The legacies of long tail and the unfolding of consolidation and concentration in the top-level domain sector

ABSTRACT

This paper addresses the evolution of the generic TLD sector against the backdrop of changes introduced by ICANN with the expansion of the TLD space in 2012 and the relaxation of rules concerning structural separation between registries and registrars. From a business and technical perspective, the market moves that have taken place can be described as part of a long tail effect. This is relevant, for policy and theoretical reasons, to inform possible future rounds of new TLDs but also to integrate this issue into broader competition policy and diversity concerns. Has market concentration and consolidation increased in the DNS registry and registrar sector? What are the consequences of the new TLD programme from a diversity perspective? The work undertakes statistical analysis from several databases and desk research to develop a ten-year mapping. The potential implications of increased consolidation trends loom over the DNS registry and registrar field. The capacity to challenge some of these features in the TLD sector needs more than the prevailing market rules. Increased political commitments should be undertaken, at least to uphold geographic and linguistic diversity in this scenario.

KEYWORDS:

• DNS
• Long Tail
• ICANN
• TLDs
• consolidation

View correction statement:

Correction

1. Introduction

The consolidation of digital markets has been heavily debated over the last years. The increase of giant technological corporations, commonly known as ‘big tech’, has resulted in a revision of current policies and rules as their power has extended beyond their immediate environments to a global systemic level. One of the amendments to contain the growth of these tech giants is centred in antitrust and competition policy (Gerbrandy 2018; Haw Allensworth 2021). The current conceptual toolkit has been deemed to have failed to keep competitive markets in the digital sector. There have been a range of initiatives at the Federal Trade Commission in the US, bills at the US Congress, and the Digital Markets Act and the Digital Services Act in the EU, as well as in China and increasingly in Latin America (Aydin and Büthe 2016; Greco and Viecens 2021) that take a broad and public stance on different digital markets. Key concepts that have underpinned the antitrust environment are also being revisited by policymakers and scholars from different disciplinary environments, as the consolidation of digital markets and their infrastructure not only poses threats to competition, but also to broader economic development and democratic institutions (Allen, Scheve, and Stasavage 2023; Crane 2022; Drexl 2016).

The top-level domain (TLD) sector that represents the most visible part of the Domain Name System (DNS) has not been at the centre of these broader debates about consolidation and concentration of digital markets and power laws. Yet, since 2012, many changes have occurred when the new top-level domain programme was launched to open the internet’s root zone to accommodate more options to reflect a geographic, cultural, linguistic and commercial diversification of this part of the internet that engages directly with web content. That programme opened a new and momentous transformation in the web environment and in the approach of the Internet Corporation for Assigned Names and Numbers (ICANN), the multi-stakeholder organisation responsible for the policies and oversight of the DNS environment. The number of registries was multiplied by five and the number of ICANN accredited registrars roughly doubled. At a first glance, these figures point at a landscape of increased diversity, which contrasts with existing studies on other DNS-related infrastructures, such as hosting servers and DNS resolvers which have evinced greater consolidation and concentration patterns in recent years (Bates et al. 2021; Radu and Hausding 2020), a trend that is also clear in other digital markets.

This paper addresses the evolution of the generic TLD sector against the backdrop of changes introduced by ICANN with the expansion of the TLD space in 2012. From a policy perspective, the growth of the internet root zone from 300 extensions to over 1,600 in five years included revisiting some of the rules concerning structural separation prohibitions of vertical integration between registries and registrars. From a business and technical perspective, the market moves that have taken place in an increasingly diversified domain name space can be described as part of a long tail. But to be viable enterprises, these niche TLD services have resorted to bundled services, virtual integration strategies and larger conglomerates of registries and registrars, incurring merger and acquisitions strategies. Has market concentration and consolidation increased in the DNS registry and registrar sector? What are the consequences of the new TLD programme from a diversity perspective? This work examines the unfolding of the registry-registrar landscape against the concept of the long tail and competition policy in the DNS environment.

There is a knowledge gap concerning the evolution and implications of vertical and horizontal moves among DNS registry and registrars. This is relevant for policy and theoretical reasons. In the first place, this is because ICANN is aiming to implement a second round of the new TLD programme, with a specific focus ‘on diversity and inclusivity of the Domain Name System (DNS), including universal acceptance of new gTLDs as well as the inclusion of more IDNs’ (ICANN 2023). This second round should be able to reflect lessons learned and whether such measures should promote more diversity and inclusion as an intrinsic value of this initiative. Secondly, because it is relevant to connect how large players in the TLD environment are also offering more bundled services, including cloud and technical infrastructure (Bates et al. 2021; Wang et al. 2021; Zembruzki et al. 2022). Services become a key differentiator as the business models and incentives are not necessarily related to a domain name and as security measures are becoming more extended. Thirdly, because the conceptual and policy tools concerning competition policy are in tension with the evolution of digital markets, and the TLD sector has gone understudied in these broader conversations. As a predominant infrastructure of the internet environment, competition within the DNS and the TLD sector have remained confined and have been scantily integrated into such broader debates. Finally, it is important to revisit notions of diversity beyond the concept of consumer choice into competition policy in the DNS registry-registrar space, as a global symbol of multi-stakeholder internet governance. While diversity as a principle for greater inclusion and global outreach has been foundational for ICANN in the development of registry and registrar separations and in the new TLD programme, its potential to enhance more participation and legitimacy in internet governance should be further explored.

The research strategy has embarked on a two-pronged approach based on a thorough review of existing scholarly, government, sectoral and industry reports and documentation, and on the analysis of data provided by ICANN though the registry reports and its Open Data Portal, and the platform developed by the DNS Research Federation, DAP.live. The study is based on a mixed-methods analysis involving quantitative and qualitative data which involves the triangulation from different data sources on the domain name sector, as well as corporate reports and industry documents to map the evolution of the industry from a historical perspective.

The paper is structured in five sections. The first sets the scene around the unfolding of the TLD sector. The second develops the main concepts that underpin the problems that frame this research and that emerge from business, political economy of communications and legal approaches. The third presents the data sets and methodology. The fourth presents empirical findings based on descriptive statistical analysis concerning different metrics on the market share evolution of gTLD registries and registrars between 2012 and 2022 linked with long tail analyses. The fifth discusses mainly the policy implications from the findings described in the previous section. The concluding section delves into the implications of this study for the DNS registry and registrar sector and reflects on diversity as a key structuring principle for future governance of this environment.

2. The evolution of top-level domain competition policies

Top-level domains have played a pivotal role in the evolution of the global internet governance regime. As semantic identifiers of the DNS, the roll-out of TLDs expanded political, commercial and legal tensions over the internet and its governance. Domains were defined as ‘administrative entities’ in the RFC 920 that described the domain requirements (Postel and Reynolds 1984). When the first domain names began to be delegated in the 1980s they were originally divided into two broad categories, the country code top-level domains (ccTLDs) and the generic top-level domains (gTLDs). The ccTLDs comprised a space reserved for domain names identified with specific countries, territories or jurisdictions based on the two-letter code of the English version of the alpha-2 specification defined by the ISO 3166. The other category, that of generic, was meant to ‘provide an organization name that is free of undesirable semantics’ (Postel and Reynolds 1984). These initial gTLDs defined at the time were .com, .edu, .gov, .mil and .org. Since then, the evolution of gTLDs has expanded. Firstly, on a case-by-case basis (1999–2011), and then through the first large-scale programme to expand the gTLD space in 2012.

The creation of ICANN in 1998 marked a point of inflection for the market and policy rules of TLDs, and for the internet governance regime more generally (Mueller 2002). The political economy of ICANN both shaped, and was moulded by, the monetisation of domain names under the gTLDs. They became the source of revenue that sustained this multi-stakeholder body and ICANN turned into the global regulator of the DNS TLD space. ccTLDs have remained largely outside the scope of ICANN’s regulation. While abiding with the technical compliance to guarantee the root zone’s integrity and RFC 1591s mandate to serve their national community (Postel 1994), and contrary to gTLDs, they have no blanket obligations concerning fees, sales channels or dispute resolution processes. Even though many ccTLDs have contractual obligations with ICANN that resemble those of a gTLD, generally they have an ad hoc contractual relationship with this body, and they have idiosyncratic policies concerning their services to their local communities. These are the two reasons why they are not considered in this paper.

In 1998–1999, when ICANN started to function, the creation of a more level playing field for the domain sector took place from the onset to counteract the dominance of Network Solutions, the dominant player of the TLD space which, at the time, was the.com registry. ICANN enforced accreditation requirements to registrars in the gTLD space to create the conditions for a level-playing field (Mueller 2002). Since the beginning of ICANN, the distinction between registries and registrars has shaped the policy processes of TLDs and the domain name value chain. The powerful registries that had generated so many tensions around the commercialisation of the DNS and the TLD space during the mid 1990s were a threat with their market and technical power, and this menace would condition the DNS regime around ICANN for almost the next two decades.

Between 1998 and 2011, a strict functional separation rule was in place where registries could not own registrars, while the opposite held true, i.e. a registrar could be entitled to own a registry. When structural separation was advocated for registries, this implied that two different legal and corporate entities oversaw the registry services and those of the sales and distribution, as two distinct sides of the value chain of the DNS market. While a registry could have the delegation of more than one top-level domain (for e.g. .com and .net as is the case of Verisign Registry), registries were not involved in the sales of those domain names. Registrars, that is, organisations (most of them for profit) would be responsible for the marketing and sales of domains, based on the registries’ policies for the use and restrictions that applied to different TLDs.

From 2011, a strict functional separation rule was changed to address this possibility in a case-by-case scenario where a registry could acquire a registrar, in preparation for the introduction of the new gTLD programme in 2012. This programme was envisaged as a democratisation of the gTLD space. The rise in the numbers of registries entailed a redistribution of power which placed them in a weaker position, as they simultaneously had to address the competition from other TLDs and to seek registrar attention for their names (Kruger 2014). The ultimate end of the new TLD programme was to offer ‘enhanced innovation, competition and consumer choice’ (ICANN 2012).

The removal of cross-ownership restrictions had been requested by the ICANN board to the CEO (ICANN 2010a). With this measure, the ICANN board was hoping to eliminate barriers that would hinder registries in competing with registrars, who were, in their turn, planning to operate new TLDs and this in way to ‘level the playing field’ for both sides as well as to produce ‘positive community impact’ (ICANN 2010b). In June 2011, Lawrence Strickling, as the assistant secretary for communications and information of the US Department of Commerce sent a letter to the chairman of the board of directors of ICANN concerning the proposal to make wholesale changes to restrictions on cross-ownership of registries and registrars for existing and new TLDs. The letter annexed a recommendation issued by the Antitrust Division of the US Department of Justice, at the request of the US Department of Commerce, for advice regarding competition issues associated with resolutions by the board of ICANN that would allow for cross-ownership or vertical integration of registries and registrars of existing and new generic TLDs.

The main concerns that emerged from the 2011 Antitrust Division Report had been already developed in a report commissioned by ICANN by CRA International (2008) that pointed at the problem of cross-ownership, particularly with regards to registry discrimination among registrars, especially when faced with price caps. The letter points more squarely to a concern that was developed previously on structural separation issues, as cross-ownership ‘may allow a registrar or registry disadvantage its rivals, foreclosing competition and harming registrants’ (US DOC 2011). The recommendation from the Antitrust Division is framed around the idea that vertical integration should be forbidden, unless carefully considered within the existing multi-stakeholder sectors. The report ends with the recommendation that ‘the current structural rule that prohibits vertical integration provides a more effective and easier way to prevent competitive abuse’ (US DOC 2011), rather than developing market power studies and a complex multi-stakeholder organisation to assess these issues in a case-by-case fashion. These recommendations were not strictly upheld by ICANN and the possibilities of vertical integration were gradually opened to incorporate the deep changes that would come about with the new TLD programme.

3. Conceptual framework

This section addresses some key issues concerning competition law and policy and how these debates have been reflected more recently in digital markets and their bridges with the DNS registry and registrar marketplace. The second part examines market and technical implications of the long tail in the DNS.

3.1. Competition law and policy debates on functional and structural separation in digital markets

Structural separation is a long-standing concept in antitrust and competition law. In the last two decades these measures have been being reconsidered in policy and scholarly debates as one of the pathways to address current market power imbalances (OECD 2019), and particularly in the platform economy as a mechanism to atone for discrimination and information appropriation (Khan 2019). Structural separations proscribe certain organisational structures and place limitations on the business lines in which a firm can engage, rather than prohibit business practices. Structural remedies are contrasted with behavioural ones in competition policy: ‘Whereas behavioral remedies seek to prevent firms from engaging in specific types of conduct, structural remedies seek to eliminate the incentives that would make that conduct possible or likely in the first place’ (Khan 2019, 982). This is in line with the recommendations issued by the US Department of Justice in the previous section. From the perspective of economic efficiency, there are three reasons to pursue structural separation: the restoration of competition or innovation that has been harmed by conduct or a transaction; the deterrence of future anticompetitive conducts and, finally, the opening of markets to competition and innovation (Melamed 2009).

Structural separation has several possibilities. One way of unpacking it is through the vertical and horizontal distinctions. Horizontal relationships are those established amongst competitors that produce a similar good or service, as is the case among TLDs. A vertical relationship is one where the components in the production are complements, as is the relationship between registries for generic or new TLDs and registrars. There is a wide spectrum of potential degrees of separation between integrated monopoly and full ownership of competitive and non-competitive components within a sector. A key assessment is whether the elimination of competitors (a horizontal concern), increases market power within this space through increased barriers to entry (European Commission Directorate General for Competition 2019).

Another relevant difference for this study is between structural and functional separation. The first refers to the division of operations into separate companies, while functional separation allows operations to be conducted by subsidiaries of the same company, subject to restrictions on the transfer of information, intermediate products, or services between the subsidiaries (Gilbert 2021). DNS registries and registrars have typically operated under functional separation when it comes to mergers and acquisitions allowed after 2011, that is, the registry and registrar functions remain distinct within the organisational structure.

The traditional concerns posed since ICANN’s creation until the new TLD programme in 2012 were founded on the basis that separation helps to maintain a clear division of responsibilities and ensures that the registry remains impartial and provides equal access to all registrars. Mergers and acquisitions by large players of small TLD registries and registrars, some with a promising user base and significant competitive potential, have been predominant in the last decade since the new TLD programme, both employing vertical and horizontal integration strategies.

3.2. The long tail and the DNS

The concept of the long tail is derived from Chris Anderson's Wired magazine article from 2004 and later book, The Long Tail: Why the Future of Business Is Selling Less of More (2006), which explain how combined sales of many niche products or services can exceed the sales of a few popular products. The notion was published and widely discussed during the years of the expansion of the internet, when digital marketplaces and platforms were beginning to mushroom. It addressed a well-known concept: the Pareto Principle, coined in the early twentieth century. This principle is used to describe the phenomenon of sales concentration and is also accepted as the ‘80/20’ rule. It states that a small proportion (20%) of products in a market often generate a large proportion (80%) of sales (Brynjolfsson et al. 2011). The long tail describes the phenomenon where niche products can grow to become a large share of total sales, popularising one of the many power laws which have been discussed as the internet has expanded to more markets and services (Hilbert 2014). Anderson (2004) proclaimed three rules for the new entertainment internet economy: make everything available; cut the price in half and lower it; and finally, help to find it. This last rule was possible thanks to search engines and platforms that were beginning to populate the internet during the early 2000s and is especially relevant for the TLD sector where keywords based on different attributes are paramount.

For the last two decades, many of the rules concerning digital markets have been assessed against the long tail (Brynjolfsson et al. 2011). Other studies have criticised this concept as inadequate, as it does not contemplate the prevalence of vertical integration as a defining characteristic of some industries, in particularly the media (Napoli 2016). In this paper, the long tail is used as a heuristic concept to improve the understanding of the socio-technical arrangements and the business incentives of the new DNS registry and registrar environment after the new TLD programme.

The long tail has been used to illustrate the effects of bundling niche products and how they inject greater diversity into a marketplace. At the same time, it has been proclaimed as a business strategy, fostered by small and big dominant players, including companies such as Amazon, that have spurred on the bundling of services over millions of niche products. In the domain name sector, the long tail suggests that while there are many high-demand and popular domain names (such as single-word or highly descriptive domains), there is also significant value in the vast number of niche or specific domain names that cater to smaller audiences. This was the original claim, and a main objective of the new TLD programme.

Brynjolfsson, Hu, and Smith (2010, 5) noted that the long tail can seemingly point at contradictory outcomes, depending on how it is addressed. The absolute long tail refers to changes in the total numbers of products that are sold. The second is the relative long tail, which focuses on the relative share of sales above or below a certain rank, where the classic Pareto Principle applies. A third refers to the exponent, that is, the slope of the log-linear relationship which provides an indication of the relative importance of the head versus the tail of the sales distribution. The empirical analysis of the work addresses the three dimensions.

The long tail’s focus on serving niche markets shares a common conceptual ground with the foundations of the new TLD programme. From a long-tail theory applied to DNS it suggests that while popular and high-demand domain names may command higher prices and attract significant attention, specialised segments provide relevant business opportunities. The new TLD programme was focused on addressing unmet needs based on identity, branding and/or broader community perspectives for a distinct online presence (ICANN 2012). Many of these needs were catering for specific communities or markets. Another shared rule is the use of internet search and discovery tools linked with an increase in the share of niche products. More generally for the internet environment, the political economy of search is vital (Mosco 2017; Van Couvering 2011; Wilken 2017). A third is availability; given the vast number of potential domain name combinations and TLDs, many long-tail domain names remain available, even after popular ones are taken.

But the existence of niche products does not unequivocally explain the diversification of the market. Consolidation and concentration help to explain the success of some services when they become bundled or aggregated (Brynjolfsson et al. 2011; Mosco 2017). In the case of the TLD sector, these are bundled with other types of service, such as hosting, security, alerts and web design, among others. This bundling is mainly accomplished through virtual integration, defined by the substitution of ownership with partnership by integrating a set of suppliers through information technology (IT) for tighter supply-chain collaboration (Wang et al. 2021). Virtual integration relies on cloud infrastructure for the DNS. The development of critical capabilities to scale and address current growth and security needs, including DNS registries and registrars that increasingly rely on these types of services provided by firms such as Google, Amazon, Akamai, Dyn and Cloudflare increases interdependencies between these two sets of players. Amazon and Cloudflare exclusively host the name servers for 22.5% and 18.7% of domains, respectively (Wang et al. 2021). A similar pattern is evinced with the evolution of external DNS hosting that rapidly overtook self-hosted DNS in the period, and the percentage of domains managed entirely by external DNS hosting providers grew from 32.9% to 65.7% between 2011 and 2017 (Bates et al. 2021).

There are natural barriers to entering the cloud market due to the large upfront investments that are needed. The scope, quality and security costs of setting and maintaining these services are costly and complex. These cloud services are becoming increasingly concentrated (Bates et al. 2021; Radu and Hausding 2020). Virtual integration and the reliance on expensive infrastructure, the ‘technical competence’ (Van Couvering 2011, 10), becomes an additional lens to understand the unfolding of consolidation and concentration as part of the long tail strategies of the TLD environment. One possible explanation for the consolidation of domain names and DNS networks is owed to complex, costly and resource-intensive processes. As digitisation becomes more pervasive, the increased security risks via exploitation of DNS services become an attractive space to manipulate vulnerabilities (Kim and Reeves 2020). The area of digital security that covers incidents disrupting the availability, integrity and confidentiality (the ‘CIA triad’) of parts of the DNS ecosystem (OECD 2022), or those practices that use the DNS for malicious activities, including the repertoire of DNS abuse practices (Korczyński et al. 2018), provide incentives for users to rely on well-established third parties to provide DNS resolver and registration services. This is more critical among users and organisational environments that have less knowledge, expertise and dedicated teams (Chandramouli and Rose 2013). Streamlining and simplifying DNS network management registration services becomes a selling point concerning network and data security (Sherman 2020; Taylor 2021), with potential implications with TLD consolidation.

4. Data section

This part describes the approaches concerning the development of the data sets for their analysis. The main data source is derived from ICANN’s Root Zone Archive and Open Data Portal. This policy has enabled researchers and organisations. such as the DNS Research Federation that has developed its own API, to access relevant data from the most authoritative source concerning generic TLD statistics as part of ICANN’s mission. This work has been based primarily on the DAP platform developed by the DNS Research Federation (DFNSRF) which for top-level domain statistics, including registry and registrar indicators, is based on ICANN’s open data for the years 2012–2020. For 2021–2022, the author used ICANN’s monthly registry archival data to develop the database for these two years as ICANN has not yet opened this data in the Open Data Portal. To develop this, the author considered December’s transaction reports for each TLD, accounting for 2,418 observations. Slight differences may be accounted for with the yearly registrations from 2012–2020 from the DAP platform developed by the DNSRF, as this averaged the registrations for each TLD. Despite these different approaches for the development of the databases (Figures 2 and 3 and Table 1) these are minor differences and have no effects on the overall position of the different TLD registries nor their market share, which is the main objective of this study.

Table 1. Domains under management top 10 TLDs.

2012		2020		2022
.com	105.937.945	.com	152.373.883	.com	164.413.895
.net	15.099.633	.net	13.689.848	.net	13.502.970
.org	9.700.000	.org	10.652.302	.org	11.170.010
.biz	2.303.447	.icu	6.178.600	.xyz	4.563.254
.mobi	1.037.343	.info	4.688.275	.info	3.966.719
.tel	287.294	.top	3.356.859	.online	2.345.820
.asia	236.185	.xyz	3.182.692	.top	2.124.008
.name	227.684	.site	1.951.838	.shop	1.527.515
.pro	146.143	.online	1.636.875	.biz	1.434.749
.xxx	136.548	.biz	1.503.275	.site	1.296.714

Source: DNSRF DAP.live and ICANN transactions report.

The research has used other industry and non-profit organisations’ statistical reports, such as those produced from AfNIC, CENTR, the Domain Name Industry Brief and nTLDstats to validate the positions of the different TLDs in different rankings and to cross-reference information concerning DNS industry moves related with mergers and acquisitions of registries, registrars and back-end registry operators (Figures 4–6 and Table 2).

Table 2. Main registry groups.

Registry company	Total Number of TLDs	Estimated TLD market share (2022)
Amazon Registry Services	54	≤1%
Charleston Road Registry Inc (Google)	45	≈2%
Identity Digital	264	≈3%
Radix	10	≈3%
Registry Services LLC (GoDaddy group)	34	≈2%
Verisign (group)	16	≈82%
XYZ LLC	34	≈2%

Source: ICANN Registry Reports. The name of the company is the commercial name, not necessarily the IANA administrative contact of the TLD.

Registrar data (Figures 4 and 5) has been used from the DNSRF API, DAP.live, which is in turn based on ICANN’s Open Data Portal and registry archive. For 2021 and 2022, the data was analysed from the IANA Root Zone Database.

5. Findings

This section presents the findings based on statistical analysis from several data sources: ICANN’s registry reports, ICANN Open Data Portal and the platform developed by the DNS Research Federation (DNSRF), DAP.live. It maps the evolution of the DNS registry and registry sector between 2012, the year the new TLD programme was launched, and 2022. For some figures, it addresses an intermediate point of data in 2020 as, by that time, the new TLDs had all been running for some years, and because in 2020–2021, several consolidation measures took place. This information is complemented with industry reports that show different corporate strategies concerning horizontal and vertical mergers and acquisitions.

5.1. The long tail in the gTLD sector

Since ICANN received 1,912 applications for new generic TLDs in 2012, the landscape of internet registries and registrars has changed from a scenario of relative scarcity of TLD extensions, to one of abundance, with legal, policy, technical and business model implications. The first impression is that the new TLD programme increased diversity. For example, geographic TLDs; a top-level domain name category created by ICANN denoting geographical, geopolitical, ethnic, social or cultural representation was enlarged from three cases before 2012 (.asia, .cat and .eu) to 66 new applications in the new TLD programme (ICANN 2023). (Several sector reports sometimes consider ccTLDs as geographic too, but that discussion is not part of this study).

While the nearly 300 TLDs (considering ccTLDs and gTLDs) that existed prior to the new TLD programme may not have looked like a landscape of lack of choice, .com domains represented approximately 50% of the registrations. In addition, the existing TLDs were all in ASCII script (Latin alphabet), used in most Western countries but leaving hundreds of millions of internet users without a DNS identifier based on their own language (OXIL et al. 2017). The new TLD programme and the inclusion in the root zone of over 1,200 extensions after the review process casts a landscape of diversity, which may point at the existence of a ‘long tail’ when considered in absolute terms (Brynjolfsson et al. 2011), as it refers to changes in the numbers from the supply-side. In the decade between 2012 and 2022, the supply of TLDs rose fivefold with the new TLD programme. But from a demand side, the total number of domains names increased by 40%, from 252 million in 2012–350.4 million in 2022 (DNIB 2012, 2023). This means that from a demand perspective, the number of domains under management apportioned to any TLD has decreased.

Figures 1 and 2 depict the presence of the long tail in the gTLD sector. In Figure 1 for 2012, before the new TLD programme was launched, there was a marked and steep head, led by the legacy gTLDs .com, .net and .org. Figure 2, from a decade later (2022), maps the diversification of players in the head of the chart, led by the same legacy TLDs from Figure 1 and complemented with new TLDs as .xyz, .online and .top, as well as legacy TLD.info (Table 1). The market share of the top five TLDs in 2022, i.e. the head of the chart of Figure 2, has shown that the slope is less steep than in 2012 and that there are more players in the head. Figures 1 and 2 develop a more nuanced perspective on the evolution of the market share based on the slope of the log-linear figure. These charts unveil differences between the head and the tail of market share distribution of these TLDs. In Figure 1, the head of the market comprising 99% of domains under management was accomplished by five TLDs during 2012 (.com, .net, .org, .biz and .mobi) respectively, while in 2022, 27 gTLDs comprised 99% of the market (Figure 2). From a long tail perspective, Figures 1 and 2 underscore the coexistence of a slightly increasing diversified market from a supply-side that may potentially coexist with ‘superstars’. But it is problematic that the remaining 1% of the tail comprises the largest volume of players.

Figure 1. Source: ICANN Open Data, DNSRF DAP.live.

Figure 2. Source: ICANN Open Data, DNSRF DAP.live.

Figures 1–3 all still show a highly concentrated market dominance of one TLD (.com). While the.com has historically been the leading TLD (in both the generic and country code sectors), the charts point at a specific relationship between the power structure of this dominant player and the rest of the market.

The long tail and the superstar effects of dominant players such as .com and .net represent two sides of a common set of problems about product variety and concentration. In 2012,. com had 78% of the market share, followed by. net with 11% and .org (7%). In 2020, the gTLD market share of .com had fallen to 72%, followed by .net with 6% and.org with 5%. Two years later (2022), .com regained some market share (75%) and that of .net and .org had remained the same. The relative share of sales points at a moderate increase of the long tail phenomenon, as niche TLDs had increased their market share over a ten-year period.

An increased number of TLD players have surfaced amongst the top and is clearer in Figure 3, which examines the relative long tail as it unpacks the market share of the top five TLDs over 2012–2022. A clear trend cannot emerge at the time of writing this work, as 2021 results have shown some rebound of.com (Figure 3), but from 2012 to 2020 there is a marked trend towards a more diversified market share amongst the largest TLDs. Between 2020 and 2021, .com increased its market share among all generic TLDs from 69% to 75%. This can be partly explained by losses from the new TLDs in 2020 (1%) and 2021 (9%), a trend that stopped in 2022 with a 7% increase in stock (AfNIC 2023).

Figure 3. Market share of gTLDs (in %) of top 5. Source: ICANN Open Data and registry Transactions. DNSRF DAP.live.

While the primacy of the three most relevant legacy gTLDs (.com, .net and .org) has been sustained over the decade since the new TLD programme, the top five and top ten gTLD registries have changed with the inclusion of new TLDs that have managed to become relevant alternatives (Table 1). None of the top 30 TLDs in 2022 belonged to the geoTLD category.

5.2. Consolidation and concentration in the DNS registry-registrar space

The TLD sector has undergone persistent signs of consolidation and concentration at the registry and registrar level, as well as in the technical operators that operate the back-end services. Table 2 develops the main ‘registry groups’ in terms of the size of their TLD portfolio and their respective market share from a global TLD perspective (i.e. gTLDs and ccTLDs as there are examples of the latter administered by these large registry companies). These seven organisations control 95% of the gTLD market share and together they group 457 TLDs, approximately 36% of all gTLDs.

Identity Digital is the company that resulted from the merger of Afilias – including .info, with nearly 4 million domain names in 2022 – and Donuts. The company comprises 240 new TLD extensions, the vast majority as niche domains, though a few are among the top 30. Though Amazon has 54 TLDs under management, it does not have a significant market share, as its TLDs have been purposed for internal products of the company. Google participated in the new TLD programme with close to 80 TLD applications and has 45 TLDs in its portfolio, with two among the top 30: .app and .dev. XYZ LLC manages.xyz, the most successful new TLD in terms of registrations among the top five gTLDs and above many of the legacy ones (Table 1). Registry Services is an entity belonging to the GoDaddy group and has four gTLDs in the top 30 in its portfolio, with .biz as a legacy gTLD incorporated with the acquisition of Neustar Registry. Verisign, with .com and .net, has created different legal subsidiaries for different TLDs and has over 82% of the global overall TLD market share. Radix is the only major registry company that is not in the United States. It runs .online, .site, .store, .fun, .tech, and .website and comprises a new TLD portfolio with several extensions among the top 30 gTLDs.

Bundling and consolidation unfolds differently for registrars. Figure 4 shows an expansion of registrars which took off when the new TLD programme became implemented. Yet the spike from 2014 to 2015 can be better understood as the phenomenon of ‘pop-up’ companies (Davis 2016) – which are, in most cases, asset-light firms (Narayan 2022) – that allows for both registries and registrars to offer services in the tail.

Figure 4. Number of ICANN accredited registrars per year based on DNSRF DAP.live (2020).

In 2020, GoDaddy operated nearly 63.5 million domain names, which represented nearly 19% of the domain name market (Figure 5) and a portfolio comprising over 412 different gTLDs (GoDaddy 2022). This registrar reported 24 major acquisitions since 2013, ranging from web hosting to domain marketplaces, to the purchase of other registrars and notably Neustar Registry during 2020 (GoDaddy 2020).

Figure 5. Percentage of domain name market share top 20 registrars based on DNSRF DAP.live (2020).

Figure 5 captures the largest 20 registrars in terms of absolute domain name registrations and the long tail at work in this segment. The aggregated market share of the top 20 registrars is 44.3%. It appears to be a much more diversified sector than those of the registries, including when viewed with a geographical lens. While the top registrars are primarily from North America, there is more geographical representation from Asian and European registrars in the list.

Figure 6 captures some of the complex technical interdependencies between smaller and large registrars that point at a higher level of concentration at the back-end services. Three companies provide third-party back-end platforms for other registrars, covering 70% of all ICANN accredited registrars. Namebright provides services for 49% of the 1,252 ICANN accredited registrars, Central NIC for 19% (475) and Tucows for 2.5% (67). The remaining 714 registrars, i.e. the 28% that comprise ‘Other’ represent those that have technical and business independence and that do not have third-party service provision, including GoDaddy. Several of these firms are ‘asset-light’ registrars that have agreements or act as subsidiaries of the larger ones.

Figure 6. Absolute market share back-end platforms registrars based on IANA Root Zone Database (June 2023).

6. Discussion

This section discusses some of the main findings and delves into some policy implications and recommendations concerned with diversity and consolidation in the TLD sector.

Though variability has occurred in the tail of these markets (Table 1), a persistent trend is that the largest three TLDs – .com, .net and .org – are all legacy extensions that have managed to remain at the head with a strong market share, especially in the case of.com. Among the factors that explain this predominance, habit or ‘the default’ is crucial. There is a long tradition of research that has pointed to the role of informal processes that have ‘standardized’ understandings around technologies and applications (Blind and Thumm 2004; David 1985). The.com fits into this category as it has become the default for most commercial ventures online. Its existence predated that of new TLDs by nearly three decades, with a consolidated base of over 100 million domains. A second issue for this permanence at the head of the chart of these legacy TLDs may be related to the perception of their price stability and predictability (Lehr, Clark, and Bauer 2020). This is despite the policy changes introduced by ICANN, which eliminated most price caps on legacy TLDs, in line with the approach undertaken with new TLDs, and the renewed US Department of Commerce Contract agreement with Verisign (2019), which have increased the prices of domain names in these TLDs. At the same time, some of the larger new TLDs have developed different approaches to pricing that may potentially have had effects on registrations and renewals, though this has not been captured with the data in this study.

ICANN has relaxed its role of price regulator since 2019, and the new registry agreement between Verisign and the Department of Commerce has allowed the company to introduce price hikes, which it started in 2021. It is to be expected that some registrants will seek alternative extensions in other TLDs. This trend, which emerged in 2022 as the top five gTLD lost market share, may help to partially address this. But price may also become a factor to achieve a different business model for many TLDs, based on uniqueness and value-added services, rather than on volume. Price controls become significantly more important, particularly in the context of ICANN’s flexibilization on vertical integration, since such companies can manipulate prices in a way that may harm competitors who are not vertically integrated, and economic theory predicts an incentive to do this (IRP et al. 2022).

The increase in the supply of TLDs with minimum market share in the tail of the chart, which is the case with hundreds of open new TLDs affects the mid-term sustainability of this sector in the case of open TLDs that rely on external registrants (AFNIC 2023). This has implications for the future round of TLDs, both in what should be prioritised in ICANN’s guidelines as well as in consideration of proposals that prove less onerous for non-profit organisations, governments and SMEs. As some literature has pointed out, the projected horizon of viability for many of these new TLDs is a decade, and the investment may be too large and far away in time for smaller actors. Halvorson et al. (2015) also point out that only half of the registries have earned enough to cover their application fees, and 10% of current registries likely never will, solely from registration revenue. Without a prospect for a sustainable future, smaller registries may be prone to face more cybersecurity threats as they will have fewer resources to invest to strengthen them. Another scenario is that these signs will not stimulate independent and small initiatives for the next gTLD round.

While this study is not yet able to trace the impact of the acquisition of the Neustar registry by GoDaddy, nor those of Identity Digital’s portfolio, nor other vertical and horizontal integration moves, these types of transactions (Table 2) may potentially undermine ICANN’s legitimacy and neutrality when faced with a second round of new TLDs that is expected to bring in more diversity. While ICANN is an organisation that is bounded by US laws on matters concerning competition and antitrust, the scope and effects of the TLD marketplaces are international. Not all national jurisdictions have refined competition norms, and it is even more difficult to address concerns for digital markets (Newman 2019). TLD market competition issues should become more centrally part of the DNS sector and constituencies that participate at some ICANN’s advisory committees and supporting organisations as its effects may jeopardise diversity and inclusion.

The underlying factors for the changes within the new TLDs and their respective market share may be accounted for by pricing strategies undertaken by registrars (Halvorson et al. 2015). But there is not enough evidence to sustain the positioning of these TLDs only against the variable of price . Some of these TLDs have been given free in bundled agreements with other services. As portrayed in Table 1, there have been significant shifts within the new TLDs getting into the top ten. Studies concerning the lifetime of domain names have pointed out that new TLDs tend to have a greater prevalence of one-year registrations than legacy TLDs (Affinito et al. 2022). A case in point is.icu, which achieved notoriety in 2020 during the first year of the COVID-19 pandemic, but fell rapidly from the top ten in the year after.

Some of the fluctuations accounted for in overall domains under management in Table 1 for new TLDs are perceived in the actual use of these registration statistics, which have a greater speculative and defensive motivation (Affinito et al. 2022). Only 15% of domains in the new TLDs show characteristics consistent with primary registrations, while the rest are promotional, speculative or defensive in nature and where 16% of domains with NS records do not yet resolve, and 32% are parked (CENTR 2022). While these characteristics affect all TLDs, new TLDs are more prone to these practices and more effort should be developed by the TLD and its environment, including ICANN, in the development of clear policies as well as in outreach, so that users understand the special features of these TLD extensions.

Concerning the security dimension with DNS abuse, the registration of domains to spread malicious content is an issue that needs to be factored both by ICANN and TLD applicants as one of the barriers for new TLDs. Korczyński et al. (2018) have shown that new TLDs are more prone to be used for malicious activities. That study has underscored that unrestrictive registration practices, low registration pricing and the possibility of bulk domain registration are lowering the barriers for abuse. These are relevant features that highlight the role of policies as crucial for shaping some of the outcomes concerning DNS abuse rather than a focus on technical and infrastructural dimensions which, ex ante, may be considered primary causes for DNS abuse. This may have potentially relevant implications as the dependencies on DNS infrastructure may be modularised both technically, and from a business perspective, and may partially help to offset greater consolidation. It may also help ICANN to (re)formulate the costs of new TLD applicants for the second round, as well as to refine its policy guidelines.

7. Conclusions

This article assesses the unfolding of concentration and consolidation in the DNS registry and registrar sector. It maps these vertical and horizontal integration trends in the DNS sector and discusses them against the long tail and competition policy literature, as well as in recent policies developed in the ICANN environment for the launch of the new TLD programme.

The long tail and competition refer to different, but complementary, market dynamics. The long tail recognises the value of supplying a diverse range of products or services, even those that serve niche markets, as was the case of the new TLD programme. The benefits of removing traditional constraints on product availability, such as limited shelf space, is aligned with the idea that competition can be enhanced by enabling access to a broader range of choices, including those catering to niche preferences, which were part of the aims of the new TLD programme of 2012 and the forthcoming version.

Yet the presence of a long tail does not necessarily hinder concentration and consolidation, as this study has portrayed. In many digital markets, there is a combination of both phenomena (Schneider and Bauer 2015). In the DNS registry and registrar sector, the market is concentrated in a few dominant players that capture a significant market share of overall sales. At the same time, there is a long tail of smaller, specialised sellers offering niche TLDs that have a very low volume of domains under management and low growth rates. The hundreds of cases comprised in this very long tail of generic TLDs that are open to the public for registration would be unsustainable unless these registries develop strategies to bundle their services, become virtually integrated or become acquired by larger registries. This raises a general dilemma where increased diversity comes at the expense of sustainability or profitability and may further trigger consolidation processes. In addition, DNS abuse is more prevalent with new TLDs rather than with legacy ones.

The recent moves of consolidation, mergers and acquisitions in the DNS registry and registrar are symptoms of a market place that faces challenges to accommodate so many players in the tail, without being able to offer better sustainability options for the ultra-niche. The relaxation of rules concerning functional and structural separation for registries and registrars over the last decade has not yet materialised in a benefit for users and smaller players, and at the same time it cannot explain the market dominance of existing players which predates this reform. Additional pressure comes from increasing security and stability requirements where asset-light firms, as is the case of the smaller registries and registrars, increase the chances of vertical and horizontal integration. This trend will probably increase in the future as has been shown in other digital markets.

The redistribution of market power in the DNS registry sector is particularly crucial for two reasons. In the first place, because the governance of the DNS registry and registrar sector under the policy umbrella of ICANN is a historical site of contestation, as well as of representation of legitimacy of global multi-stakeholder governance (Bradshaw and DeNardis 2019; Jongen and Scholte 2021; Weinberg 2000).

In the second place, because geographic diversity, particularly among registry services, is still lagging and cannot uphold notions of representativeness in the DNS sector, which is still vastly controlled and represented by companies in the north. ‘Affirmative action’ strategies should be stimulated by ICANN as market rules and further consolidation among registries and registrars will not, on their own, promote a more equitable distribution between the head and tail of this market, nor have they, on their own, the capacity to increase the demand for domain names.

In terms of the limitations of this study, one of them is the access to data to study how bundled services and virtual, as well as horizontal and vertical, integration become part of the business and sustainability of registries and registrars. Another concerns the time of the study which has been too close to the moment of the consolidation cases described with GoDaddy and Identity Digital, and these effects could not be adequately assessed, though 2020 was a pivotal year that changed the trend of decreasing market share from the largest registries. Future studies could further elaborate on these issues. In addition, further work could be undertaken on domain name pricing effects on TLD growth, as well as the effects of abuse and security issues on TLD extensions.

The potential implications of increased consolidation trends in digital platforms – and through internet infrastructure as that seen in cloud provision and DNS resolvers and hosting – loom over the TLD registry and registrar field. Security concerns for gTLDs challenge their own existence and value, though robust policy design and enforcement may be a strong factor in deterring abusive behaviour that undermines trust in the DNS. The capacity to counteract some of these trends in the TLD sector needs more than incentivizing certain behaviour to shape market rules, and more explicit norms to structure these different sectors. Increased political commitments should be undertaken, at least to uphold geographic and IDN TLDs if the sector aims to be a global and inclusive, as well as an institutional, model for multi-stakeholder governance. To achieve this, interested parties should be participating in the formulation of these policies where possible and ICANN should be actively promoting engagement from existing and new players.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article was originally published with errors, which have now been corrected in the online version. Please see Correction (http://dx.doi.org/10.1080/23738871.2023.2316406)

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

This work was supported by DNS Research Federation.

Notes on contributors

Carolina Aguerre

Carolina Aguerre, Associate Professor at Universidad Catolica del Uruguay. Co-Director for the Centre for Technology and Society at Universidad de San Andrés (Argentina) and Senior Associate Fellow at the Centre for Global Cooperation Research (GCR21) at the University of Duisburg-Essen (Germany). I am a researcher working at the intersection of technology governance and global, regional, and national policies. My work focuses on global technology governance processes involving the Internet and AI looking at institutions, actors, and socio-technical arrangements, particularly in developing regions’ and from a power perspective. I have published in different academic publishing outlets including Routledge, Digital Policy, Regulation and Governance, Palabra Clave, Information and Culture and MIT Press. My most recent publication is an edited volume titled ‘Digital Data Governance. Polycentric Perspectives’ with Malcom Campbell Verduyn and Jan Aart Scholte (release in the fall 2023). I have received research grants from Tinker Foundation (United States) (2019), National Agency for Innovation and Research (Uruguay) (2020), National Council for Scientific Research (Argentina) (2010-2012) and the German Federal Ministry of Education (2020–21), among others. I have advised bodies such as UNESCO, ICANN, the Global Forum on Cyber Expertise (GFCE), the Digital Trade and Data Hub (George Washington University), the Global Partnership on AI (GPAI) and InTouch.eu from the European Commission. I have a PhD in Social Sciences from the University of Buenos Aires (Summa Cum Laude) (2015), an MA in Communication Culture and Society from Goldsmiths College U. of London (2002) and a BSc in Social Communication Studies, Universidad Catolica del Uruguay (2000).