How much is my domain worth?
Probably less than you hope, and more than the free tools say about a few lucky names. Here is how to find a defensible number.
Most domain names are worth less than their owners hope. A name that does sell in the aftermarket sells for a median of $2,025, and only about 5% of those sales reach $10,000. Your number depends on comparable sales, not on a free calculator.
That median comes from 43,301 reported .com sales since January 2025, gathered from multiple sources. It's the price of names that found a buyer. The ones that didn't sell aren't in any list, and they're most of the world's domains. So "what's it worth?" has two honest answers: what it might sell for, if it sells, and how likely that is. Nobody can compute the second part from the name alone.
What names that did sell went for
Here's every reported .com sale of $1,000 or more since January 2025, sorted into price bands. Almost half landed between $1,000 and $2,000. Fewer than one in six cleared $5,000.
The shape matters more than the median. Prices pile up near the bottom and trail off into a long, thin tail. The top 0.3% (124 sales at $100,000 or more) drags every average upward, which is why you should ignore averages and look at the median and the quartiles. Half the sales were between $1,361 and $3,550.
Which part of the pile is your name in?
Length and structure move the number more than anything we can measure from the name alone. Short letter-only names cost more. Long phrases cost less. These are medians for sales since January 2025, with the number of sales behind each:
| Kind of name | Median | Sales ≥ $5,000 | Sales |
|---|---|---|---|
| One real word | $7,877 | 61% | 471 |
| Two words | $2,304 | 18% | 17,636 |
| Three or more words | $1,825 | 10% | 17,747 |
| Has a hyphen | $1,725 | 11% | 1,454 |
| Digits only | $4,450 | 46% | 569 |
Two things jump out. A single dictionary word has a median of $7,877, more than three times a two-word name, and about 44% of those sales are $10,000 or more. But there are only 471 of them in nearly two years. Single words are rare because they're scarce, and the scarcity is the whole price. Meanwhile "two words smashed together" is the bulk of the market: 17,636 of the 43,301 sales. If you own one of those, you're in the crowd, and the crowd's median is $2,304.
Length does something similar. For letters-only .com names: four letters median $3,000, six letters $2,550, ten letters $2,111, fifteen $1,940. The slide is gentle after about eight characters. The jump is at the short end, and it's a cliff: the 73 three-letter .com sales had a median of $39,600. If your name is four letters or fewer, nothing in this guide is a good estimate of it. Get a specialist's opinion.
How to find your own comparable sales
A comparable sale ("comp") is a real sale of a name like yours. This is the only method that holds up, and it's free. Do it in this order:
- Pin down what yours is. Length, extension, how many words, whether they're plain English, any digits or hyphens. Write it down. "Seven letters, .com, one coined word" is searchable. "Cool brandable name" isn't.
- Search sold names with the same shape. Public sold-domain databases and the weekly sales reports that marketplaces and forums publish are free or cheap to search. You want five to ten sales of names that look like yours, from the last couple of years.
- Match the venue. A name that sold on a wholesale auction site and a name that sold to an end user at a broker's asking price are not the same kind of sale. In our data, the median at one big registrar marketplace is $1,800, at Sedo it's $4,000, and at Afternic $3,188. Some of that gap is the names, not the venue, but it's a reminder that "sold" can mean very different things.
- Throw out the outliers and read the middle. If your five comps are $1,400, $1,800, $2,200, $2,500 and $38,000, the $38,000 name had something yours doesn't. The middle three are your range.
- Adjust for the obvious. Is yours a clearer word? Easier to spell? Does a whole industry type it? Move up. Hard to say aloud, trademark trouble, awkward letters? Move down.
NameFew does the first four steps for you. Type a name on the appraisal page and you get a handful of actual sales of similar names, not just a figure.
Why one number is a lie
Every appraisal tool returns a single dollar figure, sometimes to the dollar. Don't believe it. We tested our own model on 6,443 sales it hadn't seen, and half the time the real price landed within about 1.5 times of the estimate, in either direction, and the other half of the time it was further out. Add an AI opinion on top and the typical miss stays about the same. Details are in our measurement of AI appraisals.
That's why NameFew shows a range, calibrated so about 80% of comparable sold names fall inside it. The honest range for an ordinary name is wide: the AI opinion's 80% range had a median width of about five times between low and high. A $2,000 name might be "$700 to $3,500." That's not vagueness. That's what the sales show.
And there's a second, uglier reason. Even a perfect model of the middle would miss the sale that matters. In a separate test on 300 unseen July to October sales, the 23 names that sold for $8,000 or more had a median sale price of $10,400, while the AI's median guess for those names was $2,800. The five biggest misses sold for $24,000 to $40,000 where the AI had said $1,350 to $4,000. If you think your name is one of those, a range will never tell you. A buyer will.
What I'd do, in order
- Get a range, not a number. Run the name through NameFew and one other tool. If they differ by a factor of three, that tells you something about the uncertainty, not about which tool is right.
- Read the comps yourself. Ten minutes. If you can't find any sales that look like your name, that's information too.
- Price high enough to leave room and low enough to be believed. The next stage is the seller's page, which verifies you own the name by DNS and lists it for free. There's no listing-price formula we trust. There's the comps, and then your nerve.
- Don't pay for an "official" appraisal to decide whether to sell. A paid certificate is another opinion. A sale is the only appraisal that counts.
The robot-versus-human part
I built NameFew in 2020 as a crowdsourced appraisal experiment, and it never recorded a single vote. I rebuilt it in October 2026 around a simpler question: can people looking at a name beat a model trained on thousands of real sales? We think they might, on exactly the names the models miss. The human side is built but not switched on yet, so for now the robot's range is all you'll see. Once people have scored a name, you'll see their range next to it, and later we'll check both against what the name actually sells for. If you want to help settle it, the game takes a minute.
Where the numbers come from. 142,183 reported .com sales of $1,000 or more from 2012 to October 2026, merged from multiple sources, with duplicates removed. Wholesale auction venues are excluded. Figures above use the 43,301 sales since January 2025. We publish aggregates only. The sales are reported by sellers and venues, and we can't verify each one.