Thursday, 21 December 2023

The shape of things to .com: An overview of domain availability

BLOG POST

As Internet usage continues to grow, certain aspects of the underlying infrastructure - notably IP addresses and domain names - are beginning to run short of capacity in key areas. In this study, we consider the availability of registerable alphabetic[1] domains, considering the .com extension (or top-level domain, TLD) - still the most popular by a significant margin - and short domain names across the set of around 1,000 other gTLDs.

For .com, over 99.6% of all two-, three- and four-letter alphabetic domain names, and even many of the remainder are reserved or otherwise unavailable. Just under one-quarter of five-letter names are already taken, though this includes the vast majority of dictionary terms. A similar comment holds true even for domains of greater length. .net and .org also have more than 99% of all possible three-letter domains names already registered. 

For two-letter domain names, seven different TLDs are at least 98% unavailable, and for one-letter domain names, there are 27 different TLDs for which all 26 possible options are already taken.

However, across the gTLD landscape in general, significant capacity does remain, with the proportion of registered domains across all ~1,000 extensions sitting at only 13%, 7%, and 3% for one-, two-, and three-letter domain names respectively, and even smaller values for longer domain names, providing a range of possibilities for prospective registrants. 

Following on from the analysis, the following points may be borne in mind by brand owners:

  • For registrations in accordance with the traditional preference for short, memorable .com domains, available options are significantly limited. Consequently, brand owners may need to resort to brokerage or (where IP protection permits) acquisition processes to secure their preferred domain names. Monitoring of third-party activity across the landscape of pre-registered domains, and new registrations, also remains key.
  • One associated recommendation for potential new brand owners is to select a longer, unusual and/or novel term for their brand name. This not only raises the possibility of the respective domain being available for registration, but also makes it possible to secure stronger intellectual property protection and makes the prospect of brand monitoring more straightforward.
  • Additionally, it may be wise for brand owners to consider TLDs other than .com for their primary website presence. A number of pre-existing TLDs are beginning to be 'repurposed' for alternative use, including .io (primarily for technology-related brands) .ai (for brands relating to artificial intelligence), .tv (relating to television or streaming services), and .co (as an alternative to .com for company websites). We are also seeing a continued series of new TLD launches as part of the new-gTLD programme, with a new round of applications set to launch in Q2 2026. Some brand owners may find it advantageous to consider applying to run a new dot-brand extension, giving them full control over all domains across the TLD in question. Failing this, utilisation of programmes such as the TMCH and registration-alert and blocking schemes can be an effective way of defending IP and receiving early warning of infringements[2]. As an associated point, brand owners should consider registering relevant domain names defensively across key TLDs, where they are available.
  • Finally, it may also be advantageous to secure IP within the emerging Web3 landscape. In particular, the blockchain domain ecosystem provides options across both generic extensions and dot-brands.

References

[1] i.e. those containing only the characters a-z

[2] https://www.iamstobbs.com/opinion/the-new-new-gtlds

This article was first published on 21 December 2023 at:

https://www.iamstobbs.com/opinion/the-shape-of-things-to-.com-an-overview-of-domain-availability

* * * * *

WHITE PAPER

Executive Summary

Key pieces of Internet infrastructure are beginning to run towards full capacity, in response to the rapidly increasing numbers of connected devices and Internet users. One such example is the set of available IP addresses, which is seeing a transition from the old ('IPv4') system, with 4 × 109 available possibilities, to a newer ('IPv6') infrastructure, with 3 × 1038 options.

Similarly, despite the essentially infinite number of possible variations, key areas of the domain name landscape (particularly short domain names across popular top-level domains (TLDs, or domain extensions)) are running short of available options for potential registrants. In this analysis, we consider the availability of alphabetic (i.e. containing only the characters a-z) .com domains generally (still the most popular extension by a significant margin), and of short domain names across the full set of other gTLDs.

The main findings of the study are as follows:

  • For .com domains, over 99.6% of all two-, three- and four-letter alphabetic domain names, and just under one-quarter of five-letter names are already taken, with several of the remainder also reserved or otherwise unavailable. Only two two-letter domains (out of 262 = 676), and only 44 three-letter domains (out of 263 = 17,576) are not currently present in the .com zone file.
  • After .com, the most 'full' namespaces are .net and .org. All three extensions are more than 99% full for three-letter domain names, with .net actually having even lower three-letter domain name availability than .com.
  • For two-letter domain names, seven different extensions (.com, .net, .org, .biz, .law, .amsterdam and .country) are at least 98% unavailable.
  • There are 27 distinct gTLDs for which all 26 possible one-letter domains are registered.
  • However, across the full gTLD landscape (the 1,078 extensions for which zone-files are available), there remains significant capacity, due to the large number of extensions available. Overall, only 13% of all one-letter domain names are already registered, 7% of two-letter domains, and 3% of three letter domains, with the proportions continuing to drop off as the domain length increases.

Part 1: Availability of domains on the Internet's largest TLD

Introduction

It has long been known that key pieces of Internet infrastructure are beginning to run towards full capacity. IP addresses, for example - the string of four numbers between 1 and 255 representing the location of any server or other device, usually written as XX.XX.XX.XX – have a total universe of 2564 (232 or 4.29 billion) possibilities which, particularly given the growth of the range of connected devices collectively known as the 'Internet of Things', will soon be insufficient for requirements. Accordingly, the intention is ultimately to transition to a new naming system (IPv6, rather than the older IPv4), in which addresses are most usually represented as X:X:X:X:X:X:X:X, where each 'X' is a value between 1 and 216 (written as 'ffff' in hexadecimal), giving 2128 (3.4 × 1038, or 340 trillion trillion trillion) possible addresses in total[1].

For domain names - the alphanumeric strings used as website addresses - there is a similar problem. Whilst the actual number of potential domain names (which can have second-level names (SLDs) - the part of the name to the left of the dot - up to 63 characters in length, and can consist of any alphanumeric characters and hyphens, without even counting non-Latin variants) is extraordinarily large (around 1098 just for 63-character names on a single domain-name extension), certain areas of the domain-name landscape are already essentially 'full'.

Despite the continued growth in the numbers of available domain-name extensions (top-level domains, or TLDs), .com remains by far the most popular choice for organisations and other entities, and shows the greatest number of live registrations by around an order of magnitude[2], currently sitting at over 160 million. Furthermore, for many use cases there is often a preference for short, memorable domain names. Many of the shortest .com domains are already registered (or otherwise restricted), leaving few options for new users, and meaning that in many cases the existing domains are considered 'premium' and can be traded for extremely high prices.

Only three one-character .com domains are currently in existence (including x.com - recently acquired by Elon Musk following Twitter's rebrand[3] - together with q.com and z.com), and the majority of one-letter names were explicitly reserved by IANA (the Internet Assigned Numbers Authority) in the early 1990s[4]. The vast majority of other short .com domains are currently taken, with many used by major corporations for their public-facing websites and e-mail infrastructure[5]. Many instances of multi-million dollar sales of two-letter domain names have been reported[6]. Even considering domain names up to around 5 characters in length, although not all possible names are taken, it has often been reported that the vast majority of dictionary terms are no longer available[7]. Generally, domains are offered on a 'first come, first served' basis, meaning that (depending on IP protection), brands may often need to resort to acquisition processes in order the secure their preferred domain[8].

In this study, we use zone-file analysis to inspect the .com landscape, in order to consider the availability of domains. All figures are correct based on the version of the .com zone file downloaded from ICANN's Centralized Zone Data Service[9] on 11-Sep-2023. For simplicity in this study, we look only at domains containing (Latin) alphabetic characters (a-z) (sometimes written as 'LLL', for three-letter domains, for example), although numeric domains are also popular in the domain-name industry, particularly in regions such as China, where their use can circumvent language barriers and particular numbers can have special cultural significance.

Analysis

Table 1 and Figures 1 - 3 show the total numbers of registered domains for each domain length, based on their inclusion in the ICANN .com zone file. In each case, these values are also expressed as a proportion of the total 'pool' of available domain names (where the total possible number of n-length domain names is 26n).

n (SLD length)
                          
Possible no.
                          
No. registered
                          
No. unregistered
                          
% registered
                          
1 26   3   23     11.54 %
2 676   674   2     99.70 %
3 17,576   17,532   44     99.75 %
4 456,976   455,325   1,651     99.64 %
5 11,881,376   2,744,780   9,136,596     23.10 %
6 308,915,776   5,779,635   303,136,141     1.87 %
7 8,031,810,176   6,828,656   8,024,981,520     0.085 %
8 208,827,064,576   8,482,121   208,818,582,455     0.0041 %
9 5,429,503,678,976   9,609,092   5,429,494,069,884     0.00018 %
10 141,167,095,653,376   10,658,182   141,167,084,995,194     0.0000076 %
11 3,670,344,486,987,780   10,963,997   3,670,344,476,023,780     0.00000030 %
12 95,428,956,661,682,200   10,843,645   95,428,956,650,838,500     0.000000011 %

Table 1: Statistics for .com domains of SLD length n characters

Figure 1: Total numbers of registered .com domains for each SLD length (n characters)

Figure 2: Total numbers of unregistered .com domains for each SLD length (n characters)

Figure 3: Proportion of all possible domain names registered for each SLD length (n characters)

For two-, three- and four-letter alphabetic domain names, over 99.6% of the available names are already registered. Just under one quarter of the five-letter domain names are taken and beyond this, although the absolute numbers of registered domains continues to rise up to an SLD length of 11 characters, the proportion of the namespace which is registered drops off rapidly, due to the exponential growth in the number of names available as the SLD length increases.

For the two-letter domains, only two (dm.com and jh.com) (out of 262 = 676) are not currently registered. With three letters, all but 44 combinations (out of a possible set of 263 = 17,576) are registered.

The following is a list of all three-letter strings which are not currently registered as .com domains:

  • baq
  • bfh
  • btz
  • bzg
  • ciz
  • eth
  • exu
  • fkd
  • gdy
  • hfh
  • ilq
  • jig
  • jrx
  • kgr
  • kkk
  • mag
  • ndq
  • njq
  • nnr
  • oys
  • pbq
  • pqk
  • pwe
  • qag
  • qgt
  • qvz
  • qzk
  • rfc
  • ruu
  • sfj
  • soe
  • sok
  • trc
  • ucl
  • wxa
  • xjz
  • xkd
  • xko
  • ykn
  • ykz
  • zig
  • zip
  • zkb
  • zkn

This means that all three-letter combinations beginning with 'a', 'd', 'l' and 'v' are taken. Amongst the unregistered strings, some have particular relevance. 'kkk.com', for example, most recently expired in October 2022 and was subsequently offered for sale via GoDaddy Auctions. By mid-November, the domain had received a high bid of nearly $100,000, before being withdrawn from sale and blocked following concerns about the domain's possible association with the Klu Klux Klan[10].

Conclusion

The analysis shows the very low availability of unregistered short .com domain names. This finding is of great significance for organisations looking to launch new brands with a website presence, meaning they may need to resort to purchases of pre-existing domains, utilise longer or unusual brand / domain names - or use brand variants or keywords in the domain name rather than just the brand name itself, or look to TLDs other than .com for their primary website presence. The domain-name industry is already seeing growth in the popularity of other domain-name extensions, such as .io, .ai, .tv, .co, etc. as a reflection of this fact. Also relevant is the ongoing new-gTLD programme, which has seen the launch of over 1,000 new extensions since its start in 2012, continues to generate new releases (with around a dozen in 2023, including .kids, .zip, .box and .music)[11], and has a new round of applications scheduled to begin in 2026[12]. These developments have significantly increased the available domain-name space, and we may also see a growth in the popularity of dot-brand extensions.

A related point to consider is that new companies may be wise to select novel or invented terms for their brand names. Not only does this raise the probability that the associated domain names will be available, but it also has the added benefits of being able to secure stronger intellectual property protection, and making brand monitoring more straightforward and less subject to the difficulties associated with the detection of 'false positives'.

Part 2: Availability of short domains across the gTLD landscape

Introduction

In Part 1, we considered the availability of .com domains of various lengths, finding that, for two-, three- and four-character alphabetic domain names, over 99.6% of the total universe of possible domain names are already registered. In Part 2, we extend the same ideas to look at the availability of short domains across the full range of gTLDs - the global top-level domains, or domain extensions (according to the zone files published via ICANN's Centralized Zone Data Service as of 11-Sep-2023, of which 1,078 were available).

Analysis

Figure 4 shows the proportion of the set of all possible domain names which are currently already registered, for each gTLD, for one- to six-character alphabetic domain names (i.e. those consisting only of Latin alphabet characters). The TLDs are sorted by the total number of one- to six-character registered domains.

Figure 4: Proportion of the set of all possible domain names which are already registered, for each of the top 40 gTLDs, as a function of second-level domain name (SLD) length (n characters)

A number of top-level observations are apparent:

  • Overall, .com is by far the most 'full' namespace, with 23.1% of all possible 5-letter domains and 1.9% of 6-letter domains registered (followed next by .net in both cases, with 3.7% and 0.2%, respectively).
  • For 3-letter domain names, .net, .com and .org are all more than 99% full; .net actually has even lower availability than .com (17,537 domains registered out of a possible 17,576, compared with 17,532 for .com).
  • For 2-letter domain names, the .com, .net, .org, and .biz namespaces are all at least 98% taken (in addition to .country (99.26%), .law (98.67%) and .amsterdam (98.22%)).
  • There are 27 gTLDs for which all 26 possible one-letter domains are registered. These extensions are: .biz, .ltd, .icu, .digital, .company, .wtf, .fyi, .cool, .run, .capital, .berlin, .law, .casa, .beer, .fashion, .hamburg, .wales, .srl, .country, .wedding, .cymru, .garden, .luxury, .irish, .esq, .abogado and .prof.

Considering all gTLDs together, it is also possible to calculate the total proportion of all possible domain names of length n which are registered (where the total number of possible names is (26n × T), where T is the total number of gTLDs - in this case, 1,078) (Table 2 and Figure 5).

n (SLD length)
                                
Possible no.
                                
No. registered
                                
% registered
                                
1 28,028   3,584   12.79 %  
2 728,728   51,322   7.04 %  
3 18,946,928   622,712   3.29 %  
4 492,620,128   2,394,695   0.49 %  
5 12,808,123,328   5,667,908   0.044 %  
6 333,011,206,528   9,549,608   0.0029 %  

Table 2: Statistics for all gTLD domains of SLD length n characters

Figure 5: Proportion of all possible domain names registered for each SLD length (n characters)

Conclusion

Although there are some subsets of the domain-name landscape which are nearing capacity (notably .com, .net and .org for two- and three-letter domains), the overall landscape is by no means full. Even for highly desirable three-letter domain names, only around 3% of all possible names are taken, when considering the full set of gTLDs. As discussed in the first part of the study, the likelihood is that brand owners may simply need to reassess their requirements when looking to acquire business domain names, and perhaps set their expectations away from the traditional .com environment.

References

[1] 'Brand Protection in the Online World: A Comprehensive Guide' by David N. Barnett - Box E.2: 'IP addresses'

[2] https://research.domaintools.com/statistics/tld-counts/

[3] https://www.iamstobbs.com/opinion/x-trademarks-the-spot-not-a-textbook-example-of-a-successful-rebranding-exercise

[4] https://www.quora.com/Why-are-there-no-single-letter-domain-names

[5] https://smartbranding.com/ll-type-domains/

[6] https://www.globenewswire.com/en/newsrelease/2019/04/29/1811388/9865/en/Coveted-Two-Letter-Domain-Name-Potentially-Worth-Millions-to-Auction-Exclusively-on-NameJet.html

[7] https://www.quora.com/Have-all-5-character-com-domain-names-been-taken

[8] https://nz.news.yahoo.com/world-running-domain-names-gone-130011203.html

[9] https://czds.icann.org/

[10] https://domaininvesting.com/godaddy-cancels-kkk-com-expiry-auction/

[11] https://www.iamstobbs.com/opinion/music-to-brand-owners-ears

[12] https://www.iamstobbs.com/opinion/the-new-new-gtlds

This article was first published as an e-book on 21 December 2023 at:

https://www.iamstobbs.com/availability-of-domains-ebook

No comments:

Post a Comment

Unregistered Gems Part 6: Phonemizing strings to find brandable domains

Introduction The UnregisteredGems.com series of articles explores a range of techniques to filter and search through the universe of unregis...