Where to Find Real Domain Sales Data (and How to Read It)

Valuation in this industry rests almost entirely on one question: what have similar names actually sold for? Getting good at finding that data — and knowing what it leaves out — is the difference between pricing from evidence and pricing from hope.

The main sources

NameBio is the largest public database of reported domain sales, aggregating from marketplaces and auction houses going back years. It's searchable by keyword, price range, date, extension, and length, and it's where most comps research starts and ends.

DNJournal publishes a weekly list of the largest reported sales. It's the industry's public record of the high end and useful for seeing which categories are moving, but the threshold for inclusion means it's the least representative source for ordinary names.

Marketplace sold listings. Some marketplaces publish recent sales. Narrower coverage, but the data is first-party.

Registrar and marketplace annual reports. Occasionally publish aggregate statistics — total sales volume, average prices, category breakdowns. Useful for direction of travel rather than for pricing a specific name.

Forum sales threads. Investors report their own sales, including small ones. Unverified and self-selected, but it's the only place you'll see the $300 sales that never appear anywhere else.

The biases you must correct for

This is the part that matters, and it's why two investors can look at the same database and reach opposite conclusions.

1. Reporting bias — the big one

Sales get reported when they're interesting, which means large. A $40,000 sale gets published. A $400 sale usually doesn't. Marketplaces report selectively, and private sales — a large share of all transactions — are frequently never disclosed at all.

The practical effect: the visible data is skewed high, and if you calibrate on it you will systematically overprice.

2. Survivorship bias — the invisible one

Every sales database records only the names that sold. The millions that were listed for years and never sold, and the millions more that quietly expired, leave no trace.

This is the single most distorting omission. Looking at comps tells you what names like yours sold for, conditional on selling at all — which is a completely different question from whether yours will sell. A category can show a healthy cluster of $2,000 sales while 99% of names in that category never move.

3. Recency and category drift

A 2019 sale in a category that has since cooled is a poor guide. .ai names are the clearest current example — the market has shifted rapidly enough that even two-year-old data can mislead in both directions.

4. Cherry-picking your own comps

The most common self-inflicted bias. You search, find fifteen results, and unconsciously weight the three highest because they support the number you wanted. The correction is mechanical: decide your filters before you look at prices, then take everything that passes.

How to search well

Search the keyword, not the whole name. Your exact name has never sold. Its components have.

Search both words separately for a two-word name, then look for two-word sales in the same industry.

Filter by extension. .com comps do not price a .net, except as a starting point you then discount heavily.

Filter by date. Last three years unless the category is unusually stable.

Filter by length. A 6-character name in your keyword tells you nothing about your 16-character one.

Sort by date, not price. Sorting by price puts the outliers at the top and anchors you to them immediately.

Reading a result properly

For each comp, ask three things before you use it:

1. Is this genuinely comparable? Same extension, similar length, similar structure, same category.
2. What's different, and which direction does that push? Almost every difference pushes downward from a comp you liked.
3. What don't I know? Whether the buyer was a funded company, whether it was brokered, whether the number includes fees.

That third question is why a range beats a point estimate. You are working from partial information about transactions you weren't part of.

What good output looks like

After twenty minutes you should be able to write a sentence like:

Two-word service .coms in the home-services category, 12–16 characters, sold between $700 and $2,800 over the last three years, with most between $1,000 and $1,600. Nothing comparable sold above $3,000.

That is defensible. You can say it to a buyer. It gives you an asking price, a target, and a floor.

Compare it to "the appraisal tool said $4,200," which gives you nothing you can defend and signals inexperience if you repeat it in a negotiation.

The number you should actually track

Public data tells you about the market. Your own data tells you about your portfolio, and it's more useful.

Record every sale you make: the name, what you paid for it, when, what it sold for gross, the commission, the net, and how long you held it. After a handful of sales you'll have your real average, your real hold time, and your real sell-through rate — which beats any industry average, because it's measured on the inventory you actually buy.

Most investors can't tell you these numbers. The ones who can make better decisions about what to buy and what to drop, because they're calibrated to reality rather than to the visible tail of a biased sample.