Most people think of domain names as something you register when you start a business. You go to a registrar, type in your company name, hope it’s available, pay your ten dollars a year, and move on. That’s about as much thought as most people give it.
But domain names are more interesting than that, and the years I spent immersed in them as CEO of Dark Blue Sea gave me something I didn’t realise I was learning at the time: a first-principles education in how the commercial internet actually works. That education is foundational to how I evaluate angel investments today.
Let me start with the basics, because they matter.
First principles: what is a domain name?
A domain name is the real estate of the internet. Or more precisely, the vacant land. It’s the foundation for both a website and an email address. Before you build anything online, you need one. And like real estate, not all domains are created equal.
A good domain is memorable. It’s easy to remember, easy to type, and it passes what I call the “radio test”. If someone hears it once, they can recall it and type it in without seeing it written down. Businesses value good domains because they sit at the centre of brand and marketing. Your domain is your address, your shopfront, and your calling card, all in one.
Here’s the part that surprises people: dot com and many other extensions are deregulated. Anyone can register an available domain on a first-come-first-served basis for about ten dollars a year. So long as you don’t use it in breach of someone’s trademark or copyright, you have clear title. And with clear title and natural scarcity (there’s only one flowers.com) domain names trade in a secondary market much like property. Good domains can be worth five, six, or even seven figures.
Now, to understand why domain names became a serious business, you need to understand a phenomenon from the early internet that has largely disappeared: direct navigation.
When internet real estate paid rent
In the late 1990s and early 2000s, search engines were nearly useless. They were full of spam listings, and this was before Google cleaned things up. So users developed a perfectly rational strategy: instead of searching for “flowers,” they’d simply type “flowers.com” directly into the address bar. The logic was sound; it was highly likely that flowers.com would be a website about flowers, and because it was a premium name, the website was probably a good one.
This was called direct navigation traffic, and it was extraordinarily valuable. It was intent-based, just like search traffic, as the user was actively looking for something. And if you owned flowers.com, you didn’t even need to build a real business behind it. You could place advertising on the page such as banner ads, pay-per-click links, or affiliate links and generate thousands of dollars a month with minimal effort. It was the digital equivalent of rental income: you owned the asset, and advertisers paid you yield.
This was the world that Dark Blue Sea (DBS) operated in. DBS was a domain name registrar that also ran an online advertising business, both PPC and an affiliate network. We served professional domain owners: people who held portfolios of hundreds or thousands of domain names. On a single platform, they could register their domains, monetise them through advertising, and sell them in the secondary market. It was a vertically integrated platform for the domain name asset class.
When I first encountered the DBS business around 2000, what fascinated me wasn’t the domains themselves, it was the PPC advertising model. This was Google’s killer app before Google even had a revenue model. At the time, the market leader in PPC was a company called GoTo, which later became Overture and was then acquired by Yahoo.
GoTo published something called a search selection tool, which showed how many times a given phrase had been searched in the prior month. Separately, you could see the PPC bid prices through another interface. As a mathematician and natural pattern matcher, I saw an opportunity. I created what I called a “commerciality metric”. Simply, search volume multiplied by bid price.
The potential behind the clicks
The logic connecting this to DBS’s business was straightforward. Direct navigation traffic was loosely correlated to search volume: popular search phrases tended to generate more type-in traffic to the corresponding domain. The advertising revenue from a page was correlated to bid price, because the same advertisers bidding on search phrases were also buying PPC ads on domain landing pages. The commerciality metric, therefore, approximated the commercial value of a domain as an advertising asset.
After manually building a list of a few hundred search phrases and computing their commerciality, something remarkable happened. The commercial structure of the internet started to become visible.
The metric cleanly separated high-volume but low-value phrases from low-volume but high-value ones. “Free music” had enormous search volume but a tiny bid price. Those users were looking for pirate download sites, and no legitimate advertiser wanted them. “Mesothelioma,” on the other hand, had modest search volume but astronomical bid prices. Ambulance-chasing lawyers were paying a fortune for every click, because a single asbestos litigation client could be worth millions. And by looking at which companies were bidding on these phrases, you could reverse-engineer how their websites and business models actually worked.
I decided to go all in. The commerciality metric became the engine of DBS’s domain acquisition strategy. We sourced search volume and bid price databases until we had complete coverage of the English-language internet, computed the metric for every phrase, converted each phrase into a domain name, ranked them in descending order, and registered what was available.
But it was a race. Other professional domain owners were running similar playbooks, and the best names went fast. The competitive pressure was constant, and it never stopped, because domain names expire if not renewed after a year, creating a perpetual flow of new inventory. Names that someone else had registered and decided not to keep would drop back into the pool, and you had to be ready.
To stay ahead, DBS built what were, at the time, massive systems to acquire, process, and manage the datasets needed to source and value domains. We also used the data to support our intermediary business which helped other professional domain owners make better decisions about their own portfolios. All of this was pre-cloud. Everything ran on our own infrastructure, which itself required constant management. It was a data business, an advertising business, a registrar, and an infrastructure operation all at once.
At its peak, DBS was the second-largest domain name owner in the world, with 600,000 domain names. The vast majority had a high commerciality metric.
Because we were deploying the company’s capital, I played an active role in the registration process, personally registering roughly half of those domains, one by one, over a period of about two years. And it was during this process that something unexpected happened.
The pattern recognition kicked in.
What 600,000 registrations revealed
For most names, I would intuitively understand the search volume and bid price and the types of companies and business models behind the bidding. But every day, there would be an outlier or two, that is, a phrase where the economics didn’t match my intuition. I’d explore the companies bidding on that phrase, and I’d discover a new angle on a business, a variation on a business model, or a marketing technique I hadn’t encountered before.
Over two years, I saw thousands of internet business models. Not in typical places like a textbook or a pitch deck. Instead, I saw them in the raw data of what real companies were spending money on, and how much they were willing to pay. It was a comprehensive, hands-on survey of how the commercial internet worked delivered one domain name at a time.
Of course, the internet didn’t stand still. Over time, direct navigation waned as Google became dominant and users stopped typing domains speculatively. Google also became the monopoly provider of online advertising and took most of the economics. The rental yields that had sustained professional domain owners gradually eroded.
But the commerciality metric didn’t become irrelevant; it evolved. As advertising yields declined, domain sales became more important. And the same signals that made a domain commercially attractive for advertising, like high search volume, high bid prices, and strong commercial intent, also made it valuable as a brand asset to an end buyer. The metric proved a good starting point for valuing domains in the secondary market too.
What domain names taught me about angel investing
There’s a broader business model lesson in all of this that I carry into my angel investing. As a mathematician, I have a natural soft spot for data businesses. But data businesses can be deceptively tricky to monetise. The fundamental question is: do you sell the data as an agent, or do you use the data to make money as a principal?
DBS had assembled a comprehensive dataset which included search volumes, bid prices, and commerciality scores, but we couldn’t sell that data as an agent. What we could do was use it as a principal, deploying it to register domain names that we owned and monetised. The data was the competitive moat. The domains were the monetisation. That distinction – agent versus principal – turns out to be one of the most important questions you can ask about any data-driven business.
I’m always on the lookout for founders who have found creative ways to monetise their data assets, particularly those who have figured out how to act as principal rather than agent. It’s a pattern I recognise because I lived it.
There’s another story to tell about how DBS convinced GoDaddy to sell our domain names in line with the primary registration path. That deal, which happened twenty years ago, was in my view the most important development in the professional domain name space. And DBS could not have done it without the book of domains we had built using the commerciality metric. But that’s a story for another post.
Despite all this happening twenty years ago, the experience remains foundational to how I look at angel investments today. When a founder walks me through their revenue model, I’m not evaluating it in the abstract. I’m comparing it, often unconsciously, against thousands of business models I absorbed during those two years of registering domains. The patterns are still there. The outliers taught me to stay curious. And the commerciality metric taught me that the best way to understand a market is to follow the money, literally, one search phrase at a time.