Median price drop not as severe as average.
Last week I wrote about Sedo’s 2008 Market Study, which showed a 21% drop in the average sales price of domain names to $2,098. I explained that the “average” number doesn’t matter, and that we should really pay attention to the median. The median is the midpoint. If you ordered all 36,884 domain names from lowest price to highest, the median would be right in the middle (technically, the average of the 18,442th lowest domain and 18,443 lowest domain).
An average price can be skewed drastically in one direction or another by a few outlier sales.
It turns out that’s the case this year since there were fewer million dollar sales. Compared to last year, the average dropped a lot since there were fewer outliers near the top.
Sedo’s analytics team kindly calculated the media domain sales price for me. This year’s median was $500, compared to $588 last year. That’s a drop of 15%. So the “typical” domain sale is down 15% year over year compared to the 21% average decline.
You’ve probably heard the expression “there are lies, damn lies, and statistics”. The $500 median may not be representative of the entire domain name sales universe. It’s a combination of mostly investor purchases but a lot of end user purchases as well. But Sedo probably has the best and most diverse data set to work with. Some other sales venues have higher median sales prices, but they don’t let people sell domain names for $60.
Thanks to Sedo for providing this data.
Steve M says
While it’s true that the median is more accurate that the average; even that isn’t as accurate as it could be…which can really only be ascertained by comparing (large numbers of) the sales prices of the same exact domains themselves if/when they’re resold.
We see this today in real estate, where both the average AND the medium prices are down in many of the MSAs in the US…running over 35% in much of my area of central/southern CA…while the RESALE prices of the exact same houses (since there are enough to use for statistical accuracy since in the large majority of cases sales prices are readily available through county records); as gathered in the widely quoted S&P/Case-Schiller organization, show decreases running in the 20-30% range.
This is the true price decrease number/s.
Since this is admittedly very difficult if not impossible to do with domains…the median is the next best yardstick to use.
Sam Nunez says
True, but in this case, neither the median nor the average conveys fully what’s happening in the aftermakret over the last few months. We are seeing a very different picture if you analyze the lower end versus the higher end of the market.
Some at Sedo believe these shifts are fundamental and may persist post recession.
I’ll include some of our market segment analysis in my presentation at Domainer Mardi Gras next week.
TheName.co.uk says
I’d be interested in the truncated mean – where the top and bottom sales are discarded to make the average more statistically useful.
Andrew Allemann says
@ TheName – I suspect it would be very close to the median.