Spaceship

news Howard Fellman Releases Modern AI Powered Domain Name Appraisal System

Spaceship Spaceship
Most of the time, domain name automated appraisal systems don’t receive much positive commentary from investors. When a newly released AI powered appraisal service was praised by a number of respected voices in the domain community, I decided to take a deeper look.

I gave the instrument a try with 134 names, including various types of names and domain extensions. I also reached out to developer, and long-time domain investor, Howard Fellman, to learn a bit more about the instrument. While it has impressive capabilities, the comparator sales section in the current release contains many sales that are not validated at the sources given.

The Release

On April 21, 2025 Howard Fellman posted the following on his LinkedIn account:
My latest personal AI project is now in beta and ready for your stress testing and constructive criticism. It consists of a no-registration domain name appraisal tool utilizing a huge database of resale comps. It works with all extensions and is, for now, completely free. You simply provide a domain name and it will thoroughly evaluate it — providing suggestions for best use along with an accurate appraisal value. Thanks, in advance, for letting me know what you think.
The appraisal instrument is at OceanfrontDomains.com.

Guide

Using the instrument is straightforward. Here are the steps:
  1. Go to OceanfrontDomains.com. At the time of writing there is no registration requirement or cost.
  2. Enter the domain name you wish evaluated. Press Send button.
  3. It will start typing out the assessment almost immediately.
  4. If you wish to make a copy of the appraisal, go to near the top, just to the right of the top of the typed review, not the line with the domain name, and a pages icon will appear, press that for copy, then paste the results into a document to save.
  5. To clear the display before entering a new name, scroll to the bottom of results and press the Clear button at the bottom right.
  6. Repeat steps 2-4 for additional names.

Presentation

While there will be slight differences according to the type of domain name, most will include the following elements:
  • Brandability and Meaning
  • TLD Evaluation While the information here will be familiar to domain investors, it provides helpful background on the extension for those not in the domain field.
  • Line of Commerce and Marketing Positioning
  • Keyword Analysis
  • Additional Considerations e.g. Acronym, Homonym, Dual Meaning
  • Comparable Domain Sales This is a very useful section, but the instrument tends to list high-value sales, as opposed to closest comparators, in some cases. The instrument sometimes lists values that a name would have commanded, as opposed to actual sales, so you have to read carefully what is said with each comparator. The AI also sometimes list sales that did not, it appears, occur. More on this later.
  • Business Viability and Competition
  • Highest and Best Use This is an important consideration in domain names – what use is the best fit and would command the best price?
  • Highest Retail Valuation The instrument normally gives a range, and for some names multiple ranges with accompanying notes. Keep in mind that this is the highest retail valuation that would apply to a buyer seeking this exact name.
Note that the exact segments depend on the name and run, and this is an instrument in beta so there were changes over the two days I used the instrument.

My Experience

Before writing this article,I checked 134 domain names with the instrument, for each reading the full report, in at least a cursory manner, and looking at the price range(s) suggested. The domains I tested include legacy extensions (.com, .org, .net), popular generalized country code (.io, .co, .ai, .vc), along with a number of new extensions and a few other country codes. In .com, I included single-word, two word, merged and created brandables, and a few three- and four-word names. The tested names were all from my own portfolio, mainly hand registrations, closeouts, or low-value auction acquisitions, and a few purchased from other domainers. In most cases I had available automated estimates from other instruments for comparison purposes.

Here are my observations:
  • The linguistic aspects were excellent, faithfully identifying the nature and roots for each name. No names were split incorrectly, or misinterpreted.
  • The instrument did a superb job identifying how a name could be used. Most of the ways I had considered were replicated, and for some names it suggested realistic possibilities I had not considered.
  • The dollar valuations were very high compared to other automated appraisals, my asking prices, or pricing suggested by curated marketplaces where some of these domains were listed. In almost all cases this instrument had a higher value than Estibot, Graen, GoDaddy, NameWorth, BrandPa, or Dynadot appraisal instruments. Note only .com had BrandPa or NameWorth valuations.
  • It is important to realize that this instrument gives pricing based on a best case scenario of an end user attracted to this particular name. While some investors prefer to price on that basis, it is important to realize that waiting times will be longer than if the name was priced more competitively.
  • The average of the top of the range given for each domain name was $67,500. 30 of the 134 names appraised had values of $100,000 or more, and only 3 names in my trial had an appraisal less than $10,000 (all three were new extensions), and none under $5000.
  • For a few names, by no means all, it suggested wholesale prices, and these seemed completely unrealistically high. My latter trials never seemed to show wholesale, so perhaps that is now dropped.
  • I could not see evidence that the prices were influenced by how the names were currently listed.
  • In some cases the instrument prices highly even rather long names. For example, it priced my three-word name HealthSciencesCenter.com at $300,000.
  • I tried the instrument with some three and four-word phrase names, and it priced them strongly, and correctly interpreted the significance of the phrase.
  • If a name can legitimately be used in multiple ways, the instrument sometimes provides pricing ranges for each type of end user. For example, I have the name RoverWork.com and it suggested pricing if used in a pet sense, or as robotic rover rental service, or as a site for freelance work.
  • Unlike most appraisal instruments, this instrument will appraise third level names like example.it.com. It does give a caveat about third level domain names depending on the stability of the operator of the second level name. HumbleWorth is the only other appraisal instrument that handles these, to my knowledge.
  • While at first glance the section giving comparator prices was very impressive, with many sales stated that are not in NameBio. However, as various people have noted, it is impossible to validate the majority of the comparator sales, and some seem clearly wrong. More on this topic in the next section.
  • The instrument misses some very relevant comparators. For example, a term that I sold in one new extension, with the sale listed on NameBio, I tried an appraisal for the same term in another extension. It did not show the 2024 sale. I then tried an appraisal on a name that had a previously recorded 4-figure sale on NameBio and GoDaddy Appraisal, but this instrument did not show me that the name had sold previously.
  • When comparator pricing is beyond the recent past, the instrument provides an indication of what a price at the current time would likely be.
  • The instrument provided sound advice on extensions and name characteristics, and why the name has value.
  • At times it was generous in saying a name passed the radio test with flying colors, when I would have said the radio test compliance was questionable.
  • I found the “Best Use” section very helpful.
  • Overall the instrument is impressive, interpreting names correctly, writing the case for the name eloquently, setting out potential uses in a comprehensive fashion, and often coming up with realistic uses that the seller may not have considered. The two concerns are some comparator data are almost certainly incorrect, and overall the evaluation estimates, even keeping in mind that this is for the best case scenario, are probably excessively high.

NamePros Discussion

If you go to the NamePros discussion started by @equity78 on the instrument, Rick Schwartz Gives Positive Remarks on New Domain Appraisal Tool, there is a mix of positive and negative comments.

The negative comments mainly have to do with two points mentioned above: inaccurate comparator sales data, and unrealistically high price evaluations.

Some of the praise for the instrument is because of a feeling that finally there is an instrument that properly values domain names, and makes a strong case for the valuation in a detailed written document.

Under The Hood

I asked Howard Fellman and what powered the instrument, and he wrote:
It utilizes OpenAI’s very latest (and “smartest”) release, Version 4.1 which was just recently introduced on 4/4/25.

AI Hallucinations and Confabulations

It became clear early on that generative AI doesn’t just make the occasional trivial error. At times AI simply makes things up! When asked to write academic-style documents with references, occasionally generative AI will make up citations – they look completely reasonable, with actual journal names, but the papers don’t exist!

While most call these hallucinations, I think in the language of psychology they would more accurately be confabulations. That is the term that Turing Prize winner and Nobel Laureate Geoffrey Hinton, who has been called the Godfather of Artificial Intelligence, prefers. Read more about AI hallucinations in this IBM document, or about the debate on whether it should be called hallucinations or confabulations in this discussion at OpenAI..

Whatever we call them, isn’t made up ‘data’ something to be worried about? Here is Geoffrey Hinton’s way of looking at confabulations:
People always confabulate. Half-truths and misremembered details are hallmarks of human conversation: Confabulation is a signature of human memory. These models are doing something just like people.
This prompted me to think of the following analogy. Perhaps we should think of the output of this AI domain instruments as somewhat like an oral conversation with an experienced and successful domain investor. Someone who had witnessed many sales over the years, and knew the industry well. But they are operating without notes, and can be prone to the occasional exaggeration. They will get some details wrong, maybe even recall a sale that did not happen, confuse two names, or exaggerate, but overall their opinion is well worth hearing. That could be a view to this instrument. For any details critical to you, make sure to independently verify that information. That particularly applies to the comparator sales information that is error prone – many sales listed as NameBio are not in fact on NameBio. Also the appraisal value suggested should be viewed about the same as you would one opinion from a friend who knew the business. Different friends would suggest different values.

Update: In recent days it seems that the tool no longer lists NameBio as the source of comparator sales.

I should note that the following warning is given at the end of each appraisal.
This automated preliminary evaluation has been generated entirely by AI and is designed to demonstrate educational concepts highlighting the similarities between digital real estate and conventional property. The report is not intended to act as an official appraisal nor should it be interpreted as legal, financial, or investment advice. AI can make errors. It is advised to independently verify all information, consult with qualified human experts, and conduct thorough research before making any decision, or taking any action, based on this report.

Meet The Developer

So who is Howard Fellman, the developer of this domain name appraisal instrument? He has been investing in domain names for decades from a base in south Florida. Not along ago he attended the 2024 South Florida Domainers Meetup – catch the pictures from the event in this DN Journal issue.

With two brothers, 36 years ago Howard started PC Professor Technical Institute that is still going strong offering online and in-person training in Microsoft and other IT skills training. You can check it out at PCProfessor.edu.

Howard Fellman has done many interesting things in life, including being in Google’s first ever TV commercial! Yes, really! You can view it at this YouTube link.

Advice Re Using This AI Instrument

It is still early days, and the instrument is in beta and still changing, but this is how I see optimally using this impressive new entry in the automated appraisal field:
  • It provides a superb linguistic interpretation of a name, particularly helpful for those who do not have English as their main language.
  • The potential use suggestions are very well done. I think consulting this instrument will be a second ‘mind’ to confirm what you see in the name, and provide additional use possibilities.
  • It is a one-stop place to get multiple components of the name, all nicely presented.
  • Maybe this is just me personally, but I found it helped me identify and focus on the best use possibilities for a name.
  • As such, it makes the case for almost any name very effectively. I can see it as being helpful as part of the presentation to someone making an offer on one of your names. However, keep in mind that each run may produce somewhat different results. It is not like the algorithm in an appraisal like Graen that produces the same result each time, except for minor changes as registered domains change.
  • I would caution you to not get carried away by the high valuations, and to keep in mind the possible confabulations.
I plan to keep using the instrument personally, but won’t be pricing my names according to it. However, it has already made me rethink my pricing on a few names.

A Selection of Comments

Here are a selection of some of the positive comments about the instrument.

Myke O’Neill of Thrive Domains wrote on LinkedIn:
I am lost for words. This is the tool domainers have been looking for since domaining became a serious business.

One of the pioneers of the domain industry, Dante DeAngelis commented on LinkedIn
Impressive! One of the best tools I have tried.

Aron Meystedt left positive comments, and pointed out that he had tested it with symbolics.com and it went into great detail of the value of the first in any asset.

Rick Schwartz, The Domain King, wrote:
Bravo! Bravo! Now that is an appraisal tool worth its weight in gold. Great job, Howard!

Please share your own views and experiences below. To access the appraisal instrument, go to OceanfrontDomains.com.

Updates:
Apr. 30, 2025 I changed wording in several places to reflect the serious nature of the errors in the comparator sales section of the reports.
May 7, 2025 This is a beta tool, and it appears that in recent days they no longer list some comparator sales as NameBio. I have added a note on this.
 
Last edited:
48
•••
The views expressed on this page by users and staff are their own, not those of NamePros.
I'm struggling to understand the people who are giving this tool such positive comments. To be blunt, as Bob identified, this tool fabricates sales data - and then offers domainers appraisals on the made up data! What is the point of an appraisal tool that does this?

 
25
•••
17
•••
Yeah yeah, whatever :yawn:

Confabulations though.. Great word (y)
 
5
•••
IMG_20250425_203737.gif
 
0
•••
Congratulations to the developer and thank you Pro. Bob for sharing yet another insightful article 👏 👍 🙌

Indeed this tool is much much better and domain investor friendly compare to other available appraisal tools from other registrars as it provide reasons, use cases and some kind of rationalities.

Suggestions: I think they should add two important parameters in their tool:

1. Number of domains registered in other extensions just like dotDB and
2. Number of site developed/ operational

Consider these two important aspects for any domain names and integrate it appropriately in their final appraisal.

Thanks and best wishes
 
3
•••
Like you say Bob, it's great for 'use case' discovery. It does provide insights that might not be known to the domain owner.

That said there's a decided bias towards exact match .com domains. Brandables, not so much.

stlds, ntlds, cctlds it automatically depresses the max possible price based on the top sale in that tld, which could have been 10 years ago. .ai sales from even 2020 can't even be used for comps today.

That said, it's an improvement over all other appraisal tools I've tried - godaddy / estibot / etc, their valuations seem more like wholesale pricing, at best.

Though it's a step in the right direction, it could be improved by adding on the namebio / dnjournal sales data to the 'memory' of the wrapper.
 
2
•••
Yea! Another appraisal so we can have a better average of all of them. Ty Mr. Bob for the article. :)
 
3
•••
All domain appraisals, by AI or human, are flawed in principle (imo), so I don't really care about numerical values (the exception being GD's GoValue, because it's harmful to the industry). This tool would be better without value estimations, especially because those prevent some people from appreciating all the good stuff.

I mainly use AI to help me brainstorm, and hallucinating AI is much better at ideation tasks than "veridical" people around me. Fellman's AI wrapper produces interesting ideas and insights in one shot without additional prompts. Of course I can and will directly use the constantly evolving real things and my own prompts, but so what? This (evolving?) tool is addictively convenient.
 
Last edited:
5
•••
Suggestions: I think they should add two important parameters in their tool:

1. Number of domains registered in other extensions just like dotDB and
2. Number of site developed/ operational

Yes true that info not seeming to be taken into consideration by the instrument.

I will continue to use dotDB, OpenCorporates, the Domain Academy tools, and NameBio as part of research, along with TM queries, for any domain names under consideration. This new appraisal tool is certainly not an all encompassing view of relevant parameters.

Thank you for your suggestion to the developer.

-Bob
 
3
•••
I mainly use AI to help me brainstorm, and hallucinating AI is much better at ideation tasks than "veridical" people around me. Fellman's AI wrapper produces interesting ideas and insights in one shot without additional prompts. Of course I can and will directly use the constantly evolving real things and my own prompts, but so what? This (evolving?) tool is addictively convenient.
I think you have captured perfectly, and eloquently!, the right way to use this new instrument. Thank you.

I like a lot that, as you say, in one shot it analyzes the name, and makes a good pitch for the strengths of any particular name, also including negative connotations or possible confusions where appropriate.

-Bob
 
4
•••
I hope that people will look beyond the aspirational pricing and comparator sale confabulations to see the ways this instrument can be used.

I continue to test names today, and the beta nature is obvious, as they keep changing the structure of the output, so the headings listed in the article may not be what you encounter.

I've blocked out the name to not promote one of my listings, but it just wrote this for one of the names I submitted. I could spend an hour, and could not have more eloquently made the case for the name. Truly great writing and comprehensive and accurate coverage of potential use.

CB-descr.png


-Bob
 
Last edited:
2
•••
@Bob Hawkes If you're looking for a tool that generates exceptional lander descriptions, I recommend building a "Gem" in Google Gemini. With the right prompt, this tool consistently produces top-notch lander descriptions, while keeping you fully in control. I rely on it for all my Atom landers, and simply paste in each domain name. The dedicated "Gem" instantly crafts a tailored, five-paragraph description that reflects the strategic value, use cases, current market developments, and culminates in a compelling call-to-action.
 
Last edited:
2
•••
@Bob Hawkes If you're looking for a tool that generates exceptional lander descriptions, I recommend building a "Gem" in Google Gemini. With the right prompt, this tool consistently produces top-notch lander descriptions, while keeping you fully in control. I rely on it for all my Atom landers, and simply paste in each domain name. The dedicated "Gem" instantly crafts a tailored, five-paragraph description that reflects the strategic value, use cases, current market developments, and culminates in a compelling call-to-action.

This is interesting, I'm going to give this a whirl. I keep having to "retrain" claude and ChatGPT domaining. If Gemini allows me to save it so the Ai can reference our conversations and learnings over time - that would be amazingly helpful
 
0
•••
@Bob Hawkes If you're looking for a tool that generates exceptional lander descriptions, I recommend building a "Gem" in Google Gemini. With the right prompt, this tool consistently produces top-notch lander descriptions, while keeping you fully in control. I rely on it for all my Atom landers, and simply paste in each domain name. The dedicated "Gem" instantly crafts a tailored, five-paragraph description that reflects the strategic value, use cases, current market developments, and culminates in a compelling call-to-action.

Thanks for the suggestion, I must give it a try.

After putting a lot of effort into Atom descriptions, I have come to the conclusion that overall I am better served by the minimalist Atom landers, so while there are still descriptions that show on my white label site, most potential buyers never see them.

But for other presentation of names, a great description can sometimes be important.

Thanks again,

Bob

PS Do you use Atom Logo generation, or create them with something else and then upload?
 
2
•••
PS Do you use Atom Logo generation, or create them with something else and then upload?
Both. With my Ideogram Plus subscription, I can produce stunning images that go far beyond a mere logo, transforming Standard and Plus landers into visually striking showcases. However, when a Standard/Plus listing transitions to Premium, these images are not always accepted because Atom insists on maintaining a level playing field by preventing overly polished visuals (for Premium domains). In the meantime, Atom’s AI logo generation tool has improved significantly over the past few months, with the cost per generation dropping to just 10 cents. Although it still requires a few retries, it generates a sufficiently good presentation much more quickly than the Ideogram workflow, which involves prompt fine-tuning, resizing, and uploading in both 750x450 and 600x600.

For my 3,500 Atom Plus listings, most of them (about 2,500) have these stunning ideogram visuals.
 
Last edited:
4
•••
After putting a lot of effort into Atom descriptions, I have come to the conclusion that overall I am better served by the minimalist Atom landers, so while there are still descriptions that show on my white label site, most potential buyers never see them.

But for other presentation of names, a great description can sometimes be important.
I believe it's essential to showcase every facet of a domain in the description. This is easier for domains with an immediately clear focus than for those that initially seem like an empty vessel, lacking inherent meaning.

A well-crafted landing page description not only encourages users to linger, resulting in measurable positive engagement, but also plays a crucial role in search engine optimization through compelling descriptions and keywords. Moreover, high-quality content can enhance Atom's own AI training process, leading to improved discoverability of your domains within the marketplace.
 
5
•••
It flatters domain investors by telling them their junk is valuable. Like a mirror that lies and tells you how amazing you look all the time.

That's why people like this one.

Like most appraisal tools, it's not worth much.
 
6
•••
2
•••
0
•••
  • The sidebar remains visible by scrolling at a speed relative to the page’s height.
Back