Analysis Of Domains Used By Top Websites

Analysis of short domain names and popularity Short dot com domain names command a high premium in the resale market. Even seemingly meaningless 3-letter combinations sell for many thousands of dollars at domain trading forums. Such high prices are often justified based on their relative scarcity compared with longer domain names. But are these very short domains really more likely to be popular? And, if so, do some categories of short domains do better than others?

In order to calculate the effect of a domain’s length on popularity we examined the most popular websites in the world ranked by traffic. We took our list domains from Alexa, which publishes a free list of the Top 500 global domains.1 From this list we examined all dot com domains with 3 or fewer characters. We further divided these names into 5 categories: Category

Members

Description

NN.com

100

2-digit domains between 00.com and 99.com

NNN.com

1000

3-digit domains between 000.com and 999.com

LL.com

676

2-letter domains (e.g. qq.com and hp.com)

LLL.com

17,576

3-letter domains (e.g. msn.com and udn.com)

3-char.com

28,080

3-character domains containing at least one letter and one number (e.g. fc2.com and 6rb.com) 2

Next we checked how many names from each category were to be found among the Top 500. Here’s what we found: Category

In Top 500

NN.com

3

NNN.com

4

LL.com

5

LLL.com

22

3-char.com

3

1

Our list of the top 500 domains was obtained on November 9, 2006. Alexa calculates their rankings based on tracking data from millions of users with Alexa’s browser toolbar installed. Although Alexa’s rankings are by no means a perfect representation of the true relative traffic a website receives, we believe them to be highly accurate for the web’s most popular websites. These top domains generate a volume of traffic which would be difficult to hide from Alexa’s albeit non-random data source, and would also be difficult to fake by artificial means. 2

Although the number of 3-character domains is sometimes said to be 46,656, that number (36 to the 3rd power) includes all of the LLL.com and NNN.com names as well. To get the number of “true” 3-character domains you have to subtract out these other names (46,656 - 17,576 - 1000 = 28,080).

These numbers might be interesting3, but by themselves they are not all that meaningful. To find out what they mean we need to know whether these quantities are significantly different from what might have randomly occurred, and how the quantities compare to the relative scarcities of each category of name.

The global domain market There are currently about 112 million domain names registered in all of the Top Level Domains (TLD’s), including both “generic” TLD’s (e.g. com and .net) and country code specific TLD’s (e.g. .de for Germany).4 Based on this estimate of total worldwide domain names, and assuming that each of these individual domain names had an equal probability of being a high-traffic name, how many from each category might we expect to find in the Top 500 list? Category

Expected number in Top 500

NN.com

0.00045

NNN.com

0.0045

LL.com

0.0030

LLL.com

0.078

3-char.com

0.13

As the table above shows, if the distribution of high-traffic domains was completely independent of length and category, we would most likely not find a single domain from any of our categories in the Top 500. Clearly, setting aside for now the “why”, there is an extremely strong correlation between short domains and popular websites.

Scarcity versus Popularity The chart below shows the ratio between the actual numbers of domains in the Top 500 relative to the expected amount in each category. This gives us a measure of the relative strength of each category in terms of producing high-traffic websites.

3 4

For the list of all individual domain names in each of these categories, see Appendix A at the end of this document. This information comes from a November, 2006 report by Verisign. See http://www.verisign.com/static/040029.pdf

Scarcity/Popularity ratio

8,000

6,720

7,000 6,000

Multiplier

5,000 4,000 3,000 2,000

1,657 896

1,000 1.23

23.9

All .com

3-char.com

280

0 LLL.com

NNN.com

LL.com

NN.com

We have added in all .com names as an additional category. As you can see, .com names in general have just a 23% better chance of making the Top 500 than we would be expected by chance alone.5 3-character .com names “beat the odds” by a multiple of 23.9, LLL.com by a multiple of 280, and so on until we get to NN.com names, which have an astounding 6,720 times more members in the Top 500 than would be expected by random chance alone. We can see from this chart that not only is it the case that short names in general are more likely to be in the Top 500, but this relationship between scarcity and relative likelihood of popularity holds true for every sub-category of short name as well. The scarcest category (NN.com with 100 names) is, on a name-by-name basis, the most effective at generating Top 500 domains. This is followed by LL.com (676 members), then NNN.com (1000 members), LLL.com (17,576 members) and finally 3-char.com (28,080 members). Lest it be argued that scarcity is, by itself, enough to guarantee a high proportion of popular domains, it should be noted that Top 500 lists only 2 LLL.net domains (a category just as scarce as LLL.com) and not a single short domain with a hyphen, even though L-L.com and N-N.com are just as hard to find as their 2 character, unhyphenated counterparts. Clearly some scarcely populated categories are much more effective than others.

5

We are estimating a total of 58 million .com domains currently registered. 292 of those made the Alexa Top 500 list we examined.

Analysis of causes and pricing effects The correlation is irrefutable, but why did it happen, and what does it mean? More specifically, could you make a case that this correlation, however strong, arose from factors which have nothing to do with theses domains inherent worth, and add nothing to their market value? One could argue that short domains have become popular because they were the first to be registered, and therefore websites on these domains had clear “first-mover” advantages in becoming popular. But this begs the question, why did so many of the early domain name registrants pick short domains? Clearly, the owners of early websites which went to become very popular could have picked longer domains which were more descriptive of their business, or more closely matched their business name, but they instead chose short names. For example, the world’s most popular exchange rate calculator chose XE.com (Alexa rank 683) instead of CurrencyCalculator.com or ExchangeRates.com, both of which were presumably available in 1994. Of course, not all popular short domains relied on this first mover advantage. Many other companies with popular websites registered or purchased shorter versions of their company name to redirect to their own site, or to serve as their primary web address. Examples of this include Fifth Third Bank (53.com), Barns & Noble (bn.com is a redirect), and MSN.com (Microsoft Network). So does this mean that the exaggerated presence of short domains among the Top 500 is simply a matter of these names being owned by large companies, and that the domain name itself contributed nothing to the growth and popularity of the company’s website? Even if you could argue that Microsoft’s main portal page would have been just as popular at any domain, you still have to explain why they chose a short domain like MSN.com. And why did so many other companies choose to invest so much money buying and promoting short domains instead of better matching, longer domains? This seems to be a very clear indication of the value put on having a short domain as a way to access one’s websites.

Implications for domain name pricing When it comes to evaluating the natural price-premium (or inherent market value) of short domains, does it even matter why a category of domain names has such a high relative percentage of Top 500 names? Perhaps not. Whether these domains became popular as a result of being short, or whether popular sites and companies could afford to purchase shorter domains, the end result is the same: a clear indication of disproportionate value being placed on these categories of domains. While certainly not the only factor in evaluating a domain’s value, clearly the scarcity/popularity ratio should be carefully considered. Very few domain resale markets can be considered liquid, but over the past year some fairly well defined reseller prices have been established for domains in each of the categories we examined. Do these reseller prices correlate with the categories effectiveness in producing Top 500 names? Just as importantly, have the relative prices been getting closer, or farther away, from their relative effectiveness? Is there any trend? Answering these questions would require a full separate analysis, and is outside the scope of this current paper. However, it should be noted that the answers to these questions, combined with knowledge of current reseller prices, could give very strong indications as to future values of short domain names in general, as well as which of these categories is relatively under or over-valued.

Appendix A: Alexa data for examined categories On November 9, 2006, when our analysis was run, the following domain names were found in the Alexa Top 500 list in each of the following categories: NN.com

LLL.com

Domain

Alexa Rank

Domain

51.com

118

msn.com

2

89.com

289

cnn.com

34

56.com

323

aol.com

38

tom.com

51

mop.com

111

NNN.com

Alexa Rank

Domain

Alexa Rank

yam.com

159

163.com

13

ign.com

195

126.com

80

ask.com

197

265.com

151

mlb.com

227

139.com

498

ups.com

262

wwe.com

267

ibm.com

275

icq.com

285

LL.com Domain

Alexa Rank

qq.com

5

udn.com

333

go.com

29

aim.com

345

hp.com

153

mac.com

364

cj.com

456

nfl.com

375

tv.com

467

nba.com

390

yok.com

398

tvb.com

449

sun.com

483

sdo.com

491

3-char.com Domain

Alexa Rank

hi5.com

49

fc2.com

94

6rb.com

324

2ch.net

226