Finding Comparable Domain Names
Alex Tajirian
August 12, 2011
Determining which domain names are comparables is an
imperative for the market approach to valuation. The larger
the number of comparables, the more accurate the value
estimate. The problem is that identifying such domain names
is far more complicated than most appraisers— professionals
or domainers—realize.
A recent paper by professors Gilbert and Hoegl titled “In
Praise of Dissimilarity” and published by the Sloan
Management Review generalizes the concept of similar
products and markets. The article notes that Intel and
McAfee share the theme of security even though, under the
traditional taxonomy, the two companies are very different.
The shared theme, the professors say, justifies Intel’s $7.7
billion acquisition of McAfee. They also cite the way that
Facebook, dissimilar to Google in many ways, still manages
to share Google’s search engine role.
When valuing a domain name, appraisers have traditionally
relied on the prices of sold domain names whose key words
are similar to those of the name being valued. But the
number of such sales is often so limited that the analyses
don’t have adequate data.
A real estate appraiser can compare a house with recently
sold properties that have the same number of rooms, the same
square footage, and the same neighborhood. The appraised
price of a new house on the market would be the average of
the comparable houses.
Unlike real estate, however, domain name comparables are not
easily identified because there are no obvious yardsticks
for measuring how much the different factors contribute to
value. Even names that contain the same sort of words may
not be comparable. Consider the differences that can crop up
with Indian tribes. One tribe may be on the federal list of
recognized Indian tribes while the other isn’t. One may run
a casino and enjoy the profits of a casino-related Web site
while the other doesn’t. Domain names incorporating the
names of these tribes could be very far from comparable.
Statistical methods are needed to weigh a name’s different
factors and what they contribute to its value. After that
the appraiser can locate comparable names and take the
average of their recent sales prices.
Under statistical methods, domain names are divided into
buckets according to characteristics such as the number of
Google key word search results, the key word’s pay-per-click
(PPC) revenue, and the number of trademarks. (Technically
the method is called nonlinear discriminant analysis.
One such method is
tree regressions,
which I have been using since 2004.)
Comparables are more than what meets the eye. Statistical
models improve the definition of comparable domain names and
thereby turn up a larger number of similar sales. That
allows for solid price estimates instead of guesswork.