Two pitch decks sat on the same conference table, and they disagreed by a factor of a hundred. The first said the average reported domain sale in 2026 was about $3,500. The second said a four-character name had just sold for $399,995. Both decks cited real data. Both presenters believed their numbers. A buyer looking at a short name has to decide which deck is describing the market they are actually in, because the answer changes the price by six figures.
Start with the first deck, because it is the more careful one. DNJournal, which tracks reported domain sales week by week, published seventeen median figures across 2026, and they barely moved: $3,262 in late September, $3,944 in early September, $3,746 in late August, $3,508 in early August, $3,115 in late July, $3,699 in early July, $2,999 in late June, and a string of similar numbers back through February's $4,094. An independent summary of the same reports put the median at $3,462 for February through May and $3,508 for late May through September, a change of 1.3 percent. That is a flat market, described precisely. A name that sells for the median is a name that sells for about three and a half thousand dollars, and most reported sales cluster near it.
The second deck is also telling the truth, just about a different market. On July 8, 2026, goka.com — four letters, pronounceable, no meaning in English — sold for $399,995, the top reported sale of its week. In March 2025, bet55.com, a word plus two digits, sold for $145,000. These are not typical sales. They are the sales that make the top of the weekly list, which is why they get remembered. But they are real, and they establish something the medians cannot: brevity itself has buyers at prices the median never sees.
The gap between the two decks is the entire business of short names, and it exists because reported sales are not the whole market. Only about five to ten percent of retail domain sales are ever reported, according to the coverage estimates NameBio publishes. The reported set skews toward wholesale and reseller deals — professionals trading with each other, where the buyer intends to sell again and prices accordingly. An end user buying a name to build a business on for ten years is not in that dataset most of the time, because end users do not report their purchases. The median describes what resellers pay each other. It does not describe what a company pays for its own name. Those are different transactions that happen to use the same asset.
The tier bands make the distinction concrete. An analysis of NameBio, Park.io, and GoDaddy Aftermarket data published by Names.Center in 2026 put premium four-letter brandable names at $1,500 to $8,500, and premium pronounceable three-letter names at $8,500 to $22,000. A separate study of 1,163 reported single-word .com sales by MetricName found the middle half landing between $399 and $8,888. Notice what those bands share: they top out where a company's brand budget begins. A reseller will not pay $20,000 for a name they must resell at a profit. A company that will print the name on invitations, paint it on a storefront, and say it in a broadcast ad will, because the name's cost is spread over every customer the business ever serves. The same four characters are worth $3,000 to one buyer and $30,000 to another, and both valuations are rational.
That is also why the famous sales at the top of the market mislead more than they inform. Voice.com sold for $30 million in 2019. Chat.com sold for $15.5 million in 2024. Gold.com sold for $8.5 million in 2024. These are category-defining English words bought by funded companies, and they tell a small wedding vendor or a startup nothing about what their name is worth — except that the ceiling for a perfect word is essentially unlimited, which is not a useful fact when you are pricing a real asset. The honest comparison set is smaller and duller: the bands, the medians, and the handful of short sales in the same shape as the name in question. A wedding-adjacent name listed on a brandable marketplace at $2,495, as one six-letter ceremony name currently is, sets a floor for weak names. It does not set a price for strong ones.
Here is the counterargument, and it is the one that should worry any seller of a short name. Most short names never find an end user at all. The reseller market exists precisely because end users are rare and hard to reach, and a name can sit for years while its owner waits for a buyer who never arrives. Holding costs are trivial — about $10.88 a year in renewal fees at a discount registrar — which makes waiting cheap, but cheap waiting is not a strategy. A name priced for a buyer who never comes is not an asset appreciating in a vault. It is inventory taking up a slot that could hold something else. The medians, for all their wholesale skew, carry a warning the top sales do not: the typical outcome for a short name is a four-figure sale to another reseller, not a five-figure sale to a company. Anyone who prices only for the second outcome should know how often the first one happens.
So what is a four-character name actually worth? The unsatisfying answer is that it has two prices, and the seller has to choose which market they are selling into. Price for the reseller and it will move, probably for a few thousand dollars, probably soon. Price for the end user and it may sit — months, maybe years — until the one company that needs that exact name finds it, and then it is worth what a decade of brand use is worth to them. Neither choice is wrong. The mistake is confusing them: quoting the $399,995 sale to justify a price the reseller market will never pay, or quoting the $3,500 median to talk a company out of paying for the name its customers will remember. The decks on the conference table were both right. The buyer who understands which deck they are in is the one who gets the name at the right price.
Sources: DNJournal weekly sales reports, 2026 (seventeen median figures, $2,999–$4,094); audit.domains summary of DNJournal data; Names.Center 2026 short-name tier analysis (NameBio, Park.io, GoDaddy Aftermarket data); MetricName single-word .com study (n=1,163); NameBio coverage estimates (~5–10% of retail sales reported); BrandPa ceremony listings; reported sales: goka.com $399,995 (July 8, 2026), bet55.com $145,000 (March 2025), Voice.com $30 million (2019), Chat.com $15.5 million (2024), Gold.com $8.5 million (2024).