2/3rds of searches include a TLD, 44% of searches for new TLDs came back unavailable.
eNom revealed some interested stats about domain name searches in its most recent newsletter:
- Users performed an average of 2.6 searches per session.
- One-third of all searches were made without specifying a top-level domain.
- Of those searches that did specify a TLD, 20% were for new generic TLDs.
- 53% of searches for legacy TLDs came back as unavailable for registration. The same was the case for 44% of new TLDs.
The last bullet point is quite fascinating. If all the good names in legacy TLDs are taken, then can you say the same thing about new TLDs? 53% vs. 44% unavailability for “old” vs. “new” isn’t that much of a difference.
I suspect this has something to do with people searching for higher quality second level domains under the new top level domains.
Christian says
A lot of the new gTLD registries have restrictions on premium domains, which might also in part explain why 44% are getting “unavailable” when searching for domains.
Andrew Allemann says
Yes, but gets back to people searching for single keywords with high value. There are also collision list domains, but I think those are mostly out now.
Joseph Peterson says
So if you’re searching for domains in legacy TLDs, then the probability of finding something available is (on average)
47% after 1 domain lookup
72% after 2 domain lookups
85% after 3 domain lookups
92% after 4 domain lookups
96% after 5 domain lookups
98% after 6 domain lookups
99% after 7 domain lookups
99.8% after 10 domain lookups
99.95% after 12 domain lookups
That means only 1 in 2000 people will fail to find an available domain within their first 12 ideas in legacy TLDs. Actually, 9 out of 10 people find something available within their first 4 domain lookups in legacy TLDs.
If I’m misinterpreting things, please let me know.
Asfas1000 says
Indeed there is some misinterpretation.
The numbers would be valid if the searches were randomly taken from Enom’s database.
But, since these searches are done by the same person, they are not truly random : there is some association between them.
For instance, if someone is searching for decent “Insurance” domains, he won’t find anything.
While another person who searches for an uncommon term, like “microfusion”, may find a lot of available domains.
Joseph Peterson says
@Asfas1000,
You’re entirely right. I suspected that flaw would get me. Hoped somebody would pay attention and force me to retrace my steps.
Here’s why I’m wrong:
In an extreme case, if all my searches are for 2-letter strings, then even after 100 searches, I’ll find nothing available. Whereas if all my searches are for haphazard 40-character strings, then the availability rate will be nearly 100% per query.
In such an extreme case, we might see – let’s say – 1 person searching 53 times for LL stings while a different person searches 47 times for 40-character gibberish.
Between the pair, that would give us the average availability rate of 47%. Yet the first person could continue indefinitely at 0%, while the second person would consistently find everything unregistered.
And if they both kept querying domains in the same proportion relative to 1 another, these rates would remain constant – 0% for person A, 100% for person B, with 47% on average. In that case, there’s no convergence in results because there’s no mixing of behavior.
I suspect that real lookups involve more mixing of domain types than that extreme case. If the pot were stirred such that everybody searching for domains experimented with the same breadth of domain styles, then query success rates would be more like what I calculated.
But I did assume a uniform distribution. And that’s certainly incorrect.
Another situation might be this: Half the people see 74% availability per query, while the other half see 20%. If they each perform lookups at the same rate, then that gives us an average of 47% availability per query. But after 2 queries, the first group would find something 93.24% of the time, while the second group would be successful just 36% of the time. That means an average success rate after 2 queries of 64.62% – not the 72% I’d stated.
Long story short, my reasoning was fallacious. We know that different domain types have different query success rates. Yet we don’t know how those types are distributed among searchers. Much depends on that. We simply don’t have enough information.
Thanks.
Asfas1000 says
Well said.
John says
>”I suspect this has something to do with people searching for higher quality second level domains under the new top level domains.”
I suspect you are right. So much for the public receiving a second chance after .com. And it’s not a second chance if prices are through the roof either.
Asfas1000 says
Well said.
M. Menius says
Ditto, searchers are typing in valuable keyword domains to find them increasingly gone in the new tld’s. Same experience with gmail and twitter names as well.
“I suspect this has something to do with people searching for higher quality second level domains under the new top level domains.”