Dan Warner, who many in the domain industry know is a numbers/stats guy, has some interesting thoughts on winners and losers of the new gTLD’s in the “sports” category which he recently published.
Mr. Warner has done guest posts for TheDomains.com and you can read more about his background and domain experience here.
Its an interesting read but you would have to agree with Mr. Warner’s basic theory which is “The very best indicator on market size for new gTLDs is the number of contextually related .Com domains in existence. ” to buy into the winners and losers list
Dan writes:
Market size is a critical part of any registry valuation, and if a valuation cannot be clear and concise with references to the actual data used to form an opinion – it isnt worth much.
“The data is there if you know what you are looking for and have the experience to understand it. ”
What you need to know is what revenue the business will generate.
It doesn’t make sense buying a business that will always be a loser but you won the auction.
“The first thing you look for in doing valuations are the benchmarks.
Demographics, market shares, contextual contribution to total market size, search volumes, previous sales, existing customers, and more. ”
The very best indicator on market size for new gTLDs is the number of contextually related .Com domains in existence. ”
If you are applying for the .Golf registry then knowing the total number of golf related domains already in existence is fairly important.”
Com is the king of registries.”
The most accurate source of information for data mining and comparisons come from the .Com root zone file (all .com domains owned). ”
“This has led to approximately 105 million .com domains being owned. ” (actually the number is currently over 109 million-MHB)
The domains in the .com zone file are a clear indication of the domains that are likely to be purchased in new registries.”
The market size can be approximated by measuring the number of domains which have already been registered as a .com and then estimating the number of domains that will comparably be registered for domains with similar character strings. PaintingArt.com = Painting.Art.
.Rugby is a good example.
Rugby is a term that can’t be realistically shortened, it is rarely used as a plural, and without the inclusion of its 5 characters makes little sense. It is used as a category qualifier.
If an existing domain includes rugby it is almost certainly is going to be about rugby.
So the next thing to find out is how many .Com rugby related domains are there?
The answer is only 13,311 domains are in existence.
These domains include a massive amount of diversity in domains.
Included are denverbarbariansrugbyschedule.com, driving-lessons-rugby.com, fishbirdrugby.com, e2v24rugbyfeminin.com, and newyorkrugbysevens.com. These are from a fairly random sample but they do represent the depth and diversity of the domains which have been registered.
Considering .com is the overwhelming market leader, has been around since the 1990s, is analyzed by the worlds greatest statistics junkies, and everyone who every felt they really needed a domain has probably already bought a .com – what percentage of the existing .com market should a new gTLD registry expect to get?
The million dollar question is by comparison to .com how many domains will be registered?
Do you believe that as a newcomer registry that your brand is big enough to topple .com?
Are people going to throw out all their business cards, forms, and stationary to change their brand for an unproven new gTLD?
Are new buyers that haven’t had an interest in the last 20 years all of a sudden going to decide they have to buy a new domain name?
The answer of course is that each new gTLD will capture a share of the existing market.
That many existing domain owners will also want the new domain brand – but not all of them.
Change comes difficult to many and it is still the global financial crisis so let us be factual and realistic.
You can’t create markets that don’t exist – so understand the market that does exist.
If rugby currently has 13,311 .com domains and they capture 20% of the .com market that equals 2,600 domains. Not nearly enough to justify a registry or its costs.
In contrast another example Sport (and Sports) have just over 700,000 .com domains which are registered with the “sport” character string in the .com domain name. A great generic brand with strong contextual relevance to the category.
If this new gTLD captures 20% of the market over 5 years that equals 140,000 domains.
That would be a conservative and outstanding achievement.
That registry after five years (not at auction) might be worth $15 million then if it meets those targets, but right now it would using a discount rate of 35% would be more like $2.5 million.
The chart above is a segment of the full chart which can be downloaded here.
The full chart PDF contains 156 different variations of sport related character strings and their .com domain counts.
You may be astounded by the number of domains each concept represents.
There are also a lot of false positives.
“Fit” as example would include Fitzsimmon, Fitness, and Confit.
It would have an extreme amount of false positives that need to be vetted out before meaningful conclusions could be drawn from it.
However, on the other end of the spectrum if a character string like “cricket” which only has 12,349 domains registered as .com’s and your registry expects it will be selling 500,000 domains in the first 5 years – we are afraid you are headed for a train wreck.
So take the high positive counts at the top of the list with a handful of salt and take the low counts at the bottom to mean that is probably the whole market.
Dan Warner’s favorite for the sports category are .Sport (and .Sports), .Golf, .Fitness, .Dance, .Bike, .Yoga, but you can see the entire chart for all the winners and losers.
Dan did not include .Football, .Baseball, or .Hockey on his list saying “Consider that for these sports they rarely are played as adults, don’t require frequent tuition (services), and have very little equipment (products) to purchase. Not that many domains are purchased when you can only sell a ball, glove, or have a market which is too small. Understanding the market economics are the key to valuing where your registry will be at in 5 years.”
Not sure I agree with that assessment but here is Dan’s chart, to see a large picture of the list click here
And the winner is…………….
Dan,
Very informative article. You really have put a lot of thought into the analysis.
Tan
seems a bit more useful and realistic way of appraising the value of new gtlds instead of assigning some random dollar amount to keywords.. which is only done for novelty sake.
Using the same metrics/formula and Analytics:
I wonder where .tv .info and .biz will fit/rank in that puzzle?
say they didn’t already exist… and were part of that great and upcoming “next big thing” called New gTLDs ?
Hi All,
This study is less predictive than it is showing a proxy for a Maximum Realistic Market Size for a domain class.
In one valuation I did for a domain extension in the top 10 most contended applications the total .com domains were over 700,000 as the extension was three letter extension and fairly generic. Once I vetted out all the false positives there were about 200,000 left (29%)
Another extension valuation I did also had three common characters with over 2.5 million .com domains returned and the final number of eligible category domains was 250,000 (10%). Whereas if you were looking at the “cricket” character string it would include almost exclusively the total number of domains returned (near 100%).
So for extensions that include common strings like “fan” the false positives include fantastic, fans (air fans), fanta, fantasy, etc. The number of total initial .com domains returned doesn’t mean much until you process it which takes days. However, when the term is very descriptive and longer it is much more accurate out of the gate. More importantly though is when the total possible market size is too low to begin with to be viable – even before you remove the false positives.
Valuation and creating extensively research predictive domain lists, premium domains, and auction reserves is a science and an art. It isn’t nearly as strait forward as this simple view of the market. This is about market size not market value. The premium domain sale values are not part of this calculation.
Generic extensions like .info or .biz are much harder to value and considering the hyper competition in the new market I expect would face a much different competitive landscape compared with when the registrars were starving for stock. This is becoming more the age of the specific rather than the “me too” generic extensions.
Maximum Market Size isn’t enough to make a valuation. This study is a window of information to show accurate valuations can be realistically done and a warning. If a registry based on this data can’t hope to sell 10,000 domains in the first year or so they will probably go broke. Even efficient Mega Registries likes Donuts with economy of scale for costs and management, and specialist knowledge of the market, are unlikely to make them work. What chance does a stand alone registry have?
[The black line in the graph is for 40,000 domains so if you consider a 25% market penetration of the maximum market size that equals 10,000 registrations – a realistic point of basic viability]
These numbers are for the MAXIMUM market before false positive discounting – Looking at the numbers which bet would you make?
Definitely some interesting ideas there. However there are some modifying factors when it comes to sports and other niche concepts. The most important is the geographic factor. Apart from national teams and venues, there is a lot of localisation in sports so it would be important to check whether the .com registration is a brand protection registration for a ccTLD. If this is the case, then a niche TLD brand registration might not be used or even registered. Another factor would be the longevity of the current .com registration for the term. The hidden aspect of that would be the drop history of that term. The most important modifying factor would be the usage of the .com domain. If it is PPC parked or just a holding page, then the niche TLD version might not be attractive especially if it is a one hit wonder that only exists in .com TLD and no other TLD.
I still think that a lot of new gTLD registries will be hoping for a major brand protection registrations boost. This data may actually help quantify the scale of that boost. However there is a danger in relying purely on the .com domains in that there is already a very high level of speculation and non-development of domains in that set. I ran a full com/net/org/biz/info/mobi/asia/others website mapping survey to determine the locations of websites and holding page and PPC parking websites were quite apparent. Unless there is an emphasis on development in these new gTLDs and especially in the niche new gTLDs, then much of the registrtion activity will be just speculative and brand protection registrations. And that’s not good for the long term survival of a TLD. That said, these new gTLD registries and their backers need all the data that they can get.
Thank you John for such a well thought out response. It is nice when we can get peer reviewed in a forum such as this. It helps us all shape our opinions and make better use of the available data.
This single point unvetted reference of a .com zone query is a strong reference only when considering the limitations and scope of a market rather than an indication of its total market value. It provides a great proxy for the upper limits of the market rather than its true value which could be and is most likely to be less (although it is indicative of its wide potential). So the data is best used for long tail registries. What are the limits of a market rather than its potential.
I was heavily involved in the .EU landrush and sunrise. EU has its own problems including national pride and company identity. However, it wasn’t helped by professional registrants owning too much of the core domains. My own company had a lot of them.
I fully believe in professionally owned generic domains as a form of arbitrage. It serves a purpose in some cases, but the reality is that professional owners wouldn’t exist if the value differentiation between the price a domain is sold for as a common registration and its inherent value didn’t have a value that could be arbitraged. Including the traffic of typed-in domains.
EU was a market that professional ownership took too much of the pie. If domain names with the .EU extension are not being used enough on billboards, taxi cabs, television ads, and print – then the market can’t be developed easily. When those companies are spending money on advertising they are also building the brand of the domain extension they are using. In EU the best domains are being held by professionals instead of end users, but the market to capitalize on that ownership doesn’t really exist yet because they aren’t being used or advertised.
Do I believe .Fit is going to be second to .Sport (.Sports) absolutely not! That’s why we need to understand the context of the data. A character string is a list of digits, hyphens, or characters in a row – That’s it! It means that the character string exists so many times. To make that data make sense it requires significant labor to purify the data, but the maximum potential remains unchanged.
The idea of how a registry would overcome the competition from .com and erode its market share is an interesting point of debate. After 20 years of dominance by .com and then facing the switching costs of ownership for the same market there is a threshold that must be overcome. I think there is great opportunity in the new gTLD market for a reasoned approach to migration.
Regardless, if the market hasn’t existed for .com – why will it exist for a new entity in a hyper competitive marketplace in the middle of a recession? If a market hasn’t emerged in 20 years for these niche markets of the smallest new gTLDs why will it exist now?
I see opportunity for as many as half of these registries – if they are run efficiently and they hire the specialist labor that is required. For the other half that isn’t backed by data I expect roulette in Vegas has better odds and more importantly a better return on investment.
Great report Dan, but I am not sure about this “Dan did not include .Football, .Baseball, or .Hockey on his list saying “Consider that for these sports they rarely are played as adults, don’t require frequent tuition (services), and have very little equipment (products) to purchase. ”
In the US many adults play baseball, football and hockey. There is a ton of equipment to purchase in hockey. So I don’t understand the analysis here one bit, maybe its that these sports are not as popular in Australia.
I have friends in their 50’s that play hockey every week, and after paying for ice time, skates, pads, helmets etc… they spend quite a bit playing hockey.
Hi Raymond,
I agree with you on those points, but you find that compared to the size of the sports the number of domains they purchase are significantly less. Although there are significant dollars spent on .Tickets and .Shoes the number of domains purchased surrounding football, baseball, and hockey are not very high. I first ran into this when I was monetizing traffic for domain parking companies and when we designed sport related landing pages it was difficult to find enough traffic to fulfill utilization below the parent “sports” category. Really only golf, fitness, weight loss, fitness equipment, tickets, and fishing had enough domains to make building the templates worthwhile.
It might be a brand related issue too. Football and baseball teams are not usually referred to in the generic. It’s the “Seahawks” not football team. Yes, hockey has a lot of equipment but the sport isn’t very large in the number of domains. Domains for .com that have the “hockey” character string it them number 28,115 after 20 years of registrations. If you think of that as a benchmark how many domains do you think will be registered in .Hockey?
Remember it is estimated that they will need to reach almost half that number (12,000) to break even as an independent registry.
The .EU was a bit of a disaster for all those concerned, Dan,
The EU Commission seemed to rely on people who didn’t understand the domain industry and chose what was effectively the registry of a small ccTLD to run what really was, in reality, a gTLD. It was massively overspeculated to such an extent that it killed development. Approximately 78% of businesses and companies in Ireland and the UK did not get their business name despite having gone through the poorly thought out Sunrise validation process. That effectively killed the ccTLD in Ireland and the UK, the main English speaking areas. The problem for a lot of the professional registrants was that they were primarily targeting English language keywords in a market with about 27 different languages. The registry, EUrid, even tried to lie its way out of the situation but were called out by Bob Parsons and others. I though that it was going to take between five and ten years to recover but what is happening at the moment in that ccTLD (I actually track trends in the gTLDs and a few ccTLDs including .EU) is unsettling. The powerhouse of .EU is Germany. The German registration rates stalled last year and they have not recovered. France and Holland and a few other countries have maintained growth but in real terms, the .EU is not doing well. The best that can be said about .EU is that it developed into a gateway TLD for websites and businesses in the EU. That’s a very different kind of TLD from a ccTLD and is one where there is minimal natural development.
What happened in .EU highlights a very important fact for the domain industry – speculation is part of a healthy ecology rather than just the ecology. Too much of it and it will effectively kill a TLD. The much of the rapid growth in real European Union ccTLDs from 2006 to 2010 be attributed the failure of .EU (and naturally to Domain Tasting in the gTLDs). Too many domains were registered by professional registrants (who primarily targeted generics), cyberwarehousers (who targeted everything) and cybersquatters. That destroyed consumer confidence in the ccTLD. There was no reason for small businesses, the Mom and Pop operations, to develop websites. Without these websites, the ccTLD turned into something close to a Dead Zone (where there is minimal development, high rates of abandonment and overwhelming levels of holding page and PPC parked domains). EUrid’s own usage “survey” was just a marketing response to the large scale (2 million or more) TLD web usage surveys that I run periodically. EUrid’s surveys are based on a 5000 website sample and while appearing statistically valid have a very unreliable methodology and classification. I run a survey of Irish websites each month and almost twice the .EU sample size is included in that survey (typically >350K websites per survey). The usage patterns of .EU most closely resemble those of .BIZ gTLD. The core TLDs (the ccTLD/.COM axis) generally accounts for over 80% of registrations and the bulk of any country’s developed webscape.
Just on the rugby issue, the assumption that the term “rugby” would be part of a domain’s string is not accurate (I realise that it is just being used as an example). Rugby clubs tend to use the abbreviation RFC (Rugby Football Club) at the end of the club name. Context, in this case, is more likely to be found on the website rather than the domain name. Of 231 Irish websites in the latest (July 2013) survey with the term “rugby” in their web content, 155 do not have the term “rugby” as part of the domain name. But Rugby is also a town in England and a famous English public school. That leads to another contextual element (something that is all too apparent to those who work in search engine development) that a purely statistical approach to zonefiles overlooks – the geographical connection.
Rugby, like most team sports, tends to be highly localised when it moves beyond the larger country level terms. Thus there are to be high levels of ccTLD domain names in domains with websites related to rugby. People identify with their ccTLD (generally) in a way that they don’t identify with .COM. They don’t have to think about the extension so it is part of their mental geography and they can probably remember a website in their ccTLD easier than they can remember a .COM website. (Ironically this effect is something that a niche TLD like .rugby would have in its favour.) The mistake is to take the .COM zone as being a single market instead of a generics market (which may partially overlap the country markets set) and a set of country level markets. The geographical connection is the reason that why .COM is being overtaken by some ccTLDs in the primary markets of those ccTLDs. Limiting the analysis to just the .COM zone could produce some misleading results and indicate a lower maximum potential marketsize than may exist. The early web was highly generic (in geographic terms) with businesses trying to sell to the largest possible market. Now that market has become more complex in that a website owner could be selling to their neighbourhood or to their town or city. Their market could have a clear and precise geographic coverage and that website may not use the geographic area’s name or even the product’s name in the domain name.
The key element that a basic .COM zonefile statistical approach misses is the level of development in the set of domains that could indicate a potential market. If development in that set of domains is below the TLD profile or below the industry/niche profile, then the potential market might be smaller than expected. Most registries now realise that while registrations drive growth, development is the fuel. Without it a new gTLD is more likely to be an also-ran or a complete failure. Development and intended market characteristics, (combined with the basic .COM zonefile statistical approach), might be a more accurate method of estimating market potential for a new gTLD.
Hi John,
Thank you for such a well thought out response. I wholeheartedly agree with your analysis. Rugby is a great example of where ccTLD data might be more accurate and the dangers of taking only a single point of reference on .com zone file data can be dangerous.
In order to be more accurate with the zone file data and to eliminate so many false positives in the data it requires some manual review of random samples for each query result and then applying that false positive reduction to the overall count. Using “fan” as an example the total number of domains that include the character string fan is grossly distorted by the false positives (fantasy, fantastic, air fans) whereas the number of false positives for “construction” are extremely limited to almost a pure sample. In order to do a statistical sample of 1300 alpha numeric extensions alone would require manually reviewing around 50 to 100 per query – 65,000 to 130,000 domains. You can manually review around 600 domains per hour (mind numbing work) which might be 108 to 216 hours of review for the current applications. Then you would need to recompile that data, chart it, and write commentary. To extend this data to all the available ccTLD and gTLD domains in the world, and a sub analysis of only newer existing gTLDs is mind staggering in scale but very possible. As this is unpaid work done primarily for the benefit of stimulating the community to test their assumptions, to show that with exhaustive analysis of a much wider set of metrics, a fairly accurate view of the market can be assessed (valued).
[This work would be well suited to a PHD candidate whose time might not be at such a premium as ours]
The reason I illustrate this is to show how imperfect the data is. It is a single factor lens to view one market size indicator. As in most data without the analytical mind and experience to temper and test that data (like you and I) the data can be misleading or inaccurate. It takes me about one full week of analysis to write a brief on a single extension. In that assessment it tests assumptions, looks at price points, analyses ccTLD variances, finds associated phrases, etc.
Here are most of the things required in a standard extension analysis:
Premium Domain List
Domain Auction Valuation
NPV Calculator
Realistic Market Size
Retail Price Elasticity
Variable and Fixed Costs
Break Even Analysis
Extensive Market Data
Most Likely Customers
Critical Mass Sensitivity
5 year Financial Projections
Five Market Forces
Value Chain Segmentation
Buyer Segmentation and Motivation
Risk analysis and Discount Rates
Best Practice Rules of Engagement
Most Likely Domain Registrations
Associated Phrases
Geographic Domains
Human Name Spinning
Trademark Awareness
Search Data and Bid Prices
Phrase Density Analysis
So to be frank looking at a single view of the .com Zone data is a bit of fluff comparatively. It doesn’t pass some basic hypothesis testing standards or deep peer review.
In my view the only thing it really should do is provide a reality check. For those applicants that believe they will achieve 100s of thousands of registrations for a new gTLD because they wrote it on the back of an envelope and now believe their own assumptions must be true – wake up!
Get the facts from experts who understand the full nature of the market and can give you unvarnished, data backed, and unbiased (less biased) opinions that don’t serve their own interests (as many technology providers may be encouraged to do).
What clients should want (and rarely do) is consulting that tells them the best and least biased opinion they can. Consultants are by their nature commercially encouraged to tell clients what they want to hear and already believe is true. [Find out what the client wants to hear, go find data to prove their opinion, ignore conflicting data if you can; parrot what the client wants to hear back to them – get paid].
Summary:
Get professional external advice and appreciate specialist knowledge is required
Ensure the assumptions are based on actual data and is provided with the review
Don’t lie to yourself and then believe your own lies
Realize most current applicants have a grossly distorted view of the market
Net Present Valuations of Registries do not measure what will be paid at auction on a given day. They give you a data backed framework to understand the realistic returns of a registry; what you can afford to pay at auction and still make the profit you require; and a strategic framework to achieve realistic results.