Selling Domains on Flippa – The Data Driven Guide Part 2
I think its a fascinating marketplace and that while its still emerging in the domain sales field it definitely has the potential to become a major player on the retail side, especially in domains ripe for development.
In my last post I looked at the marketplace as a whole and how domains were fairing.
I analysed 142 domain sales of the 1,105 domains listed that occurred in the two month period I conducted the study.
Today though, I want to focus on what influences the sale price of a domain on Flippa.
Here comes the data…
I’m a data nerd, trust me humans are rarely better at determining prices than a good ole bunch of numbers.
So instead of looking at all the sales and figuring out whether they looked good or not I ran a regression analysis on a number of factors I thought might influence the sale price of a domain on Flippa.
Regression is the holy grail of statistics and economics – it is used globally as primary means to measure the relationships between different pieces of data.
If you have good data to predict sale prices you can create a model (algorithm) that can be used to determine what influences prices on Flippa and this model could be used to predict the likely selling price of another domain coming up on the selling block.
By way of an example, an economist might gather data such as IQ, parent’s educational attainment, socio-economic status and years in education to predict a person’s income.
If you have this historical data and a person’s income for a large number of people you can then isolate how much of an impact each extra year in education has on an average person’s income.
This is how we have proven the causal relationship between more education and increased income.
I gathered data on the domain length, number of keywords, exact match search volume, CPC, competition, Alexa ranking, PageRank, number of other extensions registered, the domain’s TLD and whether it was in DMOZ for all 142 domain sales. I also collected data such as the number of days the auction lasted and whether the seller bought any upgrades.
I knew that even if the analysis did create a good model for domain prices on Flippa, that due to the small sample size the likelihood was that the results couldn’t be extended out to any domain that goes up for sale on Flippa.
So the predictive nature of the regression model was pretty much out, but perhaps the analysis could tell us something about the domains that had already been listed and sold.
Unfortunately the data I plugged into the regression analysis was in no way predictive of sales prices.
All this data only predicted 25% of the variation in prices from domain to domain. That by the way, is pretty bad – but not unlike many other domain sales platforms.
Very few of the data points I collected were close to being statistically significant.
Adding a buy it now option and setting the auction length to 30 days were among the best predictors of a higher sale price.
Controlling for the other factors mentioned adding a buy it now option, on average increased a domain’s sale price by $385 and setting the auction length to 30 days bumped up your return by $485.
While the dataset is small and their level of significance was certainly not at the level I would want; I can recommend you choose these two options when selling on Flippa.
As seen with other sales platforms data points such as search volume, number of other extensions registered for the domain and the domain’s TLD were the most influential on the sales price – all being good measures of the value of a domain but in terms of Flippa specific influencers they were no more influential than any other platform.
Even if the evidence isn’t overwhelming it is more compelling than the other options, plus adding a buy it now price or letting your auction run for 30 days won’t cost you a dime.
The failure of this model doesn’t mean that Flippa sales in general can’t be predicted with good data or that the other factors I tested don’t matter, logic shows that not to be true.
But what it does show, is that even with a large dataset there is unlikely to be any particular types of domains Flippa buyers like over other sales platforms.
Whether or not Flippa is the right place to sell your domains is a tough question.
I would say based on my previous post’s data, the high selling costs and the unpredictability of the market definitely not for a full portfolio.
But if you are selling a top of the range premium domain, their substantial web presence, smooth and trusted system and optional upgrades might be a nice place to host an auction if you are willing to promote that auction in traditional domaining spheres.