A crystal ball for estimating domain name renewal rates.
The U.S. Patent and Trademark office has granted patent number 9,094,467 to Verisign for “Method and system for predicting domain name registration renewal probability”.
Simply: take a number of factors about the domain name and its history to calculate the chances of it being renewed.
Verisign has to run these calculations frequently, and not just for its own forecasting. Because of the renewal grace period, it can’t give final renewal rates until 45 days after a domain expires. As a public company, it gives some guidance to analysts on each quarterly investor call even though it’s not 45 days after the end of the quarter.
Some of the factors the patent embodiments consider:
- registration information
- hosting details that can include nameserver data, webserver data, and/or mail server data for the domain name
- domain lookup statistics for the domain
- information about a website hosted at the domain
- length of time the domain name has been registered
- number of continuous times the domain name registration has been renewed
- the length of time the domain has been registered without becoming unregistered
- status of the website hosted at the domain
- purpose of the website hosted at the domain
You can view the patent here (pdf).
I’d be more impressed if they could predict the MONETARY renewal rates of domains over the next few years. I see they plan to raise .net’s another 10% or so.
I’d be very interested to peer inside those calculations.
I’d would interesting to see the results after the calculation and one thing is for sure that it will add life to renewal rates & associated KPIs along with the domains.
“A method for determining probability of a domain name registration renewal includes receiving a plurality of inputs associated with the domain name and …assigning a weighted value to each of the plurality of categories and calculating the probability of the domain name registration renewal”
So basically they’ve patented predicting renewal rates via linear regression with no feedback loop. Which means competitors will be forced to limit themselves only to the hundreds of other more flexible and accurate predictive analytics algorithms that have been invented in the 100 years since linear regression was cutting edge. Good work, Verisign data scientists!
Reading through it and parts of it are too simplistic (it is a patent and the claims in US patents are always universal in aspiration if not almost completely speculative). There are other factors that influence the renewal probability. From reading it, I don’t think that they have a reliable and working renewal prediction model.