|
||
|
||
In the tenth month of the revolutionary expansion of generic top-level Internet domains, global registrations in new gTLDs reached more than three million addresses, providing the clearest illustration yet of the strong international appetite for new, relevant addressing options. As we near the first full year of new gTLD availability, focus now shifts to another critical metric—renewals—which we expect to show similar strength based on history and data analysis.
If registrations demonstrate consumer appetite, renewals reflect consumer satisfaction. Both factors are critical to the long-term success of new gTLDs. While every hour brings new registration data making the case for a strong and growing global enthusiasm, we will have to wait four to five months before we have a full picture of the renewal marketplace.
If history is any guide, however, we have good reason to anticipate renewal figures that are impressively strong.
Several key factors drive renewal rates:
On each of those factors, new gTLDs in general (except, perhaps, any that are offering outright free names) and Donuts gTLDs specifically present a strong case for renewal.
Age and Size Are More than Just Numbers
The age and size of a TLD are important factors in renewal rates because the better names in a TLD are registered first. Put simply, TLDs with fewer names registered in them have a higher renewal rate because the 20,001st name registered (example: GreenBay.PIZZA) is of better semantic quality than the 100,000,001st (example: Samza-OrderIn.COM). Comparing the median Donuts new gTLD zone file size of 11,054 (for gTLDs launched prior to May 31 of this year) to .COM’s more than 100 million, it is reasonable to anticipate strong renewals on that factor alone.
A qualitative analysis of new gTLDs further supports this point. Looking at the zone file for .PHOTOGRAPHY, for instance, one sees a gTLD dominated by short, descriptive names that are directly related to photographic content or the photography business. More than half are keyword domain names, and the remaining typically are small business company names. An examination of the market dynamics of existing TLDs reveals that domains of this type are renewed at the strongest observed rate.
Country code TLDs (ccTLDs) that are operated as country codes, also typically smaller than legacy generic TLDs, are another indicator with higher renewal rates. Names in these TLDs renew at a blended rate of approximately 80 percent.
The recently re-launched .CO domain gives us a look at renewal rates for a larger TLD. The first 100,000 names registered in .CO posted a renewal rate of 92 percent, while the next 180,000 renewed at 72 percent, netting out to a blended rate of 78 percent for these first 280,000 addresses. The case for renewal in .CO becomes even stronger if we factor in the rate of names that were re-registered shortly after non-renewal. More than 80 percent of those first 100,000 .co names that did not renew on their first year anniversary were subsequently re-registered as new names within the first 60 days of non-renewal. More than a year after the first 100,000 .CO names were registered, 98,000 were still registered.
Related to the size of the TLDs is the breakdown of names registered by year. Names registered in 2014 in .COM are names registered in that TLD’s 29th year of existence. Names registered in 2014 in a new TLD are names registered in that TLD’s first year of existence. Names registered in the first year have a higher renewal rate than those first registered in the 29th year.
Meaning Matters
There is a tendency in analyzing TLD performance to focus on the distinction between “real” users who register domains for the purposes of immediately establishing businesses and identities online, and domain investors (or “domainers”), whose principal goal is asset monetization, either through resale or advertising.
While that distinction may matter in assessing the qualitative impact of a new TLD, it has less bearing on the quantitative reality, at least in the new gTLD marketplace. As it turns out, the renewal behaviors of both professional domainers and non-domainer registrants are driven by many of the same considerations. In both cases, semantic value—that is, what, if anything, an address means—is a critical factor in determining renewal rate.
In legacy domains like .COM, registrants renew at a rate lower than that of many other gTLDs. But because early registrations in any gTLD, especially now in new gTLDs, are based on the semantic value of the underlying terms, and not merely the traffic the name might generate, the purchase motivation for any buyer is different than those purchasing in .COM.
Accordingly, the first 50,000 or so names in each TLD will have an extremely high semantic value, and thus a high renewal rate, regardless of who the registrant is.
Things get even more interesting when we compare .PHOTOGRAPHY, for example, with the population of .COM names that were newly registered in 2013. For these .COM names, the percentage of active sites with unique content is just 23 percent, and a full 30 percent of them fail to resolve. The renewal rate on these .COM names is just 52 percent. In .PHOTOGRAPHY, already over 20 percent have active content (a nearly 14 percent increase since late June) and only eight percent don’t resolve. It will be fascinating to compare a full year of .PHOTOGRAPHY with the population of .COM names newly registered in that same year. Current data suggests the .PHOTOGRAPHY names will handily outperform the contemporary .COMs in renewals.
Domainers and new gTLDs
Domainers are an unusual market segment. As the “fire-starters” of any registration trend, they not only are hyperaware about TLDs but act as an unpaid sales force by selling domain names to other types of users at a margin above what they paid the registrar for the name. In the first year of any TLD the proportion of domainer registrations is therefore skewed to a higher percentage. But over time, and this is happening steadily already, domainers get good names into the hands of end users.
Why are these domainers registering names in .COM and in new TLDs? To satisfy one or both of two criteria: 1) semantic value, and 2) traffic value.
Domainers register cheap .COM names and park them, hoping to make more per year in PPC traffic than the few dollars they paid to register the name. Will domainers renew a .COM address at the registrar’s $8 renewal price for that second year? Not if the traffic value is less than $8. They won’t even renew for $1 if the traffic degrades below that value.
Now let’s look at why domainers register new names in new TLDs. It’s not because of the traffic, because there is almost none yet. The entire TLD is new, so no residual or legacy links to currently unregistered (but previously registered) names, or any other source, even exist. In fact, according to our own analysis of six of the earlier released Donuts gTLDs (.PHOTOGRAPHY, .EMAIL, .COMPANY, .TIPS, .TODAY, GURU), usage of registrations for PPC declined by an average of 9.36 percent over a four-month period ending November 1. So the reason for domainer interest is the other component of domain name value: the semantic content of the name itself. It’s meaning these domainers are buying, not traffic.
The domainers bought the names because they believe they’ll be able to re-sell the names for a higher value later (the names are only parked because there is nothing else to do with them while waiting for TLD awareness to increase and the names to be resold at a profit). They are speculating. And they are paying 10x (or more) for these new TLD names, compared to .COM names.
Traffic fluctuates over time, but meaning does not.
Unlike the quantity of traffic to .COM names, the meaning of words does not change very fast. “Pizza” means “pizza” all over the world—it meant the same thing last year and it will next year. The meanings of “Fast.PIZZA”, “Chicago.PIZZA”, “Order.PIZZA” do not change, so they hold their value. With increasing awareness of the .PIZZA TLD, or more Google attention for semantically valuable names, semantic value on the Internet increases. This is why domainer registrants will largely renew, because a) to not renew means losing their investment, and more importantly b) they believe the name’s future value (minus the future renewal fee) is greater than its present value.
The Price Equation: Easy Come, Easy Go
The price we pay for something is directly proportional to how much we value that thing. In the same way that you care more about your Mont Blanc pen than the cheap disposable you bought in a pack of 20, you are likely to place greater value on a domain for which you paid a premium than one you bought among dozens at a bulk introductory rate for pennies.
Names sold for higher prices have a higher renewal rate (when the second year price is the same as first year price) than do names sold for lower prices because names sold for lower prices are more often registered purely for traffic reasons while names sold at a higher price engender more attachment and have more meaning. A buyer paying $20 for a name has a sunk cost incentive to renew when it comes time for renewal. Registrants think twice about deleting names that cost them $20 and think three times if they cost $200, but don’t think twice about deleting a name they got for $1 or free. As an example of this principle, if .PHOTOGRAPHY names were $1000 per year each there would be fewer names in the zone, but renewals would be very high.
Diversity Creates Stability
The more diverse a gTLD’s registrant base, the less likely it is to be impacted by any individual’s renewal rate decision. Renewals generally go up for TLDs in which the ratio of addresses to owners is low (as opposed to domains that have higher numbers of portfolio registrants). Our ratio of average domains per registrant is low, at three names per registrant, and that ratio is getting smaller as our TLDs mature.
What that means from a practical standpoint is that Donuts domains have not been concentrated in a few hands and therefore are not likely to be cancelled or dropped en-masse.
Another Consideration: Deciphering “Blended” Renewals
Names that already have been renewed at least once have a higher renewal rate than will names that have yet to be renewed.
This is an important distinction to understand when analyzing renewal rates. Is the rate you see the blended renewal rate, or a specific cohort’s renewal rate? The blended renewal rate for .COM is about 72 percent, while the renewal rate of .COM’s first-year cohort is, now, close to 100 percent (names registered 29 years ago that have been renewed each year since then are very likely to renew next year), and the renewal rate for the .COM 2014 cohort will be approximately 50 percent. All those cohorts together make for a blended 72 percent rate for all names in the .COM TLD.
The same data is born out in the same way in the .CO.UK domain name, according to this graph:
Renewal rates correlate to the age of the domain – Domains between three to four years old follow the average renewals rates of around 70% and this increases up to over 90% for domains aged 10 years and over. (Source: Nominet)
What Should We Expect?
In March 2015 we’ll see the renewals start on hundreds of thousands of short, descriptive, meaningful domains that were purchased for real money by a wide variety of registrants, and were not purchased for traffic harvesting purposes. The semantic value of these names is not decreasing, and usage of these names and general awareness of the TLD increases steadily every month. Based on the factors discussed here, we conservatively predict an initial renewal rate of greater than 80% for Donuts gTLDs.
Sponsored byDNIB.com
Sponsored byWhoisXML API
Sponsored byIPv4.Global
Sponsored byRadix
Sponsored byCSC
Sponsored byVerisign
Sponsored byVerisign
A World-Renowned Source for Internet Developments. Serving Since 2002.
Paul, great analysis and great conclusion—especially around the importance of “meaning” with regard to new TLDs where existing traffic is less of a factor. I might add that domains with multiple words may seem to have more meaning, but in reality have power because their message is diluted. Domains using new TLDs are naturally shorter due to much higher name availability and furthermore they are shorter because the extension is not wasted. How much clearer is rays.pizza then famousrayspizzanyc.com?
Some great arguments, Paul.
However I do think that many of the initial registrations - especially for .guru - was due to curiousness with no initial plan to use the domain name. Also, out of the 3.2m domains registered, around 650.000 .xyz, 85.000 .realtor and 75.000 .berlin were given away for free. So the total renewal rate will be lower than a standard ccTLD rate, if payment is requested for renewals. Any thoughts on this?
The difference between the .xyz renewal rates and the .realtor renewal rates should be very interesting. Many of .xyz registrations ar robot registrations where the registrant neither asked for the registration or paid for the registration. The .realtor gTLD is really a "community" gTLD. Where registrants identify with "their" TLD, the renewal rates tend to be good. Thinking of TLD registrations purely in monetary terms misses the one thing that drives renewals and growth in ccTLDs. The free domains issue with .berlin gTLD is quite different from that with .xyz gTLD. The .berlin landgrab was attributed to two companies. Many of these domains are speculative and they may be renewed repeatedly unless they are sold. Thus as a group, many of the .berlin free domains may, for the first year at least, not follow the classic boom and bust cycle if .berlin continues to grow.
Thank you, Paul—I agree: the distinction you make between meaning and traffic is a very valid one (and my own strategy—which I often refer to as “The Wisdom of the Language”—is strongly focused on meaningful names).
There are, however, a couple details that merit further inspection.
The first of these concerns the meanings of the TLDs (rather than the meaning of the second level domains). This is no insignificant matter—and for most of the new, privately held top level domains it is by and large uncharted territory. All of a sudden, private corporations (some of which are relatively small companies) are now on par with the governments of entire countries. There is a risk of balkanization—into not only hundreds, but thousands upon thousands of tiny fiefdoms. Which laws will apply, and how such laws will be applied is very much open to question. For example, in one dispute I know of, the rationale used went along the lines of “Facebook has a right to the name face.book and MasterCard has a right to the name master.card”. I find this quite shocking, but nonetheless this reasoning was actually applied (and AFAIK it was even applied beyond the court’s own jurisdiction). I refrain from an exact citation, because I find the decision completely absurd and do not wish to give it any credence whatsoever.
Another example is the “rebranding” of many country-code domains. Prominent cases for this are TV (Tuvalu) being rebranded as “television”, or CO (Columbia) being rebranded as “company”. Although this is mainly a “marketing trick”, there are nonetheless many people who were thus convinced to register domain names in these registries on the basis of such pitches.
And there are other prominent cases in which a TLD changed its own meaning. For example: The ORG registry was once reserved for non-profit enterprises, but this limitation no longer exists.
In general, though, I would agree that the way the users of any language interpret strings tends to be the “meaning” of the string that actually matters in the marketplace of ideas. In this vein, I would say it may be even more important that the usages of the strings in question are consistent, reliable, etc. than that the usages are widely distributed. Of course, both together are a good basis for establishing a cohesive linguistic community—but if one had to choose one or the other, than I would lean towards “conforming to rules” rather than “used all over the world”. Such rule-bound conformity can be the basis of the reliability people seek in adhering to conventions. This is one reason, I guess, why the “hyperlocal” geographic names (such as BERLIN) are so successful—people feel that the involvement of the local government leads to a more reliable top level domain.
Now there are two more points I wish to also address.
First: Ray, “famousrayspizzanyc” is not defined in English—it is a meaningless string. “Rays” is defined, but I think it is ambiguous (for example: it might refer to sunlight or perhaps the rays from a microwave oven). Globally, English has a relatively “loose” grammar, and in any case languages are always changing. One thing that I find fascinating, however, is that individual words don’t usually have any meaning by themselves, but rather only when they are used in specific and/or particular contexts (indeed: this was one of Ludwig Wittgenstein’s most significant insights).
This brings me to my last point. When I first wrote about “The Wisdom of the Language” (I think that was in early 2007), I noted that Google’s algorithms are actually not very good at understanding meaning (and I also wrote that there may actually never be algorithms that are able to do that—in part simply because language is a “moving target”). While Google’s algorithms are indeed able to recognize that when people type in “amazon” that they probably want to visit amazon.com (versus, e.g., a wikipedia.org page about the Amazon region), it will often push content to the top of its results pages based on metrics other than meaning (and of course there is also a vast industry built around finding out what the “secret recipe” for “Google search results” is—as well as for other “secret recipes”, such as Coca-Cola).
But I myself do not care what the top results at google.com are. When I look for a hotel, then I will want to see the top results from a hotel domain. When I want a new bike, I may very well be more inclined to check out .bike domains. Or I may choose to search by local geographic area (if that is what I want at the moment). Google.com simply doesn’t matter very much to me.
And I also doubt that Google will change its focus on its own bottom line. As far as I know, Google is itself interested in acquiring control over many top level domains—and I presume they will be very interested in promoting these… much in the same way as they now promote the youtube.com video platform, their Chrome browser, mobile telephones using their Android software, etc. They have a responsibility to their shareholders, and they do not have such a responsibility to help me find the best bike.
Very positive post Paul.
Perhaps some industry adoption of some of the new gTLDs might help with your cause. It will take more than just Google pushing new gTLDs. At this moment the launch of the new gTLD Program is not where I truly expected it to be.
From a search perspective Google is all about trust, safety and semantic meaning especially if the TLD has high quality content that enhances the TLD’s semantic meaning (See ccTLDs counting in local searches—e.g. A search in Germany will return mostly .DE results).
For a new gTLD to be a success you would need thousands of users using the specific gTLD for its semantic purpose. Usage, high quality content and mass market use is key. Any industry adoption for any popular highly-searched gTLD would aid this process and also help promote all new gTLDs because it would provide the gTLD with great content with a high trust and safety factor. Legal and trusted content will be important too. For example, less takedowns requests will play a big factor in the future of copyright-focused strings (copyright enforcement will play a role in the future of search since Google will continue to improve its algorithm to demote illegal sites and gTLDs with a history of low quality content and illegal activity).
I do not see how the new gTLDs that have been launched thus far will accomplish this scale and trust factor since nearly all launches have been too similar and provided no differentiating factor beyond the novelty of the name to the right of the dot.
As one of the biggest proponents of the value of new gTLDs, I do hope your assessments are correct. Looking forward to seeing the renewal rates of new gTLDs. With limited demand, many substitutes (.legal, .law, .lawyer, .attorney, .lawyer, .legal, .esq), limited complementor products to differentiate beyond .COM, and high supply, it does seem the jury is still out.
The biggest threat to new gTLDs are apps and closed ecosystems. Mobile is the present and the future. On mobile phones, 86% of users’ time is spent in apps and only 14% is spent on the open web according to mobile-analytics firm Flurry. This divide will continue to increase. While apps may be complementors to new gTLDs the writing is on the wall about how the future will unfold. Apps are a huge win for users since they are fast and easy to use. Convenience and “getting the job done” is a huge incentive for users to adopt any product. Until we have real industry adoption from each string’s corresponding community then new gTLDs will just be niche success stories at best.
Thank you for your in-depth analysis.
Constantine Roussos
.MUSIC (DotMusic)
I hope your company employs a statistician to build renewal-prediction models with this post being just a marketing tool.
Christopher, thanks. We agree with your point on free giveaway names and pointed this out early in the article.
Constantine, correct that it’s not all about Google. Our point of view is that buy / renewal decisions are at the level of the registrant, and while Google love may play a role, semantic meaning is the predominant factor in those decisions as of today.
Paul/Mason,
You have touched upon several key factors that will drive renewal rates, such as the age and size of the TLD, its semantic value, price and registrant demographic/distribution of registrants but your article downplays the biggest key factors: gTLD usage, gTLD domain development and gTLD domain marketing.
If a registrant has invested in a domain the likelihood of its renewal sky rockets. I am not sure the statistic on .PHOTOGRAPHY—that 80% of .PHOTOGRAPHY domains do not have any active content—is a number that is positive.
Also another issue is one of domainer speculation since many domainers have registered a big portion of new gTLD domains. If these domainers are convinced new gTLDs will bring future ROI to them then they will hold on to these domains.
However, the more “synonym” TLDs that are launched (e.g. .PHOTOGRAPHY, .PHOTO, .PHOTOS, .PICS etc), the less each of these domains will be valued on average, given the difficulty of differentiation and issue of user confusion. This also holds true for singulars and plurals, which can add to user confusion. Let us say a registrant registers ShareMy.PHOTO and another registrant registers ShareMy.PHOTOS. The average user would have difficulties remembering whether the site is singular or plural. Over time registrants will have to take into consideration the cannibalization and revenue loss created by user confusion because of the synonyms and singulars/plurals issue with new gTLDs.
The most desired goal for increasing renewals rates is increasing gTLD industry adoption and optimizing usage rates to create a more trusted and prevalent gTLD brand. Using the traditional domain “factory” approach of commoditizating gTLDs that relies solely on the gTLD’s semantic meaning as a value proposition does not help drive renewal rates either.
Should be interesting to see how the renewal rates for new gTLDs will unfold. Pricing will play the biggest factor in my opinion. Going from a free domain to a paid domain (e.g. for .XYZ) will strongly affect renewal rates. Each gTLD is a case-by-case scenario though. We will see some winners and some losers that is for sure.
Constantine Roussos
.MUSIC (DotMusic)
Constantine, we think you’re right about pricing, as we discuss in the article. Free names, particularly those not even asked for, won’t see much of a strong renewal trend.
On your usage point, this is a transition period. If you look at .COM adds in the past 12 months, you see about a 40% usage rate, if you combine active content and redirects to active content. .CO, by the same measurement, has about 20% usage. Our example TLD of .PHOTOGRAPHY has about a third of all registrations in active usage. Impressive enough for a new TLD, particularly in less than a year after launch. We think that usage trend will continue its upward rise.
You are right. The usage rate will continue to rise because it is below the .COM average, but it will level off. It makes sense that renewal rates for new TLDs with a lower volume count will be higher because the best available names almost always get renewed. However, what might be concerning is whether new gTLD growth rates can offset their attrition rate as more new TLDs enter the marketplace. We have seen the negative trending effect of this attrition in .NET, .TEL, .MOBI and .ASIA. Without major differentiation, marketing push, lack of branding and an increasing supply of substitute alternatives, new TLDs domain counts will grow, likely reach a maximum level and then drop to a stable domain count range that will remain quite constant over time. The risk of new gTLD commoditization will likely leave domainers as the primary new gTLD registry customers for most new gTLDs. Since .COM domains are cheaper, have a significantly higher resell value and have a higher parking value than new gTLDs, domainers will just hold on to only the top new gTLD domains of high semantic value. We are all very curious to see how it all develops in 2015 as more new gTLDs get launched and the first new gTLD renewal rates are seen. Constantine Roussos .MUSIC (DotMusic)