[Editor’s note: I have been critical of both the number of .art domain names with premium prices and the number of tiers. In this article, Kurt Pritz, Member of the Strategic Planning Board and Operations team member for .art, provides an update on how the strategy is working. There are currently about 15,000 .art domain names registered.]
First, let me reiterate the purpose of the .ART premium names strategy. .ART set its standard domain price fairly low, $9. That was done to provide affordable names to the art community. Not surprisingly, over 30% of the .ART registrations to date are of the first-name—last-name type or some version of that. One would expect that where artists are self-identifying with their work. These name combinations are rarely premiums and so serves the art community well. We have also found that families have adopted the first-name—last-name idea, securing names for their kids as well. This .ART cachet with families is something being pursued as a product. The result is that .ART is selling many names to the expanded art community at low prices.
Second, .ART set out to determine a predictive value for each domain. It has always surprised me that domains are typically priced as if they were sold at the Dollar Store rather than the grocery store. It can be intuitively concluded that domains all have different values. The question is whether the differences are sufficient to warrant varying the pricing.
Without making a pre-determination, .ART set out to answer that question through an exhaustive “big-data” collection and analysis approach that gathered hundreds of millions of words and word-combinations from online sources (e.g., Google AdWords and Wikipedia) and static sources (e.g., art reference materials). This occurred over months of online data collection and included billions of queries.
We scored each word combination against 14 separate criteria such as word length (the shorter the better), and frequency of Google search (the more the better) in order to create relative values. We then compared these values against every prior premium domain sale in the past, and, after appropriate normalization, calculated a prospective market value for each of the possible domains. (A paper can be written on the previous two paragraphs.)
Considerable work was done after the fact to ensure the valuations were stable and immune to small environmental fluctuations.
The result, as you have written, was 3.5 MM premium names. This caused .ART to identify these assets as “inventory names.” Does .ART expect to sell them all? No. (Even if they were sold as standard, .ART wouldn’t expect to sell them all.) But there are two positive outcomes, and I see no negatives ones.
One positive outcome is that .ART could build a business plan that could:
- serve the art community by selling domains cheaply to those in the art world (and those outside it) that wanted the sort of first-name—last-name domain that has become popular and is an obvious choice for those displaying their work online; and
- derive needed revenue from more-highly valued names that were sold to those who valued those names.
The second positive outcome was in the proportion of revenue generated. Shortly after GA opening, only 7% of the names registered were .ART inventory names but that represented 80% of the revenue. As time went on, the percentage of inventory names registered decreased to 5% but the revenue generated stayed above 70% of the total. These results are without the benefit of any premium name auctions, promotions or other concentrated efforts.
The vast, vast majority of these inventory names is priced above, but not too much above, standard prices. Over 96% of the identified names come from the bottom 3% of the pricing tiers. If .ART had followed traditional premium names models, these lower-valued inventory names would have been priced as standard names and much of that revenue would have been foregone. (That revenue is not a windfall; as other registry operators will tell you, it is vital to ongoing operations.)
I have found that, compared to other registries, .ART prices premiums either modestly or in-line with others.
The .ART inventory name model also provides an excellent opportunity for registrars. Registrars realize higher margins on inventory names and some have seen revenue ratios approaching that of .ART. When using .ART rebates and tools that point registrants towards inventory names, the impact on revenue can be significant.
The algorithmic prediction that there is a “long tail” of modestly valued premiums falls into line with marketplace thinking that is over ten years old. When I managed engineers (if such a thing is possible), our mantra was to trust the numbers. The .ART algorithm was carefully designed. Even though this algorithm is, in some ways, the first of its kind, the results should not be disregarded in favor of instinct or past practice.