Dynadot

analysis Is GoDaddy Appraisal Value A Useful Filter?

Catch.Club Catch.Club
When seeking names on the buyer requests section at NamePros, or filtering a list of expiring names for consideration, domain investors often use filters to make the size of the list manageable. For example, they may seek only names with a certain minimum age, maximum length, or minimum GoDaddy appraisal value.

In the NamePros Blog analysis Does Domain Name Age Matter, I looked at how useful domain age was as a filter. The correlation was very weak, although it was true that the vast majority of names that sold at retail sales prices of $2000 and up, were aged by a number of years. Just to stress, it is not that age makes a name more valuable, but simply that names that are more valuable tend to also be aged.

Therefore, domain age can be a useful filter, although some good names will be excluded when an age filter is employed.

In this study, I used the same set of domain name sales, but looked at the GoDaddy Appraisal value for each name, to see if automated appraisal value was a useful filter.

The Experiment

I wanted a selection of mainly retail sales at a single venue over a variety of prices. I used the same set of sales employed in the age study. A selection of Sedo .com sales over a variety of price points were used. Sales data was as reported in NameBio. The data is mainly from 2022 and a few from late 2021.

To get a dataset of manageable size, but with domain names with a variety of prices, I selected the first 25 sales starting at various price points from $2000 to $50,000 – see the details in the earlier study. There were a total of 189 sales used in the in the analysis, that sold at prices from $2000 to $1.6 million.

I used the free GoDaddy Domain Appraisal instrument to get an appraisal value for each domain name. That was done recently (early March 2023). Some automated appraisals, for example Estibot, reset values after a sale is reported. To my knowledge, GoDaddy do not seem to do that, and by using Sedo sales data makes it even more unlikely. The fact that no appraisal was exactly the same as the sale price, and few were even close to the sale price, supports the fact that their algorithm does not reset values by NameBio reported sale of that exact name.

How Sales Prices and GoDaddy Valuations Compared

The majority of Sedo sales in this sample sold at prices well in excess of the GoDaddy Appraisal value. For each sale I calculated the ratio of the sales price divided by the appraisal value, with results shown in the graph below.
Image-Ratio.png

Only 18.9% sold at a value less than the appraisal estimate.

A name with an appraisal of $524 sold at $25,000, and another name with an appraisal of $4951 sold at $125,000, a ratio of more than 25x.

There were 15 names, from the 189, that had a GoDaddy appraisal of >$25,000, so precise ratios could not be calculated for these sales. In all cases, the sales of these names were at more than $25,000.

GoDaddy Appraisal did get some values about right. For 41.4% of the sales, the sales price was within a factor of 2 of the appraisal value, that is a ratio between 0.5 and 2.0.

Note that the results of the agreement will depend on the sales data used. I suspect if one looked at sales of $2000 and less the agreement between sales price and appraised value would be significantly better.

Correlation

Shown below is a plot of the sales price versus the GoDaddy appraisal value. Note that 7 sales were off the scale of the graph.

Image-RegressionFull.png

The apparent grouping of data is because of the way the data was selected, sales at about $2000, $5000, $10,000, $15,000 and so on. This was done to have a manageable number of sales, but at a variety of price points.

There is a correlation between sales price and appraised value, but it is very weak. The R2 value is just 0.215, a bit better than the correlation of price with domain age found in the previous analysis, but in a scientific sense it is a poor correlation.

To better see the region where most of the data is clustered, I plotted only the data up to $30,000 below. The graph has a line showing equality of sales price and appraisal value, and you can see the majority of sales are on one side.
Image-Regression-Detail.png

Use of Appraisal Value for Filtering

So, how does using appraised value work as a filter? If we regard this dataset as typical, if you limited consideration to names with $2000 and up appraisal value, you would eliminate 19 of the names, about 10% . The highest sales price in the excluded names would be $25,000, with 10 of the excluded sales in the 5-figure range.

If you set a more rigid filter with $2500 and up appraisals, you would exclude 13.8% of the names that sold $2000 and up.

An even more rigid filter, only names with an appraisal of $4000 and more, would eliminate 23.3% of the names that sold $2000 and up, including several $25,000 sales.

Keep in mind that this is not a random sampling, but one that weights higher value sales more heavily. Also, all sales are from Sedo, rather than a mix across multiple sales venues.

Only you can decide if the worthwhile names that are excluded are an acceptable cost to a much shorter list of names to consider.


My sincere thanks for NameBio as the source of data for this analysis. Also, appreciation to Sedo for making much of their sales data open to the community. And thanks to GoDaddy for making their appraisal tool freely available.
 
Last edited:
50
•••
The views expressed on this page by users and staff are their own, not those of NamePros.

Recons.Com

Top Member
Impact
18,870
Thanks. This is useful... Unfortunately, the result just proves that the valuations are mostly useful as very ballpark estimates and, e.g., a name valued at $1200 can be in fact better than the one valued at $2000.

And if you create a valuation tool that weeds off the really bad ones and follows some simple rules and then assigns values from $1250 to $2500 to 95% of those names, then your tool also will be able to get 40-50% of the actual sales within 0.5 to 2 range
 
8
•••

Fayaz Ahmed

DomainRobin.comUpgraded Member
Impact
456
Thanks a lot @Bob Hawkes for this post. Very interesting as always, especially for a data buff like me. :xf.love:

As for me, I do check GoDaddy Appraisal data, but not for the appraisal value, but for the sales records. Some of those records are not available on namebio.

Also, for gTLDs, I noticed GoDaddy appraisal value to be very poor. Perhaps we can do the same analysis over different TLDs to see how it differs. Also, a Sale-Through Rate (STR) from randomly selected profiles against the appraisal value segments over a period of time would be very interesting. :unsure:



A few more observations:

Note that the results of the agreement will depend on the sales data used. I suspect if one looked at sales of $2000 and less the agreement between sales price and appraised value would be significantly better.
I was thinking the same. Perhaps a segmented analysis would provide in a more insightful result. For example, how does it perform for $500-$1000 sales, then $1000 - $2000 sales, $2000 - $3000 sales etc.

To get a dataset of manageable size, but with domain names with a variety of prices, I selected the first 25 sales starting at various price points from $2000 to $50,000 – see the details in the earlier study. There were a total of 189 sales used in the in the analysis, that sold at prices from $2000 to $1.6 million.
I think it creates a representation bias. For example, the number of sales from $2000 to $5000 would be a lot higher than the number of sales from $25,000 to $50,000.

We can perhaps eliminate this bias by taking the number of records according to the ratio of the number of total sales from that segment. Segmented analysis as I've stated above would also eliminate this bias.

Only you can decide if the worthwhile names that are excluded are an acceptable cost to a much shorter list of names to consider.
I think that's the best conclusion at the end of the day.

At the same time, I think it's also important to keep in mind that many of the sales may not be end user sales. Because a domain that most expert domainers would appraise to be mid $XX,XXX, can be easily sold at mid $XXXX to an investor. Also, some domains get sold because of SEO or marketing values that may be out of the scope of any appraisal tool.

So the appraisal tools cannot possibly figure out all that different reasons that's often hidden behind the scenes. So at the end of the day, as domainers, we'll have to decide ourselves how we value our names, or how we find valuable names.

A side note: I was checking out GPT-4, I think we'll get far better appraisal tools powered by AI within a year or two, or may be even sooner.

Again, thanks a lot for this analysis. 🌹
 
Last edited:
9
•••

karmaco

Top Member
Impact
11,969
The GD tool serves no purpose other than possible keyword strength/hotness and sometimes comparable sales.

It values most names put in between 1 and 2 K. It undervalues valuable names, refuses to value premiums- β€œover 25K” is vague to say the least.

It would be nice if they admitted these are by far flipper sales not end user sales that make them obtain the figures.

The reason only 18.9% sell for below the β€œvalued amount” is most domainers aren’t chumps or at least about 80% πŸ˜… aren’t.

Another disservice of the tool is it over values crap names and unregistered names by design. Their intent is to sell at any price point not to give an actually accurate tool out for free. Its always about GD.
 
5
•••

GFBK.com

Established Member
Impact
13
Thanks for your effort, it's definitely worth for many domain invester to watch and learn
 
4
•••

NameInbox

Established Member
Impact
519
Thanks Bob,
Awesome content, as always.
Awesome Cory Monteith GIF
 
3
•••

ThatNameGuy

Restricted (15-30%)
Impact
1,875
Bob....like you I'm a stats and analysis guy but i tend to use the GD appraisal more as a tool to help sell my domains. With regards to "filtering" I'll also use Estibot, SAW and Nameworth on occasion. That said however, end users can relate to Go Daddy, but I've had enough personal experience with them that I can discredit them as easily as build them up.

Thanks for your analysis Bob.....your credibility is exceptional(y)
 
4
•••
Impact
35,228
Thanks for all of the input. Just thought I would reiterate a few points.
  • This is not meant to be an evaluation of the instrument overall, but rather simply possibly useful information on whether we exclude too many names when we use a cutoff based on it.
  • I think most buying names at auction, or wholesale, are looking for names that will sell retail at least $2000+, and probably in most cases $5000+, therefore it makes sense to look at price ranges I did.
  • Looking only at names that sell $2000+ one would probably expect any tool to overall underrepresent the prices. This is because the tool is trying to valuate a population of names of that type, but looking at names that sold we have cherry picked the subpopulation with sales.
  • Keep in mind this is based just on 189 sales, a few of which did not have a meaningful GD valuation. And only 25 or less at each price point.
  • I did all of this by hand, so thought about each price as I checked out the GV estimate. Often the GV estimate was nearer to what I would think the name would sell at. Some of the sales when for what I considered high prices for the name, just showing that someone who sticks out for high prices will occasionally get them as it only takes one buyer who wants that particular name badly. Of course, many similar names never get sold.
  • Yes, it would make sense to somehow weight the different price points with number of sales in those. If one did that, the instrument would be seen to cut off more names, as more of the sales really coming from the low price tier(s).
  • While I would love to evaluate the tool in the $1000 to $2000 range, Sedo sales in that range are not reported, and there is no other retail venue with a decent number of unbiased sales in that range (yes BuyDomains, but they price most in narrow range so hard to do this sort of study).
  • While at first glance these results may seem in conflict with the Afternic statement re how close it comes to sales value, keep in mind almost certainly most of the sales in that comparison were closer to $2000 and below range. For those, I suspect vast majority are appraised within $1000 of correct value.
  • Also, keep in mind that the GD appraisal was 'trained' or based on, I presume, Afternic and GoDaddy sales. There may be differences, I think are, in Sedo sales. So this is a harder test of the instrument.
  • As others stated above, I share the opinion that the tool tends to make very low worth or worthless names in low $$$$ range, so too high, while rating much more valuable names not much different, low to mid $$$$.
  • The GD valuator does a superb job of splitting words. Amazing really. I've used other valuations and GD usually splits 2-word names better. They also tend to at least in a qualitative sense get right which keywords are more valuable. It can be tricked by nonsense combinations of two words that don't go togther, though.
  • I don't see any sense in evaluating how GD appraisal does on new gTLDs until they modify the algorithm to take into account the TLD. Right now, the same SLD in almost any new gTLD gets appraised about the same which makes no sense.
  • I find GD appraisal helpful in giving me additional comparators and an idea of how much keywords are worth. It is also somewhat helpful as a second opinion. If I simply asked it which of two names is more valuable, and it gave one at $1200 and one at $2500, most of the time I agree the higher appraised one is more valuable, even though disagreeing with the exact valuations.
Thanks for all of the input.

-Bob
 
Last edited:
7
•••

donparadoks

New Member
Impact
8
This is still good point to make some sells up or below.
 
3
•••
Impact
35,228
I thought this was, .... interesting. Afternic offers advice on how to improve your chances of sales. So they today offer me the following advice to reprice 10 of my names. On one of my names, they think I should reprice, because I have it at $595 but their valuation instrument say $591. I think they have much more faith in the precision of the instrument than I do:xf.grin:. The names I blacked out had more significant differences, but still I ignored the advice.

RepriceAdvice.png
 
5
•••

Fayaz Ahmed

DomainRobin.comUpgraded Member
Impact
456
Thanks for all of the input. Just thought I would reiterate a few points.
  • This is not meant to be an evaluation of the instrument overall, but rather simply possibly useful information on whether we exclude too many names when we use a cutoff based on it.
  • I think most buying names at auction, or wholesale, are looking for names that will sell retail at least $2000+, therefore it makes sense to look at price ranges I did.
  • Looking only at names that sell $2000+ one would probably expect any tool to overall underrepresent the prices. This is because the tool is trying to valuate a population of names of that type, but looking at names that sold we have cherry picked the subpopulation with sales.
  • Keep in mind this is based just on 189 sales, a few of which did not have a meaningful GD valuation. And only 25 or less at each price point.
  • I did all of this by hand, so thought about each price as I checked out the GV estimate. Often the GV estimate was nearer to what I would think the name would sell at. Some of the sales when for what I considered high prices for the name, just showing that someone who sticks out for high prices will occasionally get them as it only takes one buyer who wants that particular name badly. Of course, many similar names never get sold.
  • Yes, it would make sense to somehow weight the different price points with number of sales in those. If one did that, the instrument would be seen to cut off more names, as more of the sales really coming from the low price tier(s).
  • While I would love to evaluate the tool in the $1000 to $2000 range, Sedo sales in that range are not reported, and there is no other retail venue with a decent number of unbiased sales in that range (yes BuyDomains, but they price most in narrow range so hard to do this sort of study).
  • While at first glance these results may seem in conflict with the Afternic statement re how close it comes to sales value, keep in mind almost certainly most of the sales in that comparison were closer to $2000 and below range. For those, I suspect vast majority are appraised within $1000 of correct value.
  • Also, keep in mind that the GD appraisal was 'trained' or based on, I presume, Afternic and GoDaddy sales. There may be differences, I think are, in Sedo sales. So this is a harder test of the instrument.
  • As others stated above, I share the opinion that the tool tends to make very low worth or worthless names in low $$$$ range, so too high, while rating much more valuable names not much different, low to mid $$$$.
  • The GD valuator does a superb job of splitting words. Amazing really. I've used other valuations and GD usually splits 2-word names better. They also tend to at least in a qualitative sense get right which keywords are more valuable. It can be tricked by nonsense combinations of two words that don't go togther, though.
  • I don't see any sense in evaluating how GD appraisal does on new gTLDs until they modify the algorithm to take into account the TLD. Right now, the same SLD in almost any new gTLD gets appraised about the same which makes no sense.
  • I find GD appraisal helpful in giving me additional comparators and an idea of how much keywords are worth. It is also somewhat helpful as a second opinion. If I simply asked it which of two names is more valuable, and it gave one at $1200 and one at $2500, most of the time I agree the higher appraised one is more valuable, even though disagreeing with the exact valuations.
Thanks for all of the input.

-Bob

Excellent clarifications with great details!
Thanks a lot 🌹
 
4
•••

Fayaz Ahmed

DomainRobin.comUpgraded Member
Impact
456
On one of my names, they think I should reprice, because I have it at $595 but their valuation instrument say $591
LOL. I had similar experiences as well.

I'd understand if they suggested $593, because it's a prime number, but why $591!
Just kidding :ROFL:
 
4
•••
Impact
35,228
$593, because it's a prime number, but why $591!
Yes, it is funny indeed.

But you suggest something I never thought about, maybe it is time to try prime number pricing! My math side likes the idea! B-)

Here is a list of prime numbers up to 7919 in case someone wants to try prime number pricing. Because of role of prime numbers in blockchains, I bet it would attract at least a little attention.:-D

https://en.wikipedia.org/wiki/List_of_prime_numbers
 
Last edited:
6
•••

twiki

Top Member
Impact
25,037
Thanks. This is useful... Unfortunately, the result just proves that the valuations are mostly useful as very ballpark estimates and, e.g., a name valued at $1200 can be in fact better than the one valued at $2000.

And if you create a valuation tool that weeds off the really bad ones and follows some simple rules and then assigns values from $1250 to $2500 to 95% of those names, then your tool also will be able to get 40-50% of the actual sales within 0.5 to 2 range
As an owner of a more or less similar tool I can agree. It gives you a ballpark.

It can't really give you an exact price, though sometimes it matches.

I had(and still have) a 50% precision - which is good enough to me. But that means 50% of sales more or less in the right zone. About 25% out of that actually being very accurate. But that's about all the precision.

The rest 50% were off, either too big or too high. Sometimes by much.

There's a certain degree of randomness in sales data. Plus very, very, I mean very large gaps between sales.

Data looks more like the universe... that is 99.999% space and just the rest being stars. (Edit: I only have a little amount of data, but even if I had 10x that would not solve the problem, still way too large distance between data points... see the chart Bob made.)

I've been using both math-based algo and AI (machine learning) to get the most accurate appraisal I could. It never went past this. Wondering why? See above line.

AI is as good as the data you feed it. The data, I'm afraid is not enough by a large margin. Note, GD probably still has the biggest data sample there is, internally.

Good differentiator though between something to look at and a piece of cr*p ... in most cases.

Edit: GD appraisals have improved a bit lately, that I can tell as I often compare various prices with that.
 
Last edited:
3
•••

NigelD

Established Member
Impact
246
Thanks Bob for this very useful study. It's timely too as I was scratching my head recently trying to work out just why GoDaddy tool was valuing .uk domains so low. I even watched a video on youtube featuring @Joe Styler and he was asked at 6.10 about comparable sales and using info from namebio and dnjournal and Joe says the tool will bring in own sales data from afternic and public information. Here's the video:

So I took a look at the five highest reported .uk sales on namebio and godaddy still values each domain at a tiny fraction of its actual sold price.

The sales were: $56k, 39k, 21k, 21k, 19k
Go Daddy appraisal tool: $1723, $3700, $1992, $308, $167

I took at look at the top sale of $56k on estibot and yes estibot have valued that domain at $57k so have taken into account the public sales data.

Now we are dan users based in uk with a large .uk portfolio and, for now, we have decided against adding our domains to the godaddy network. Godaddy need to take into account past .uk sales when calculating .uk values. We also need a greater presence in Godaddy UK search results for the .uk domain.
 
6
•••
I feel the appraisal tool might be a bit more useful if it actually updated on a regular basis.
There doesn't appear to have been any updates since Feb 15, 2021, which means all of the appraisals are over 2 years old.
 
6
•••

Edukar

Established Member
Impact
74
Thank you, Bob, the article is wonderful. But I don't use GD evaluator. It's disgusting and has nothing to do with sales. I did the same work as you did a year ago, and I became convinced that this tool cannot be trusted. I like the Alter appraiser. He sins by overcharging the domain more often, but as I understand, his algorithms prescribe the search for similar domains on the market, so the pricing of this tool suits me/ Judge for yourself, I sold Vitaxir.com for $2050 (this is at a discount, the first price was $2200), the Alter appraiser evaluated the domain $2154, the Godaddy appraiser - $1085. I'm sorry for your wasted time, but GD evaluator is not worth the attention
 
Last edited:
6
•••
I personally use GoDaddy Appraisal Value for preliminary domain filtering on a huge domain list and I set the filter value to 1K to 1.2K only. You would unlikely overlook any valuable domains with a relative lower threshold setting.
 
Last edited:
5
•••
It's a useful value marker because generally the higher the appraisal the better the domain. But I would never price my domains retail based solely on Godaddy estimate.
 
5
•••

NigelD

Established Member
Impact
246
maybe its geared towards the .com market - in my example above a two letter, good letters, .uk domain sold for $19k and that sale is listed on namebio. Godaddy tool says $167. You cannot buy any 2 letter .uk domains for that price. Someone needs to amend the algorithm or simply remove valuations for .uk domains as this tool is not helping domain investors in the UK.
 
4
•••
I sold a dot com domain last year for $51,888. GoDaddy appraisal tool value - $1,686.
 
Last edited:
6
•••

Edukar

Established Member
Impact
74
I sold a dot com domain last year for $51,888. GoDaddy appraisal tool value - $1,686.
Can you name this domain? If this is a secret, I would appreciate it if you use Alter.com appraiser and specify the price here
 
1
•••

WillW

Upgraded Member
Impact
1,167
I've sold names that have low GD appraisal value.
I haven't sold and may never sell some names that have a high GD appraisal value.
I use it as a basic guide when I look at expired domains or auctions. That's about it.
As far as a pricing guide, it's a non factor and I wouldn't recommend anyone invest or price names based on appraisal values alone.
 
3
•••

ReallyBigIdea.com

Top Member
Impact
1,644
Govalue sucks. Absolutely useless .
 
4
•••

Joe Styler

Aftermarket Product ManagerTop Member
βœ” GoDaddy Staff
βœ” Afternic Staff
Impact
4,785
Thanks Bob for this very useful study. It's timely too as I was scratching my head recently trying to work out just why GoDaddy tool was valuing .uk domains so low. I even watched a video on youtube featuring @Joe Styler and he was asked at 6.10 about comparable sales and using info from namebio and dnjournal and Joe says the tool will bring in own sales data from afternic and public information. Here's the video:

So I took a look at the five highest reported .uk sales on namebio and godaddy still values each domain at a tiny fraction of its actual sold price.

The sales were: $56k, 39k, 21k, 21k, 19k
Go Daddy appraisal tool: $1723, $3700, $1992, $308, $167

I took at look at the top sale of $56k on estibot and yes estibot have valued that domain at $57k so have taken into account the public sales data.

Now we are dan users based in uk with a large .uk portfolio and, for now, we have decided against adding our domains to the godaddy network. Godaddy need to take into account past .uk sales when calculating .uk values. We also need a greater presence in Godaddy UK search results for the .uk domain.
That video is old. I don't even own those clothes in the video anymore :) The tool is due for an update of data soon so if you watch it will show updated data (more recent sales loaded in) in the next couple months.
How is having the appraisal impacting your choice to not list domains on GoDaddy?
 
2
•••