The Expert’s Algorithm

Feel free to display the infographic on your own Facebook, blog or website – if you are I’d appreciate a link in return.

 

Wouldn’t it be great to know what the top domainers look for in a domain.

Not what they say they look for, but what they actually buy.

Since the success of my analysis of DomainMarket.com’s portfolio and their pricing algorithm I’ve been obsessed about getting my hands on data from the industry’s top players.

This kind of data is normally pretty hard to come by and as most top domainers like to keep the majority of the buying and selling activities private.

When they do publish their data they tend to hand-pick which ones are known to the world.

This restricted data can be interesting but in terms of academic rigour it’s wide open for manipulation – we want to know about the successes and the failures but it’s in a domainers best interest only to publish successes.

What I needed was a complete dataset of purchases by the best people in our industry.

I got thinking about whether there was a time when only the most experienced domainers competed against each other in a public auction.

I found this gem of a dataset from the various Moniker run auctions at exclusive domaining conferences over the year.

I downloaded years worth of auction sales (1414 domain sales) from T.R.A.F.F.I.C and DomainFest and gathered a search volume, CPC, etc. for each domain.

I plugged the data for each event into a regression analysis and came out with a bunch of models that told me using real data what the top domainers look for in a domain name.

Some of the data I would normally gather e.g. PageRank and Alexa couldn’t be gathered because of the historical nature of the sales but for the most part I felt I covered most of the areas I would with fresh sales.

It turned out that only 3 metrics were consistent statistically significant influencers of how much an expert domainer was willing to pay for a domain; exact match search volume, the number of other extensions registered for a domain and whether was the domain a .com.

Interestingly CPC and the length of the domain had no statistically significant relationship with the sale price.

The below chart represents the relative importance of each metric.