
concentration is staggering. Approximately 94% of the AMBIONIC 
(Asia/Mobi/Biz/Org/Net/Info/Com) market is on just 1.29% operators 
(registrars/resellers/hosters). Of the hosters in that market, 650,340 
are one hit wonders in that they only host one domain name (typically 
their own). This is often due to automated control panel setup for 
domain names.

The registrations by registrars metric is obsolete and completely 
unreliable. There has been a wave of acquistions that have seen 
registrars and hosters change hands. Some of the larger portfolio 
operators have webhosters which act as resellers and it is not unusual 
to see some of the top ten positions in various country level markets 
being occupied by the registrars and resellers of various portfolio 
operators. This often the portfolio operator's registrars and hosters 
competing with themselves.

The line between registrars and resellers has also become somewhat 
blurred due to the common ownership of registrars and resellers by 
portfolio operators. While the gTLD market can be described by registrar 
market share, there is not a one to one registar to reseller mapping and 
some resellers (not owned by the portfolio operators) will have 
registrations from multiple registrars for historical reasons.

Even the Registrars/Resellers model has some complexity because many 
resellers are not "resellers". There are different types of hosters.

Some of these are small businesses hosting their own domain names and 
they are not really resellers as such. They are often brand protection 
registrations in that they businesses protecting their own brands. At 
ICANN level, small businesses don't really appear because the IP 
constituency is focused on trademark and other IP rights. The bulk of 
brand protection registrations are small businesses, often without any 
registered trademarks or service markets protecting their own brand 
across the "must register" TLDs.

The next layer is made up of web developers. They host the domain names 
of their clients and develop websites for them but they are not 
primarily in the business of domain name registrations. Some of them can 
have a few thousand registrations. They may evolve into domain name 
registration/hosting businesses.

Once a hosting business has enough registrations, it may make the leap 
from being a hoster to being a registrar. With gTLDs, that generally 
meant becoming an accredited ICANN registrar. As country level mrkets 
have become increasingly dominated by their local ccTLDs in the last 
fifteen years, hosters are more likely to become ccTLD registrars for 
their local ccTLD rather than ICANN registrars. This shift was caused 
primarily by ICANN's failure to take action against large-scale Domain 
Tasting in the mid 2000s. People (including some of those that ALAC is 
meant to represent) found they could no longer get the domain name they 
wanted in .COM but could in their local ccTLD. Since then, the ccTLD has 
replaced .COM as the first choice TLD in most countries. The US market 
is the powerhouse of .COM and the legacy gTLDs. If it ever shifts 
towards .US ccTLD, then the legacy gTLDs will be in serious trouble.

When it comes to portfolio registrations and domain name investment, the 
percentage of these registations in most gTLDs is much smaller than 
people expect. This because people don't typically rely upon statistical 
data. As a group, the "for sale" domain names in a TLD have a high 
non-renewal rate. The auctions and resale part of the market has a few 
large players but there is a blurring of the lines between registars 
that park expired domain names on PPC with a "may be for sale" notice 
and genuine auction/sales operations. One of the larger auction/sales 
operations has 1.35% of the .COM market. Contrary to the purist view, 
having auction/sales sites in a TLD is an indication of the health of a 
TLD. While it is a source of friction, it shows which TLDs that end 
users are in adopting. The NGTs have been a bit of a disaster in terms 
of secondary market sales.

> -  Effects of vertical integration on industry concentration. Compare 
> degree of concentration in each of the  Regions.

Concentration by region is also problematic due to ICANN's broken 
registrars model. The top of the market is dominated by portfolio 
operators that operate in mutiple country/regional markets 
(Transnationals like Godaddy, EIG, United Internet, CentralNic etc). One 
of Godaddy's acquisitions was the Host Europe Group and that operator 
had been buying up hosting businesses and registrars throughout the 
European market. Other players like United Internet and CentralNic had 
also been acquiring smaller players.

When it comes to regions like Latin America/Carribean or Africa, they 
may not even move the needle when it comes to ICANN registrar 
marketshare but some of the countries will have their own thriving ccTLD 
markets. The markets in Africa seem to have skipped much of the Desktop 
Internet phase of development and moved instead to the Mobile Internet 
phase. It also means that many gTLD registrations will be hosted outside 
the registrant's country. This is often a function of economics and 
infrastructure development.

> -  effects of new IDN Registries on market shares in the relevant 
> language areas.

IDNs seemed to be a good solution but their uptake has not, even in the 
Chinese market, been spectacular or on anything like the scale that 
their proponents expected. What is interesting is that there has been a 
trend to use numbers in registrations in the Chinese market as the 
pronounciation of the numbers will be phonetically similar to that of 
various words.

One of the big problems with the CCT was that many of the metrics 
suggestions were from people who were unaware how difficult it was to 
get data for these metrics. ICANN's Domain Name Marketplace Indicators 
group has been working on developing usable metrics and it might be 
better to look at some of this work rather than indulging in a 
Golgafrincham-like effort on what colour to paint the wheels on the next 
round of new gTLDs.

Regards...jmcc
-- 
**********************************************************
John McCormac  *  e-mail: jmcc at hosterstats.com
MC2            *  web: http://www.hosterstats.com/
22 Viewmount   *  Domain Registrations Statistics
Waterford      *  Domnomics - the business of domain names
Ireland        *  https://amzn.to/2OPtEIO
IE             *  Skype: hosterstats.com
**********************************************************