IT.COM

analysis Domain Name Sales 2015-2019 - Part 2: Average Prices

Spaceship Spaceship
Over the past five years the average price for domain name sales listed in the NameBio database is about $1473, compared to $1306 for year 2019 alone. In the first part of this series I looked at domain name sales volume, while this time I look at average price trends.

I start by considering apparent average prices, that is simply the average of all sales over $100 listed in the NameBio database for different domain extensions and time periods. It is important to keep in mind that these only reflect the venues that report sales information to NameBio.

NameBio-reported domain sales are a mix of wholesale and retail sales. This can mask true trends. For example, if more wholesale acquisitions are in the database for a certain time interval, the average price may trend down, even though the actual retail prices are going up.

In the last part of this report I extract a subset of NameBio sales data to get a handle on retail-only price trends. The all-time average retail sales price is probably of the order of $6400.


Legacy Extensions

The graph below looks at apparent price trends for all domain extensions, as well as individually for each of the major legacy extensions, .com, .net and .org. Keep in mind that these are not retail-only prices, but rather a mix of wholesale and retail sales.
01Price_AllLegacy.png

The main trend is one of relatively constant prices. Note that the sale of voice.com for $30 million significantly influences the 2019 average. If that single sale was removed, the .com average price would be about $277 less.

The average prices for .net and .org are less than for .com, but have remained relatively stable, although .net has been edging down since 2017.


.IO, .ME, .CC and .US

With the same price scale as used for the legacy trend graphs, I had a look at the four-year trend in four different country code extensions.
01_US_CC_IO_ME.png

The .IO extension, after rising for several years, was down in average price in 2019.

As mentioned in Part 1, .ME has had a strong 2019 in sales volume. Most of that is due to an increase in number of sales, rather than average prices, however. Prices in the extension have remained relatively constant.

The .CC extension is trending upward in average prices.

The .US extension is down in 2019 compared to 2018, although not a huge change from the four-year average.


.CO, .TV, .AI and .DE

The scale I used for the legacy domain extension graphs was insufficient for average prices for some extensions. Keep in mind that the graphs below have a maximum price of $6000, compared to just $2000 in the earlier graphs.
03Price_CO_TV_DE_A!.png

The .DE and .TV extensions both showed strong, and relatively constant, pricing over the four years.

While sales volume is up in .AI, it is surprising that the average price has gone down significantly in that extension over the last two years. I think this is primarily due to a change in expiration sales, resulting in more wholesale acquisitions within the sales.

The .CO sales price pattern is strong, and seems to be edging upward in recent years.


All Country Code and All New gTLDs

If we lump together all country codes, the price trend is fairly constant year-to-year, and not much different from the average value for .COM presented earlier.
04Price_Country_New.png

If we take all new extensions, the average price is significantly higher than in the legacy extensions, and is fairly constant over the four years in the graph. These sales are a mix of registry and domain investor sales, however.


Retail Sales

As wholesale prices have edged upward, more and more retail sales are included within the publicly-available NameBio database. This is simply because more acquisitions are now above $100 in value.

To try to get a sense of how retail pricing is changing, I restricted the venues in a NameBio search to sales from the following venues: private, Sedo, BuyDomains, Uniregistry, NamePull, Domain Market and Afternic. While not every sale at these venues is a retail sale, the vast majority should be. Some retail sales do take place at other venues. With this selection of venues, only 5.1% of the overall .COM sales are considered retail, accounting for about 5550 sales in 2019.
05_Retail.png

Since I had not recorded sales from just these venues from 2016 and 2017, I could only readily present 2019 and 2018 data by year. I also include data from the three and five most recent years as a group, and for all time.

At first glance it looks like 2019 was a great year for .COM average retail prices, but that is deceptive. In such a relatively small group of sales, the voice.com sale has a huge impact. I show with the arrow the average price if that single sale was excluded.

Without the voice.com sale, it appears that while retail prices in .COM are relatively constant, not much different from the 2018 average of $6243, and the all-time average price of $6455.

The .COM average retail price is higher than either the new domain extension or overall country code average retail prices, although not by a huge factor. In 2019 the new extension retail prices averaged $4495 and the country codes $4413, compared to a voice-sale extracted value of just under $5900 in .COM.

It should be kept in mind that actual retail sales, in all cases, are probably substantially less since, except in auctions, Sedo only report to NameBio sales of $2000 and above, unless individually reported by seller. Also, Uniregistry only report their top-value sales, as is true for several of the other retail venues.


Other Extensions

I had a look at price trends for numerous other extensions. I report on some of them below.
  • While edging downward, .INFO prices are overall fairly constant. 2017 was the best year for that extension, at an average of $851, up from $396 in 2016. The average prince in 2018 at $650. and slightly down at $604 in 2019.
  • .APP prices are strong, from $4261 in 2018 to $6121 in 2019.
  • While .XYZ had a poor year volume wise in 2019, there was not much difference in average prices between 2018 at $1408 and 2019 at $1331.
  • Some of the country code extensions with smaller numbers of sales had the highest average sales prices. For example, in 2019 the average sales price for Australia’s .com.au was $40,900!
  • Canada’s .ca also had a strong average price of $9318 in 2019, and $7127 in 2018.
  • The .UK extension was up sharply in sales volume in 2019, and that was partly due to a relatively robust average sales price of $1936. However, that average price is lower by more than a factor of two compared to .CO.UK in the same year that averaged $4213.
  • The .LY extension is used in many domain hacks. Prices have been fairly steady, ranging from a high of $566 in 2017 to a low of $414 in 2016. The average in 2019 was $470.
  • Although the volume of sales in .GG is up in 2019, average prices are not, dropping from $1679 in 2018 to $835 in 2019. That is probably influenced by many more wholesale acquisitions, however.
  • The average price for .BIZ domain sales in 2019 was $462, down substantially from the $925 in 2018, although only slightly down from the average sales prices in that extension in 2016 and 2017.

Notes:

Nothing in this report should be considered domain investment advice, and is offered for educational information only. If any information in this report is important for investment decisions, you should independently verify the data.


Thank you once more to Michael Sumner of NameBio, for creating and maintaining the domain sales data resource used for this analysis.
 
Last edited:
47
•••
The views expressed on this page by users and staff are their own, not those of NamePros.
fantastic information - thanks
 
3
•••
3
•••
Thanks for your time & efforts putting this together, @Bob Hawkes!
 
5
•••
Thank you @BobHawkes! As always...your posts are invaluable! I drink your "kool-aid." :-P
 
2
•••
3
•••
3
•••
2
•••
2
•••
Thanks for the info and the time it took to put this together...useful to domain investors!
 
3
•••
Thanks @Bob Hawkes great work as usual!

I want to comment on this part:
Note that the sale of voice.com for $30 million significantly influences the 2019 average. If that single sale was removed, the .com average price would be about $277 less.

Yes that is main problem when using average values, they are usually skewed by very high sales. More accurate is using median sale price, which gives more accurate picture. But finding median value is not easy thing to do, you need to have a list of all sales, sort them, then take middle value. At Namebio that cannot be done because they limit the results to only 100.
 
Last edited:
2
•••
Part 2 of Domain Name Sales was as informative and as insightful as Part 1 and, as detailed!

A great contribution to the Forum, thanks Bob!
 
4
•••
Thank you for your kind comments.

At Namebio that cannot be done because they limit the results to only 100.
With a subscription plan it is 2500, so possible for some TLDs, but still challenging with big datasets.

The approach I use to find the median in NameBio for big sets is to first set the minimum value to my guess where the median is approximately. See how many of the set is above that in the search. For example if the original set had 20,000 I guess a median of $500 and with that there are 12,000 above. Then I try a bit different number until I get 10,000 above and 10,000 below.

For a brief period of time NameBio did provide median in the stats, but it was slowing the searches down too much, as I understand it.

The median values are much lower - I compute it for each day in my daily tweet and for com or all sales it is typically $230 to $320 most days (it has edged up from a couple of years ago). This is just telling us that the majority of sales, even above $100, in .com and overall are actually wholesale acquisitions.

Bob
 
Last edited:
2
•••
Thank you Bob for your time and valuable information.
 
3
•••
Thank you for your kind comments.


With a subscription plan it is 2500, so possible for some TLDs, but still challenging with big datasets.

The approach I use to find the median in NameBio for big sets is to first set the minimum value to my guess where the median is approximately. See how many of the set is above that in the search. For example if the original set had 20,000 I guess a median of $500 and with that there are 12,000 above. Then I try a bit different number until I get 10,000 above and 10,000 below.

For a brief period of time NameBio did provide median in the stats, but it was slowing the searches down too much, as I understand it.

The median values are much lower - I compute it for each day in my daily tweet and for com or all sales it is typically $230 to $320 most days (it has edged up from a couple of years ago). This is just telling us that the majority of sales, even above $100, in .com and overall are actually wholesale acquisitions.

Bob

You did well to get approximate retail values when you restricted venues to only Sedo, Afternic, Uniregistry..etc. The average price jumped from around $1200 to around $11,00? But I think that number is still skewed by top sales even without voice.com. Thats why median is prefered in real estates:
https://www.tripsavvy.com/median-vs-average-what-the-difference-2682237
 
Last edited:
2
•••
But I think that number is still skewed by top sales
I agree. It is biased, because of the Sedo not reporting under $2k and Uniregistry only their top sales, and also skewed in the sense that it is an average, whereas more typical of what an individual investor might encounter would, as you say, be the median. However, if one is combining with sell-through probabilities strictly speaking it is the average, not the median, that should logically be used. However, the vast majority of individual domainers will never see the big sales, so I agree with your point that it is skewed

FYi for 2019 .com only, retail only, here are the numbers for the 5551 sales
  • average sales price is $11.300
  • with voice sale removed average price is $5896
  • median sales price is $2500
If one looks back at previous years and longer period the average price is slightly higher. I did not compute the median for those, but I suspect not much different.

Thank you for your input.

Bob
 
2
•••
.de average price surprised me
 
0
•••
Bob, thank you for this research. Any reasons you use average instead of median?
 
1
•••
Bob, thank you for this research. Any reasons you use average instead of median?
If combining sell-through rates it really is average prices that should be used for consistency. However, that said, the high outlier prices introduce effects that most typical domainers will never see. The problem, of course, is that they are also impacted by the wholesale acquisitions that are in the database, and with acquisition prices going up this has probably changed over the years.

I do track median prices every day (do one Tweet per day that gives them) and at some point I am planning to do a long term study of how they have varied (I have been doing this daily for a bit over a year and a half).

I am not sure why, but I believe the data will show that overall median prices have generally been edging upward. It used to be that the daily median was often in range of $220 to $235, whereas lately more often it is $265 to $330, with exceptions. This is for all sales >$100, that is wholesale dominated.

One answer to your question is that NameBio stats provide average prices, and not median, so it is much easier to provide average prices. You can calculate median prices of course by hand for each dataset.

In this reply I give the average and median price comparisons for retail-only .com sales in 2019. I suspect that the similar numbers in other years would not be very different.

Since it seems that NameBio now contain more than 50% wholesale transactions, at least in .com, the median is pretty dominated by the wholesale market.

Thanks for your note.

Bob
 
Last edited:
1
•••
Thank you for detailed replay.
If combining sell-through rates it really is average prices that should be used for consistency.
Honestly I did not get this part, could you explain this point?
 
0
•••
Thank you, Bob for this valuable info! Would you consider / are you able to share the data in numeric / table form? This could be extremely valuable especially to those of us working with ccTLD's and new gTLD's. Thanks again!
 
1
•••
Thank you for detailed replay.

Honestly I did not get this part, could you explain this point?

It is essentially if you are using average sell-through rates for a population, to get sales volume, it is the average price they need to be multiplied by. Perhaps this analogy might help....

Imagine a simple model in which I throw a dice and depending on results (which we will interpret in terms of domain sales) if I roll a 1, 2 or 3 there is no sale, if I roll a 4 or 5 we will in simplified model assume a $10 sale, and if I roll a 6 it will be a $100 sale. Assuming the die is unbiased, in this we would expect a 50% sell-through rate. The price when there is a sale is $10 for 2/3 of the sales and $100 for 1/3, resulting in an average selling price of $40. However the median sales price is $10 since 2/3 of sales are at that level. But if I had say 500 opportunities, resulting in 250 sales, I would not multiply by $10 to get my sales volume, but rather the $40, since that is the average of my sales.

Now, having said all of this, for any one person really they should be using their personal sell-through rate and their personal average sales price. Those with say 5+ years of results and sales of 100+ per year can have rather good personal numbers to use. For someone starting out, or with a very small portfolio, probably the best that can be done is to use industry-wide average numbers, but it is important to realize that those are not necessarily very accurate for your own case. It probably makes sense to try to adjust the average prices to make them somewhat applicable to your own situation. Some do that with median, and it does at some level make sense that you line up all the sales and pick the middle one, but strictly speaking, you should, as indicated in above example, be using the average of a population that is similar to your own situation.

I hope this helps. Thanks for quesition.

Bob
 
1
•••
Thank you, Bob for this valuable info! Would you consider / are you able to share the data in numeric / table form? This could be extremely valuable especially to those of us working with ccTLD's and new gTLD's. Thanks again!

Thanks for your comment and request. It is something that I have considered, and it would be possible as the data I use in the analysis I keep in a MacOS Numbers document. That said, I think I am not inclined to generally distribute it (although may at some point change my mind).

It is partly just I put some work into developing it over past couple of years, plan to add each year to add to trends, and just feel a little selfish giving that away.

But there are some other concerns. I can express the various caveats (like what is included and not in NameBio, changes over years, time related impacts, etc.) but once I generally release it I see a potential danger someone attributing it to me.

Finally, really the data is NameBio and I can I think stand within bounds of fair use (reference only averages, properly credit the source, and only write about it if I am in some way adding something to the data, and not simply taking data). I have not had this discussion with NameBio, but hope they feel similarly. I worry if just giving my summary away, I lose the certainty of similar fair use principles.

Now your specific question re ccTLDs and new gTLDs, I want to stress that while I did look at a number of TLDs not presented in these reports, not very many, so I think how helpful it would be is rather limited.

Also, anyone can access for themselves from the NameBio database almost the same data (admittedly not on a year by year basis for the earlier years since they use a rolling presentation for that). NameBio allows you to readily look up data for past year, past 2 yr, past 3 yr and past 5 yr. Not quite what I presented, but you could derive almost the same thing from the data with any spreadsheet. For example, if you took the past 3 yr and subtracted past 2 yr, that would give you yr 3 data only.

Anyway, a long answer to say, yes the data is in a Numbers document but at least at this point I am not inclined to just make it totally freely available. But I continue to consider the request.

Bob
 
0
•••
Back