Learning from the Master – How Mike Mann Values a Domain Name
I’ve developed an algorithm using a small number of easy to gather data points that values domain names with incomparable accuracy to the dozens of other models I have built and any other domain valuation tool available publicly.
It’s no coincidence that this algorithm in particular preforms so well, it is after all based on DomainMarket.com’s publicly posted domain valuations which are in turn based on DomainMarket.com’s internal valuation algorithms.
When I spoke to Mike Mann – domaining legend, current owner of approximately 350k domains and the founder of DomainMarket.com he confirmed that DomainMarket.com have developed internal models for valuing domains.
I analysed a random 10,000 domain sample from the Mann’s billion dollar portfolio which he had published with buy it now prices.
These buy it now prices represent what experience in the market tell the experts a domain is worth – it’s value.
They are in fact a much better data point for analysis than retail sales which even if controlled to one platform are erratic and inconsistent.
By analysing asking prices you control for the whimsical nature of auctions, the ability of the buyer to negotiate, inconsistent pricing, oversupply leading to unpredictable prices and the sales platform.
I gathered 14 data points for each of the domains in the sample; the usual kind of data domainers look at before purchasing and ran a whole pile of statistical analysis on the dataset.
By figuring out what one of the industry’s most experienced veterans believes impacts a domain’s value you can then extend those learnings to your own buying decisions in a much less costly way than the trial and error that builds that experience.
Put nerdishly: I was modelling a model.
Put normally: I was building a mathematical version of Mike Mann.
Just 4 variables
I’ve always been irked by the proposition that you needed a colossal pile of data points on a domain name to be able to value it effectively.
Of course the only way to disprove this theory was to include those data points in my analysis and show them not to be predictive of domain value so that’s what I did.
Time and time again my research shows that the same 6 metrics are the only 6 metrics that are statistically significant predictors of domain value, the rest are just noise.
In DomainMarket.com’s dataset just 4 of those predictors were significant (in order of importance):
- Search volume
- Number of other extensions registered for domain
- TLD (I had a limited amount and unrepresentative dataset for TLDs ~99.5% of domains were .com)
I think it’s amazing that 66% of the variance in DomainMarket.com’s asking prices can be explained using these 4 simple data points.
This level of prediction is unrivalled in any other model I have built or seen predicting domain values.
The average buy it now price for the random sample was $4,224.
Exact match search volume was by far the most influential predictor in determining the buy it now price; holding all other variables constant each extra unit of exact match search volume resulted in a $2.18 increase in the buy it now price. On average each domain had 1,104 units of search volume.
This is an incredible model that gives unprecedented insight into what drives the value of domains.
If you want to achieve success like Mike Mann then you have to think like Mike Mann and buy domains like Mike Mann.
To do that you need to buy domains based primarily on the above data points.
Plug: I’ve developed a more rounded proprietary algorithm based on buy it now prices from a number of portfolios as well as retail and wholesale domain sales data which I’ve built right into the DropMining product.
We rank each of the 200,000 daily dropping domains in order of value, saving you time by cutting down those 200,000 domains to a select few high value prospects which you can then analyse with the human touch and your domaining experience.
Go sign-up to be notified when we launch.