Relative Value of a
June 1, 2011
The relative value of
a top-level domain (TLD) should be measured by its price
premium over an identical domain name with a different
TLD. The measure can to used to give new
generic TLD (gTLD) applicants a better understanding
of the drivers of past performances and can be used by
investors to register and buy domain names.
A TLD’s relative value premium
does not necessarily translate into profits for the applicant.
Registrars and registries can be profitable by selling
junk domain names and names that infringe trademarks.
Conversely, high market prices do not necessarily imply
profitability. Since domain name registrations have a
fixed cost, market prices and revenue from registrations
are not related.
To determine whether Domain.org
is more valuable than Domain.info, for example, you just
need to compare their market values. But how do you determine
that one TLD is more valuable than another? Using low-high
ranges, medians, and averages is like comparing apples,
oranges, and bananas. Sheer number of registrations won’t
help either, given the already mentioned problem of junk
names and brand-infringing names. Furthermore, the number
of initial registration can be misleading.
To determine the relative
value of TLDs, you need to estimate their relative price
premiums. All else being equal, a higher premium means
a more valuable TLD. Statistical
regression trees can be used to measure the premiums.
They are robust to outliers and take into account nonlinearities
in the data, but before using them on a new gTLD wait
until market prices have filtered out the irrational exuberance
that can greet a .mobi and the like.