I’ve been writing about domain names for a long time, and valuation has always been one of the industry’s most debated subjects. Put the same domain in front of five experienced investors and you can easily get five very different prices. Automated appraisal tools were supposed to make that easier, but in many cases they simply created another number for investors to argue about.
Youness Kasmi has taken a slightly different approach with DNRater.com. Rather than trying to tell an investor what a domain is theoretically “worth,” much of the focus is on answering a more practical question: What should the Buy It Now price actually be? DNRater has since expanded into other areas of domain research and discovery as well. I wanted to talk with Youness about how he approaches domain valuation, what AI can and cannot do, what he has learned from years of investing himself, and where he thinks domain tools are headed next.
Sully: Youness, you’ve been involved with domain investing for quite a while. How did you originally get started with domain names?
Youness: Thank you Mike for reaching out. It is a pleasure to have this interview with you.
It actually started when I was in engineering school. I was studying civil engineering and needed to work online to help pay for my school fees. In 2012, I launched A2S Forum, an MMO forum where I was writing ebooks about whatever I was learning at the time, whether it was building websites, SEO, making money online, and so on. Members could also request topics for new ebooks.
Around 2014, someone asked me to write an ebook about domain flipping. At that time, I was mostly buying domains because I wanted to build websites on them, so I started researching how domain flipping worked. I found a way to approach it and wrote the ebook. People really liked it, and that was basically my introduction to domain investing.
A few years later, around 2017, I started making videos on YouTube. That brought me a completely different community, especially from the MENA region. Some of those people eventually discovered my older ebooks about domain investing, and they started asking me questions.
Between 2018 and 2021, I spent a lot of time trying to teach domain investing through videos.
The problem was that many people were completely new to the industry. They would watch me explain how to find domains at auctions or how to contact owners directly about domains that were sitting unused, and then they would try to reproduce the same methods.
But there was one big problem. They often didn’t know what a domain was actually worth in the market. Some would spend $1,000, $5,000 or even more on a domain and then get stuck with it for years.
I felt partly responsible because my content was what brought many of them into the industry.
At that time, around 2019, I was doing the valuation work manually. For every domain, I would research which companies could potentially be interested in it, what those companies were doing, how much they could afford, whether they had invested in domains before, and then combine that with my experience from previous negotiations.
The problem was that this could take almost a week for one domain.
My brother is a data engineer, and he asked me a simple question: why don’t you try using AI to do what you are doing manually?
This was 2019, so there was obviously no ChatGPT. AI was very different from what people know today. I started researching how I could train a model to reproduce the research process I was doing manually.
That eventually became the foundation of DNRater.
When the model was ready, I tested it against some of my own domain sales. I also asked friends including Hiren Patel @QualityNames and Muhammad Aamir Saddiqui @DotCorner to test it against private domain sales that weren’t publicly known. Their feedback was extremely useful for validating what we had built.
So DNRater wasn’t really born from the idea of “let’s build another domain appraisal tool.” It came from a problem I personally had, and more importantly, a problem I saw happening to people I had taught.
That is also why the product has always been much more focused on helping someone decide how much to pay rather than pretending that a domain has one objectively correct value.
Sully: What frustrated you about the existing domain appraisal tools enough that you decided to build DNRater?
Youness: The biggest problem for me was that many appraisal tools were answering a question that wasn’t necessarily the question an investor was asking.
They were trying to answer, “What is this domain worth?”
But when you are an investor looking at an auction or negotiating with a domain owner, the question is usually much more practical: “How much should I actually pay for this domain?”
There is a big difference between those two questions.
I also didn’t like the idea of relying too heavily on historical domain sales as the main source of truth. A domain that sold for $20,000 in 2017 doesn’t automatically tell you what a similar domain should be worth today. The market changes, industries change, companies change, and the amount companies are willing to spend on domains changes.
I wanted to build something that looked at the current market and the potential buyers behind a domain rather than simply finding a few historical sales and calculating another number.
That is what pushed me toward building DNRater.
Sully: When you are trying to determine a BIN price, what are some of the factors that matter most?
Youness: For me, the most important thing is understanding who could actually buy the domain.
A domain can look amazing to another domain investor, but that doesn’t necessarily mean there is a company willing to pay a lot of money for it.
We look at things such as the meaning of the domain, the keywords, the industries that could use it, companies operating in those industries, the size and financial position of potential buyers, their existing branding and naming patterns, and whether the domain fits something they could realistically use.
There are also linguistic and commercial factors. Is it easy to understand? Is it easy to pronounce? Does it have commercial intent? Does it work for a product, company, service or category?
And then there is the market itself.
The goal isn’t to find a magical formula where you put in a domain and get a perfect price. The goal is to collect enough relevant signals to make a much more informed buying decision.
That is also why I use the term BIN price. I’m not trying to tell someone, “This domain is objectively worth $17,438.” I’m trying to answer, “If you want to buy this domain today, what would be a reasonable BIN price to start from?”
Sully: Domains are unique assets, which makes valuation difficult. How do you build a model around something where there may never have been a truly comparable sale?
Youness: That is exactly why I didn’t want to build the model around comparable sales alone.
With many domains, there simply isn’t a perfect comparable. You might have a two-word domain where nobody has ever sold that exact combination, but there could still be hundreds of companies that could potentially use it.
So instead of asking only, “What similar domain sold?” we can ask, “Who would want this domain today?”
That means looking at companies, industries, keywords, business models, financial information, existing domain usage and other real-time signals.
I think this is especially important because the value of a domain isn’t completely independent from the market around it.
For example, a keyword connected to an industry that suddenly becomes very important can have a completely different commercial environment from the same keyword several years earlier.
This is one reason I wanted DNRater to work with fresh information rather than simply training everything around historical domain sales.
Historical sales are useful. I don’t think they are useless. But I don’t think they should be treated as the entire market either.
Sully: You’ve recently criticized AI models that return very broad valuation ranges such as $8,000 to $32,000. What do you think those models fundamentally misunderstand about domain pricing?
Youness: The problem with a range like $8,000 to $32,000 is that technically it may sound intelligent, but as an investor, what are you supposed to do with that information?
If I’m bidding on a domain at auction and the current price is $2,000, does that mean I should bid $3,000? $8,000? $15,000? Or $30,000?
The range doesn’t actually help me make the decision.
I think some AI models are trying to protect themselves from being wrong by giving a huge range. But for an investor, uncertainty is exactly the problem they are trying to reduce.
Of course, no model can know the exact price a buyer will eventually pay. A domain is ultimately worth whatever a buyer is willing to pay.
But there is still a difference between acknowledging uncertainty and giving such a wide range that the result becomes almost unusable.
That’s why we focus on the BIN price. It gives the investor a practical number that they can use as a starting point.
I always tell people not to treat it as an absolute truth. Use it as information. If you agree with it, use the number. If you have additional information about the domain, adjust your decision accordingly.
For my own audience, I have generally suggested looking at the acquisition price as a relatively small percentage of the estimated market value, often around 5% to 8%. The exact percentage depends on the domain, as well as the buyer’s financial situation and cash flow. The main idea is to leave enough room for the uncertainty involved in selling.
Sully: What types of domains are the hardest to value accurately?
Youness: Brandable domains are probably the hardest.
With a keyword domain, you can often identify the meaning, commercial intent, industries, potential companies and other measurable signals.
But with a brandable domain, a lot comes down to human psychology.
Someone can look at a name and immediately imagine a beautiful company, product and brand around it. Another person can look at exactly the same name and feel nothing.
The value can depend on the visual identity, the logo, the landing page, the copywriting, the product and even the personality of the person making the buying decision.
That is something we are still trying to understand better.
Actually, this is connected to another project I started called DNPoll. The idea was to understand the psychology of buyers by asking them simple questions and learning what types of names they naturally prefer.
I started working on DNPoll around 2019 but never fully launched it because we were still struggling with the psychology side of the problem.
Humans are much harder to model than domains.
Sully: DNRater now includes tools for discovering recently expired domains. What have you learned from analyzing the names investors allow to expire?
Youness: One interesting thing is that an expired domain isn’t necessarily a bad domain.
There are many reasons a domain gets dropped. The owner might have stopped a business, forgotten about the renewal, changed their brand, lost interest in the project, or simply decided that the renewal cost wasn’t worth it.
So I don’t look at an expired domain and automatically think, “The previous owner didn’t want it, therefore it has no value.”
At the same time, the expired market can teach you something about what investors are actually willing to hold.
When you analyze large numbers of expired domains, you start seeing patterns. Some names continue to have commercial potential even after being dropped, while others may look attractive at first but have very little real buyer demand.
For me, that is one of the interesting parts of analyzing the expired market. You’re not just finding domains. You’re also observing what the market is willing to keep, what it is willing to abandon, and where there may be gaps between perceived value and actual demand.
Sully: You also built Corporate Domain Radar to track registrations by major companies. What could domain investors learn from watching the naming behavior of companies like Apple, Amazon, Microsoft and Disney?
Youness: Corporate Domain Radar was actually a separate project I built in the past, and we don’t offer it anymore, but it came from an idea I found very interesting at the time.
I wanted to look at the domain market from the other side. Domain investors spend a lot of time watching what other investors are buying, what is selling, and what is expiring. But I was interested in watching what actual companies were registering.
Large companies don’t necessarily register domains because they are about to launch something. They might register a name defensively, protect a brand, prepare for a future project, or simply secure variations of a name.
So one registration by itself doesn’t tell you very much.
What becomes interesting is the pattern over time. You can see how a company names products, what words it repeatedly uses, what types of names it protects, and sometimes which naming trends are becoming important to that company.
I think that gives domain investors another source of market intelligence. Instead of only looking at what domain investors think is valuable, you can also observe what real businesses are actually registering and protecting.
Although we no longer offer Corporate Domain Radar, the idea behind it is still connected to how I think about domains: I am more interested in understanding real demand and real buyers than simply looking at domains in isolation.
Sully: What do you think domain marketplaces could do better to help investors sell inventory?
Youness: This is actually something I started thinking about back in 2019 when I built DNPoll.
One problem I saw with domain marketplaces, especially brandable marketplaces, was that buyers could search for a business idea and get thousands of names. Having more inventory sounds great, but at some point too much choice becomes a problem. If you are starting a company and you have 5,000 possible names in front of you, you can spend hours comparing them and still not know which one you actually want.
So with DNPoll, I wanted to completely change the way the buyer discovers a domain.
Instead of making someone search through thousands of domains, we show them two names at a time. They simply choose the one they prefer. It can be based on anything they naturally notice: the domain name, the logo, the symbol, the colors, the TLD extension, or simply the feeling they get from the brand.
The interesting part is what happens after each choice.
If you choose the domain on the right, we keep that domain and replace the one on the left with a new one. Then you choose again. You might choose the right one again, then the left one, then the right one. After around ten choices, we have a small picture of what you naturally prefer.
But the real idea was not only to help one buyer find a domain.
The model can learn from every buyer who plays the game. If thousands of people make these choices, we start collecting data about what different types of buyers actually like. We can learn whether certain buyers respond more to the domain name itself, the logo, the visual identity, the TLD, or other characteristics.
Then when a new buyer comes in, if their choices start looking similar to patterns we have already seen, the system can potentially predict which domains they are likely to prefer much earlier in the process.
So instead of making the buyer look through thousands of domains, we are trying to bring the right domains to them.
At the same time, it gives sellers something they don’t normally get from a marketplace: a way for their inventory to be matched with buyers based on actual behavior and preferences, rather than simply appearing in a huge search result.
That’s what I liked about the idea. It could potentially be a win-win. The buyer gets to a name they like much faster, while the seller has a better chance of getting the right domain in front of the right buyer.
I had to pause DNPoll when the valuation problem became a bigger priority for my audience and I decided to put my own resources into DNRater. But I still think the basic problem it was trying to solve is very real: domain marketplaces have become very good at giving buyers more choices, but I think the next step is helping them make the right choice faster.
Sully: You’ve spent years educating people about domains, particularly in the MENA region. What are you seeing from newer domain investors today that is different from when you entered the business?
Youness: The biggest difference is access to information.
When I started, if you wanted to learn domain investing, you had to spend a lot of time on forums, reading old discussions, following sales, learning from other investors and making your own mistakes.
Today, someone can discover domain investing through YouTube, X, Discord, Telegram or other communities and immediately have access to tools, marketplaces, historical data and now AI.
That is a huge change.
But the easier access to information also creates a new problem. People can move very quickly without necessarily understanding why something works.
I’ve seen newer investors become very interested in whatever is currently getting attention, whether that’s a particular extension, a certain keyword, AI-related domains or another trend.
The fundamentals haven’t really changed, though.
You still need to understand demand, understand the buyer, control your acquisition price and be prepared for the fact that selling a domain can take time.
For the MENA community specifically, I’m happy that domain investing has become much more accessible. When I started teaching it, many people around me didn’t even know that domain names could be treated as an investment.
Now I see people entering the industry much earlier and with access to tools that we simply didn’t have when I started.
The challenge is making sure that having more tools doesn’t replace learning how the market actually works.
You can try DNRater and learn more about Youness’s work at DNRater.com.