domainregistrar.ai · the Domain Desk
The name market has two faces. We show you both.
Real sale records. Dated pricing evidence. Names listed by their holders. Settled through Escrow.com.
Learn how the market works before you decide what anything is worth.
Browse the shelf →The aftermarket has two faces.
Face one — the expiry stream
A number beside the renew button. A lapse. A very different price later.
Registrar-run expiry auctions and drop-catch services acquire names the moment they lapse and resell them. Low appraisal figures can sit beside the renewal decision. If a name later expires, it may enter an auction or drop stream and subsequently clear at a very different price. The desk records each event separately — appraisal, expiry, auction, sale — and lets the sequence speak for itself.
Venues in this stream: GoDaddy auctions, Namecheap, DropCatch, Dynadot, NameJet, SnapNames, Park.io, Sav.
Face two — the owner stream
The holder lists it, with a record, and keeps the spread.
The same market, entered through the front door: the name is listed by the person who holds it, carries a dated record of what each number knows, and settles through escrow. The buyer deals with an informed owner rather than an expiry stream, and the desk records the resulting price separately.
Venues in this stream: Sedo, Afternic, BuyDomains, Uniregistry, private, Flippa, Atom, Dan.
Descriptive median spread between the two streams across the whole corpus. Not a claimed causal uplift.
Adjusted spread across 8 cells matched by TLD, year band and name length (geometric–arithmetic, weighted by cell size).
Spread across five appraisals of one .ai name, five opinions, observed 2026-08-19 — and every third-party figure hid its data date.
Methodology shield — read before quoting the number.
The two populations are not automatically like-for-like: TLD mix, name quality, year, length, venue reporting thresholds and selection bias all differ. The raw 4.6× is a description of the corpus, nothing more. Matched by TLD × year band × SLD length (cells with at least eight sales on each side), the spread is 1.6–2.0× — and it is concentrated in this decade:
186 / 82 sales (.com, 11–15 chars)
.com cells
.com: owner n=46 median $2,000 vs expiry n=1,311 median $452
Before 2020 the matched .com cells show no spread at all; from 2024 the .com spread is 4.4×. The desk publishes the script and re-runs it as the corpus grows. Full method on the disclosures page.
One name, five opinions, one day. Each is recorded as an independent opinion with its own transparency; the desk never averages them.
A name is worth what an informed buyer pays an informed owner. This desk exists to make both informed.
What the public record says.
Principal
More than 1.0M domains held.The largest registrar's annual report describes a portfolio of "more than 1.0 million domain names" alongside its aftermarket platform.
Primary-source confirmedAftermarket
Gross-basis revenue. Principal role.Aftermarket revenue is recognized gross because the company "take[s] control of the domain" — it sells as principal, not only as agent.
Primary-source confirmedCommission
15% / 25% structure.Since Feb 1, 2023: 15% when the name points to the company's nameservers, 25% when it does not.
Primary-source confirmedAppraisals
Tool methodology and data presentation changed.April 2023 re-base ("a ton of new data"); March 2024 cap at three free appraisals a day, comparable sales removed.
Primary-source confirmedMarket landmarks — not comparables.
Famous prices are interesting; contextual prices are useful. Before the landmarks, here is where the market actually clears in the desk's corpus — by stream, as quartiles.
Owner-listed
Expiry stream
All classified sales
| Landmark | Price | Year | Grade | Basis |
|---|---|---|---|---|
| AI.com | $70,000,000 | 2025 | Primary-source confirmed | broker announced the price; buyer identified himself publicly (Feb 2026); closed Apr 2025 |
| CarInsurance.com | $49,700,000 | 2010 | Primary-source confirmed | disclosed by the acquirer in public filings |
| Insurance.com | $35,600,000 | 2010 | Primary-source confirmed | disclosed by the acquirer in public filings |
| VacationRentals.com | $35,000,000 | 2007 | Primary-source confirmed | disclosed by the acquirer in its public offering documents |
| PrivateJet.com | $30,180,000 | 2012 | Party-reported | announced by the buyer; not independently confirmed |
See all 15 documented landmarks →
| Landmark | Price | Year | Grade | Basis |
|---|---|---|---|---|
| Voice.com | $30,000,000 | 2019 | Primary-source confirmed | seller and buyer both announced the price |
| Internet.com | $18,000,000 | 2009 | Primary-source confirmed | disclosed by the acquirer |
| 360.com | $17,000,000 | 2015 | Multi-source reported | widely reported; parties did not publish terms |
| Chat.com | $15,500,000 | 2024 | Party-reported | seller-stated figure |
| NFTs.com | $15,000,000 | 2022 | Party-reported | announced by the sellers' side |
| Rocket.com | $14,000,000 | 2024 | Multi-source reported | reported from the buyer's filings coverage |
| Hotels.com | $11,000,000 | 2001 | Multi-source reported | widely reported at the time |
| Tesla.com | $11,000,000 | 2016 | Party-reported | reported figure; buyer never confirmed |
| Casino.com | $5,500,000 | 2003 | Multi-source reported | widely reported |
| Whisky.com | $3,100,000 | 2014 | Party-reported | announced by the parties |
Cost basis on the record.
A landmark is half a record until you know what the string cost the seller. For a hand-registered name the basis is a registration fee and its renewals. The creation date is public record; the multiple that falls out of it has no peer in any asset class a person can buy with a renewal fee.
| Name | Created | Record sale | Years | Fee basis | Multiple | Chain |
|---|---|---|---|---|---|---|
| AI.com | 1993 | $70,000,000 2025 Primary-source confirmed | 32 | $495 | 141,414× | partial |
| Voice.com | 2001 | $30,000,000 2019 Primary-source confirmed | 18 | $285 | 105,263× | registrant held (MicroStrategy) |
| Chat.com | 1996 | $15,500,000 2023 Party-reported | 27 | $420 | 36,905× | partial |
| Rocket.com | 1991 | $14,000,000 2024 Multi-source reported | 33 | $510 | 27,451× | partial |
| Crypto.com | 1993 | $12,000,000 2018 Multi-source reported | 25 | $390 | 30,769× | registrant held 25 years |
| Connect.com | 1999 | $10,000,000 2022 Party-reported | 23 | $360 | 27,778× | partial |
| Gold.com | 1994 | $8,515,000 2024 Multi-source reported | 30 | $465 | 18,312× | partial |
| Green.com | 1999 | $7,500,000 2026 Multi-source reported | 27 | $420 | 17,857× | partial |
| Toys.com | 1998 | $5,100,000 2009 Multi-source reported | 11 | $180 | 28,333× | partial |
| Bot.ai | 2017 | $1,200,000 2026 Marketplace observed | 9 | $150 | 8,000× | partial |
| HighLevel.com | 1998 | $1,000,000 2026 Party-reported | 28 | $435 | 2,299× | partial |
| PrivateLLM.com | 2023 | $250,000 2026 Marketplace observed | 3 | $60 | 4,167× | partial |
| DomainPrice.ai | 2024 | $26,000 2025 Party-reported | 1 | $179 | 145× | registrant held · basis $179 · agreed, not settled |
How to read the multiple. Fee basis = $15 a year × years from creation to sale, the benchmark cost of holding a hand-registered name — an assumption, stated. "Registrant held" means the record shows the seller registered the string; "partial" means the ownership chain between creation and sale is not fully on the record, so the true basis may be a later purchase and the multiple lower. Sale prices carry their own grade. For context: a public equity bought at IPO and held thirty years has returned in the hundreds or low thousands of times at the very best; only a coin bought in 2010 lands in this range — and it needed capital and timing. A domain needed a renewal fee. The arithmetic that follows: at a 145× outcome, one name in 145 — under 0.7% of a hand-registered book — returns the cost of the whole book. That is a statement about the record above, not a forecast; most names never sell, and nothing here promises that any will.
How the desk grades a figure. Primary-source confirmed: the price was disclosed by a party to the transaction in a filing, press release or first-person statement. Multi-source reported: reported consistently by independent trade press; no party disclosure. Party-reported: stated by one party (seller, buyer or broker) without independent confirmation. Marketplace observed: a figure the desk observed on a marketplace or tool on a dated visit; the provider does not disclose its data date. The desk never averages across sources and never infers a missing date. Adult-category records are excluded from this board by policy.
Case files — how to read a price.
The panels are hers; the receipts are the desk's. Fictional character; the numbers are not.
“Same domain. Different price. Different story.”
Evidence on the shelf.
Browse all →
A few names, each chosen because its record proves something about how the desk works — not because it is the biggest ask. The sell, if there is one, happens on the name's own page.
You've seen the record. Now act on it.
Buy
Read the record. Agree terms — Buy now, Offers from, Make offer, or lease-to-own on larger asks. Settle through Escrow.com; the name transfers when escrow releases.
Sell
Prove ownership with one DNS record. List on the desk after review. Pay only when it closes: 10% direct, 15% all-in when a finder brought the buyer. Never above 15%.
Introduce
Bring the buyer through your finder link. Earn 10% of the verified settlement — not on clicks, not on inquiries, on the close.
Put the record to work.
Escrow-settled · no custody · economics published before the transaction