How to Price a Domain Name With Comparable Sales

To price a domain name with comparable sales, find recorded sales of names that share its extension, length, word type and meaning, note where and when each one sold, adjust for timing and for wholesale versus retail venues, and then set a range rather than a single number. The comps tell you what buyers actually paid. Your job is to explain why your name sits where it does among them.

This guide walks through that process step by step, with a worked example built entirely from recorded sales. The example prices a hypothetical developer-term .ai name, so you can see each decision, including the ones that move the number the most.

Why comparable sales beat gut feeling and list prices

A domain has no cash flow of its own, no rent and no earnings. Its value is whatever a buyer is willing to pay, which is why the most reliable evidence is what buyers have paid for similar names. Real estate agents work the same way: they look at sold houses on the same street, not at asking prices in the window.

List prices are the trap. A marketplace can show a name for 50,000 USD for years without a single offer. Sales are different, because money changed hands. Every figure in this article comes from a recorded sale with a venue and a date, taken from public reports of domain sales between May 2021 and May 2026.

It also helps to know the shape of the market before you start. Across 128,867 recorded sales in that window, the median sale was 1,900 USD and 86.8 percent of sales were below 5,000 USD. Only 297 sales reached 100,000 USD or more. Most names, even good ones, trade in the low thousands. If your first instinct says six figures, the comps need to prove it.

Step 1. Describe the name precisely

Before searching, write down what the name actually is. The features below decide which sales count as comparable.

• Extension. A .com sale is weak evidence for a .ai name and the other way round. Markets differ by extension.
• Length. Count the characters before the dot. Short names behave differently from long ones.
• Word type. Is it a dictionary word, a developer term, an AI term, a brandable invented word or an exact match keyword?
• Style. One word, two words, acronym, numeric, hyphenated.
• Singular or plural. Plurals usually trade below the singular, though not always.
• Meaning. Which industry would buy it? A name that means something to developers belongs with other developer names.

For our example, take deploy.ai. It is a single English word, six characters, a .ai extension, a common developer term (to deploy code), singular, with no digits or hyphens. There is no recorded sale of deploy.ai in the corpus, which makes it a realistic case: most names you price have never sold before.

Step 2. Gather candidate comps

Search for sold names that share the extension and the meaning. Matching letters is not enough. "Deploy" shares almost no letters with "server" or "kernel", yet a buyer shopping for a developer tool name would consider all of them. Good comps share the job the word does, not the spelling.

Here are the recorded sales of developer-term .ai names that fit our example.

Ten sales, seven distinct names, a range from 2,424 USD to 72,999 USD. Taken at face value that range is useless. The next steps turn it into an estimate.

Step 3. Score how close each comp really is

Not every comp deserves the same weight. Rank them by how many features they share with your name, and write down the reasons. A simple scoring habit keeps you honest.

• kernel.ai is close. One word, six characters, a core developer term, sold in 2024.
• server.ai is close. One word, six characters, a developer and infrastructure term, with a 2025 sale.
• stack.ai is close in meaning, but it is shorter (five characters) and "stack" is used far beyond engineering (tech stack, full stack), which widens the buyer pool.
• debug.ai is close. Five characters, a pure developer verb, like deploy.
• workflow.ai is relevant but different. Eight characters and a business term as much as a developer one.
• webhook.ai is a narrower technical compound. Fewer buyers, which shows in the price.
• node.ai is short and technical, but sold early, in 2022.

Deploy, like debug and kernel, is a verb or noun every software team uses daily. It is more general than webhook and more technical than workflow. That places it in the middle of the group.

Step 4. Adjust for timing

Timing is the adjustment people skip, and in .ai it is the largest one. The same names sold at very different prices a few years apart.

The wider .ai market moved with them. The corpus records 152 .ai sales in 2021, 547 in 2023 and 2,070 in 2025. Demand for AI-related names grew sharply, and short developer terms were among the first to be bought.

So the 2021 and 2022 sales of stack.ai, server.ai and node.ai say little about what a buyer would pay after 2024. Drop them from the core set or treat them as a floor. The comps that matter for deploy.ai are the 2024 and 2025 sales: stack.ai at 72,999, kernel.ai at 14,599, server.ai at 29,400 and webhook.ai at 2,424, plus the 2023 debug.ai and workflow.ai sales as context.

Timing cuts both ways. A comp from the peak of a trend can overstate value if demand has cooled. Look at the dates of the most recent sales in your segment, not just the highest ones.

Step 5. Separate wholesale from retail

Where a name sold matters as much as when. Expired auctions and investor venues (DropCatch, NameJet, Park.io, Dynadot, Namecheap, Whois.ai and similar) are where investors buy from each other. End-user marketplaces (Afternic, Sedo, Atom.com, BuyDomains, Spaceship.com and similar) are where businesses buy the name they want to build on.

Across the whole corpus, retail venues show a median sale of 2,288 USD against 1,725 USD for wholesale venues, and an average of 6,437 USD against 3,312 USD. For .ai specifically the gap is wider: sales at retail venues had a median of 9,995 USD, against 2,000 USD at wholesale venues.

Look back at our comp table. Nine of the ten sales happened at wholesale venues. That means they mostly describe what an investor paid, not what a startup would pay for the same name. A clear example of the gap in .ai is zip.ai, which sold for 1,610 USD at Park.io on 2021-08-10 and then for 200,000 USD at Afternic on 2025-03-27. Timing and venue together explain most of that difference.

The practical answer is to produce two numbers. A wholesale figure tells you what you could likely get from another investor in a reasonable time. A retail figure tells you what an end user might pay if you wait for the right buyer.

Step 6. Set the range and the reasoning

Now put it together for deploy.ai, using only the recent, close comps.

• The closest recent wholesale sales are kernel.ai at 14,599 USD (Dynadot, 2024-10-08) and server.ai at 29,400 USD (Namecheap, 2025-02-25).
• debug.ai, an equally narrow developer verb, reached 15,000 USD (Whois.ai, 2023-06-05), before the strongest part of the .ai run.
• stack.ai at 72,999 USD (Dynadot, 2024-06-03) is the upper marker, but it is shorter and has a broader meaning, so deploy.ai should sit below it.
• webhook.ai at 2,424 USD (Namecheap, 2025-10-08) is the lower marker, a narrower compound with fewer buyers. deploy.ai should sit well above it.

A reasoned wholesale estimate for deploy.ai would land roughly between 12,000 and 35,000 USD, with the likely figure near the middle of the kernel.ai and server.ai sales, around 20,000 USD. Confidence is medium: there are four or five good comps, but they are spread out and most come from investor venues.

The retail figure is higher and less certain. Retail .ai sales in the corpus show a median about five times the wholesale median, and single names such as zip.ai show far larger jumps. A retail asking range of 40,000 to 75,000 USD for a strong developer verb is defensible from the evidence, but it depends on finding a buyer who is building a deployment product. That is a matter of time and outreach, not certainty.

Note what the rationale does. It cites each comp by name, price, venue and date, says why each one ranks above or below the subject, and states what would change the number. That is the difference between a number and an argument.

Step 7. Write down what would move the number

An estimate is more useful when it says what would raise or lower it. For deploy.ai:

• Raise it. A recent retail sale of a comparable developer verb in .ai, a known company actively building a deployment tool, or another 2026 sale of a short developer term above 30,000 USD.
• Lower it. Recent wholesale sales of developer terms below 10,000 USD, a cooling in .ai sale counts, or trademarks that narrow the pool of safe buyers.

If you cannot find at least three good comps, say so, widen the range and lower your confidence. If no comparable sale above the 1,000 USD reporting floor exists at all, the honest conclusion is that names like this usually trade below that level, and no specific figure should be invented.

Common mistakes when pricing with comps

Using the single highest sale as the anchor

Every segment has an outlier. In .com, icon.com sold for 12,000,000 USD (Hilco Digital Assets, 2025-04-24). That says nothing about a typical one-word .com. Anchor on the cluster of close comps, not the record.

Mixing extensions

A .com one-word sale is not a comp for a .ai one-word name, and a .ai sale is not a comp for a .io name. Use other extensions only as loose context when your own extension has too few sales.

Ignoring plural and length

Plural forms and longer names usually sell for less, though the evidence varies. rebate.com and rebates.com each sold for 500,000 USD (Sedo, 2022-12-01), so plurals are not always penalised. Check actual pairs before applying a discount.

Forgetting the venue

Mixing wholesale and retail sales into one average produces a number that matches neither market. Keep them apart and present both.

How SoldComparables helps with each step

The steps above can be done by hand, with patience. SoldComparables runs them for you. It searches 128,867 recorded sales from May 2021 to May 2026, uses word meaning rather than only letters (so "deploy" finds "server", "kernel" and "debug"), and gives each comp a relevance score with up to three plain reasons. It labels every sale by venue tier, separates wholesale and retail, and returns a low, likely and high value with a plain English rationale that cites the comps by name, price and date. If there are too few good comps, it says so and lowers confidence.

If you want to go deeper on the inputs, see our guide to domain comps and the full description of the valuation methodology. To try the process on your own name, run a domain appraisal search.

Frequently asked questions

How many comparable sales do I need to price a domain?

Three good comps is a practical minimum, and five to eight is better. Good means same extension, similar length and word type, and a sale in the last two or three years. With fewer than three, widen the range and treat the estimate as low confidence.

Should I use asking prices as comps?

No. An asking price is a hope, not a sale. Names can sit at a list price for years without an offer. Only completed sales with a known venue and date show what buyers actually paid.

How old can a comparable sale be?

It depends on how fast the segment moves. In .ai, sales from 2021 are far below later sales of the same names, so they work only as a floor. In stable segments, sales from three or four years ago can still be useful with care.

Why give two numbers instead of one?

Because there are two markets. Investors buy at wholesale prices to resell, and end users pay retail prices for a name they plan to build on. A wholesale figure tells you what a quick sale might bring, and a retail figure tells you the upside if you wait.