A number of domain-name appraisal sites use
the length of a domain name in their appraisal model as one of the
predictors of domain market value. Our empirical analysis indicates
that length is insignificant in predicting the market value of a
Using a sample of 95 dot-com domain-name transaction
prices obtained from eBay, AfterNIC, and proprietary data from our
escrow service, we performed a number of statistical tests. The
domain names in our sample ranged from 2-19 alphanumeric characters.
Below are some of the findings.
- As expected, we found
a negative relationship between the number of characters in a
domain name and prices. However, a linear regression with length
as the predictor is able to explain about 9% (R-squared of 0.09)
of the variations in prices, while a log transformation of the
predictor values yields a meager 11% R-squared.
- A number of appraisers
use domain-name groups as a predictor
of Value. To test the explanatory power of such groupings, we
divided names into groups of five, six, and seven characters.
Thus, a five-character group would have domain names with 2-6
characters as group 1, 7-11 characters as group 2, etc. For example,
“bundled.com” would fall under group 2 within 5-character grouping.
As expected, we found a negative relationship
between the groups and prices. However, the grouping variable
was able to explain only about 6% of the variation in prices
of domain names in our sample.
- Adding length to the set
of predictors that we use, we found that a linear regression of
the combined predictors does not add explanatory power. A linear
version of our model explains approximately 81% of the variations.
- Using a non-linear, tree-structure,
discriminant-analysis regression technique (the primary modeling
technique used for our appraisal services), we found a very minor
influence of the length predictor — i.e., lower predicted prices
for some domain names with greater than 9 characters.
Topic tags: appraisal/valuation
Connect & Share