IT.COM

data Domain Data: Just 66% of YCombinator's Class of 2018 Use .COM

Spaceship Spaceship
It’s very rare that a brand new startup with minimal funds would initially create their online presence on a premium .COM domain. Bootstrapped businesses may, therefore, opt to go with an alternative extension, create a brandable name or use a subsequent imperative word such as “Get” or “Buy”.

One of the most active funds for startup capital is YCombinator. Founded in 2005, YCombinator acts as a seed accelerator, offering early-stage investment to around 1,900 companies including AirBnB, DropBox and Reddit.

Twice a year, summer and winter, YCombinator releases a list of companies it has invested in during that season’s batch. These lists can be a good indication to naming trends and domain name preferences amongst the startup community. Below, we’ll aim to find out what type of domain preferences YCombinator startups have in 2018.

We have a dataset of 234 domain names, gathering data from YCombinator’s list of both S2018 and W2018. The list of companies is publicly available on YCombinator’s website.


Domain Extensions

In a 2016 Domain Data article, we analyzed the list of every single company that YCombinator had invested in from 2005 to 2016, which totaled over one thousand names. In that article, we calculated that 86.45% of domains were .COM.

In YCombinator’s class of 2018, that figure has dropped to 66.23% with many more companies seemingly happy to go with other domain extensions instead. We have a number of startups opting to use New gTLD alternatives such as mojo.video, emojer.app, and buycoins.africa.

Both .CO and .IO remain popular with startups, with a combined 17% of this year’s YCombinator startups using them. Some of these names are one-word domains that would cost at least five- or six-figures to acquire as a .COM, but as .CO or .IO, they were likely relatively cheap to acquire. Some examples include Torch.io, Atrium.co, and Ben.co.
pasted image 0.png


Several years ago, there was a trend amongst startups to create their online presence on a .LY or a .ME domain. These aren’t prominent extensions at all now, with one .LY and three .ME domanis being used by the class of 2018.


The Domains

With .COM established as still being the most popular extension, it’s interesting to see what type of .COM domains are being used by these startups.

There are 16 companies using .COM domains that would be considered to be one word - either dictionary .COM’s or a “brandable” word. Usually, these brandable names match the company name such as Trusu.com, Delphia.com or Berbix.com.

However, there are a couple of examples of premium .COM names being used. Artificial Intelligence company Molly smartly acquired the Molly.com domain name and cargo company Shone has Shone.com.

The large majority of these .COM domains consist of two words. Some of these two-word names are exact-match domains to the company, for example, Sharpest Minds owns SharpestMinds.com. However, many startups have opted to use an imperative verb (Try or Get, for example) along with their brand name. Here are some examples:
  • GetOxygen.com
  • GetShogun.com
  • GetStation.com
  • OnGrin.com
  • OnHexel.com
  • SoBump.com
There are six companies using “Get” as a prefix, three using “On” and one using “So”.


Length

In our 2016 article, the average length for 1,100 YCombinator company domains was 8.56 characters with several business using three-letter .COM’s. In our current data set, just two companies are using three-letter domains (not .COM): Ben.co and GFI.org. The shortest .COM is JITX.com.

The average length of a domain in this data set is 9.05, slightly longer than our 2016 data. The longest domain used is twenty characters: DataDrivenBioscience.com. This is the exact match domain name for that company.

As we established above, the most frequent non-.COM individual extensions in our list are .IO, .CO and .AI. Traditionally, these extensions may have been used as direct alternatives to shorter .COM domain names that are unattainable.

Therefore, in theory, the .IO, .CO and .AI domains in our list should be shorter than .COM’s average length of 9.05 characters. The average character length for .IO is 7.3, .CO is 8.5 and .AI is 5.8. This data would support our theory.


Aftermarket Domains

How many of the domains used by YCombinator’s class of 2018 were acquired on the aftermarket? As it turns out, NameBio has listed just three sales of domain names from our list.

The first is JITX.com, used by JITX, a circuit board design company founded in 2016. The company looks to have paid $5,988 for the name in 2016, around the time they launched. DomainIQ’s WHOIS history shows the domain was briefly in the possession of Jonathan Bachrach, JITX’s co-founder before it went under privacy protection.

The second name is OnUniverse.com, acquired by website builder Universe. This name was acquired in June 2015 from BuyDomains for $2,188, a year after the company was founded. Since March 2017, the company has raised $7.3 million in funding but has opted to keep the same domain.

Our third and final domain from the aftermarket is Torch.io, purchased on the Park.io platform for $550. It’s unclear as to whether Torch.io was purchased by an investor and subsequently sold to the company that’s using it now, Torch Labs. The only data available shows the domain was purchased in February 2017 on Park.io for $550. According to a LinkedIn page, Torch Labs was founded in September 2017.
 
34
•••
The views expressed on this page by users and staff are their own, not those of NamePros.
5
•••
66% .COM, but still shunning nGTLD I see. 4% :smuggrin:

Brad
 
5
•••
So in 2018 out of 234 startups only three bought an aftermarket domain. One was a LLLL.com, another a two-word .Com and the third a one-word.io. So several years after the "new" TLDs were launched and with perhaps 15 million aftermarket domains in the nnewer extensions, what were the odds in 2018 of selling an aftermarket new TLD domain to a Ycombinator startup?

ZERO
 
5
•••
So in 2018 out of 234 startups only three bought an aftermarket domain. One was a LLLL.com, another a two-word .Com and the third a one-word.io. So several years after the "new" TLDs were launched and with perhaps 15 million aftermarket domains in the nnewer extensions, what were the odds in 2018 of selling an aftermarket new TLD domain to a Ycombinator startup?

ZERO

I agree that new tlds have little chance with tech startups, but assuming that NB gets every sale listed is wrong.

I sold 35 names in 2018 for average of over $4K each and I would be surprised if any of those made it to NB.

Try anem/com and jouz/com for example that sold for combined $37K+. Both would qualify as startups (wind energy and vaping device). And both are 4L. Double of what yCombinator startups have bought ;)
 
7
•••
Generally, I would caution not to use YC data as a proxy for the overall market. There is, obviously, YC coaching bias here, where they learn the same things. Few years back they were told that .ly was good idea, now they are told .io or .ai are good.

If you read the article by Molly founders, YC specifically advises against spending large amount on a domain, as they deem it unnecessary at this stage. And apparently, they advise that adding get or on to the brand is fine, which is a poor advice, imo.
 
7
•••
So just using the data above:

.AI > all new gtlds combined
.IO > all new gtlds combined
.CO > all new gtlds combined
 
2
•••
0
•••
1
•••
Excellent article, although as noted by earlier commenters there are biases so this does not (necessarily) represent the overall startup field domain choice. There are undoubtedly also geographical biases based on source.

Even recognizing that NameBio certainly has a minority of all aftermarket sales, I found it amazing that only 3 of the 200+ are clearly aftermarket sales (and 1 of those perhaps not).

Bob
 
2
•••
Of those that opt to add a verb, it's worth noting their strategy is basically to have wild success and then pay a huge sum for their one-word .com later. I invested in a YC startup that used 'get' and they've been trying to negotiate with the .com owner since the first time they raised money. That owner effectively has stock options in the company. If they blow up, he's going to get a huge purchase price. Though I guess he's better off than other shareholders in that if they fail, at least he'll still have his domain lol
 
7
•••
Excellent article, although as noted by earlier commenters there are biases so this does not (necessarily) represent the overall startup field domain choice. There are undoubtedly also geographical biases based on source.

Even recognizing that NameBio certainly has a minority of all aftermarket sales, I found it amazing that only 3 of the 200+ are clearly aftermarket sales (and 1 of those perhaps not).

Bob

Is it amazing? Depends. If only 1 in 50 names get reported to NB, 3 names from NB reports might mean that 3x50=150 out of 200+ were purchased in aftermarket.
 
2
•••
Excellent article, although as noted by earlier commenters there are biases so this does not (necessarily) represent the overall startup field domain choice. There are undoubtedly also geographical biases based on source.

Even recognizing that NameBio certainly has a minority of all aftermarket sales, I found it amazing that only 3 of the 200+ are clearly aftermarket sales (and 1 of those perhaps not).

Bob

As far as new gtld's go, percentage wise it's about the same as when DNGeek was keeping track of them, last time - https://dngeek.com/2018/01/88-newly-funded-startups-domain-names-rootcloud-com-firstagenda-com/

https://www.namepros.com/threads/ge...coming-mainstream.1062072/page-3#post-6545214

That last one 3/88 3.4%. All the others were low as well, from about 2% - 3% of startups picking a new gtld.
 
2
•••
If only 1 in 50 names get reported to NB

Yes if that is true, then the numbers are, as you say, not at all surprising.

I'm not saying that it is impossible the ratio is that low but with Sedo reporting those >$2K unless one party paid extra fee for privacy, and considering their share of domains for sale, it seems to me a 2% ratio requires (at least) one of the following:
  • Selling rate at Sedo and other venues that report to NameBio is much lower than the rate at other venues that do not report.
  • A very high ratio of buyers/sellers pay the fee to keep sale information private.
  • Vast majority of sales are below $2k at Sedo.
I am just thinking out loud here. If NB data does represent 2% of sales, domaining is doing better than might first seem to be the case. I think it is possible that one, or even all 3, of the possibilities I list are true. Clearly in their famous blog post Verisign assumed the ratio must be very low.

Anyway, thank you for your contributions. Much to think about.

Bob
 
1
•••
Yes if that is true, then the numbers are, as you say, not at all surprising.

I'm not saying that it is impossible the ratio is that low but with Sedo reporting those >$2K unless one party paid extra fee for privacy, and considering their share of domains for sale, it seems to me a 2% ratio requires (at least) one of the following:
  • Selling rate at Sedo and other venues that report to NameBio is much lower than the rate at other venues that do not report.
  • A very high ratio of buyers/sellers pay the fee to keep sale information private.
  • Vast majority of sales are below $2k at Sedo.
I am just thinking out loud here. If NB data does represent 2% of sales, domaining is doing better than might first seem to be the case. I think it is possible that one, or even all 3, of the possibilities I list are true. Clearly in their famous blog post Verisign assumed the ratio must be very low.

Anyway, thank you for your contributions. Much to think about.

Bob

Thank you.

First, even at Sedo, under $2K sales represent considerable portion.

Second, Sedo is really small and majority of the sales via it are from landing pages, i.e. the site itself contributes next to nothing. I have had just 1 sale in last few years there and that was a name that was not even listed there. Just a company hired their broker. With almost 500 names listed, they cannot contribute even a single sale, so don't even see point in adding more there. In the meanwhile, Afternic sold tens of names for me without even a landing page.

The matter of fact is that most sellers don't see point in taking extra effort to share the sale and provide a proof on top.

With that, NB can create a bit distorted picture of the domain market, as it is disproportionately affected by Sedo $2K+ reports and Godaddy Auctions $xxx sales which are mostly just reseller restockings.

Again, I sold 35 names last year via direct, Afternic/GD, Sedo, BB, BR, BP and zero of those got reported.

So, to conclude, while fun to read, the above statistics are too limited and have too much bias in it to draw any conclusion.

And also think how the facts are represented here: .com got "just 66%", while all the rest got the "whopping 34%".

In other news, Golden State Warriors got their ass whooped by Denver Nuggets, as the Warriors managed to win by the margin of just 31 points. And although they managed to set NBA record for a first quarter by getting 51 points, their chances of threepeat look even slimmer after this miserable performance.
 
4
•••
Overall I am not that surprised by the big picture breakdown by TLD. Certainly not surprised that 66% chose .com, and the 9% and 8% that chose .ai and .co for moderately large startups in this era with so much AI related makes sense. Tracking carefully ngTLD sales for past year, I was not surprised that only 4% chose a ngTLD. I might have guessed the cc would have been a bit larger. The one number that surprised me most was the 2% .org. I know most would not consider it but with trends to crypto/blockchain and crowd/peer where it does find use in companies, surprised not more. I might have thought one or two would have started on .net so that surprised me too.

A superb analysis by James. Nice to have full list at YC where can click to go to the websites. Thanks to all.

Bob

Ps I guess it would not be that hard to deduce from Whois creation dates, but I wonder what fraction of these went with a hand reg?
 
Last edited:
3
•••
It's OK, we can still sell domains to each other for ~$9. :)
 
2
•••
1
•••
It was mentioned by @Recons.Com that YC give certain advice re what to spend (or not spend!) on domain names and also re extensions. No doubt that influences the choices.

Does someone know typically do startups in this class hire early on branding experts to help them with a naming decision, or do they tend to go it alone, or perhaps just search the brandable sites for the best available option?

I think established firms most of the time would use consultants and or in house branding and marketing experts before a rebrand. I suspect, but could be wrong, that is less true for early phase startups.

Bob
 
2
•••
Got any data on how many from past rounds are still around, how many upgraded out of their non.com or from 2 word to 1 word . etc ?
 
3
•••
Thank you, James, for the data and great explication.

The statistics have changed in one year in favor of the extensions other than .com. I think the ratio of .io has increased.

NPs can compare the posts to see what's changed:
Domain Preferences of Y Combinator Companies
 
4
•••
Back