How often does a listed domain actually sell?
GoDaddy puts a valuation on every one of its auction listings. We take a full copy of all 999,999 of them every day and follow each listing until it ends, so we can say what fraction of them go anywhere.
Why an appraisal on its own cannot answer it
A comparable-sales figure is a price you observe given that a sale happened. Domains that never sold are not in the comparables as zeroes. They are absent completely. So if 2% of listings sell in a year and the sales that do happen median $1,500, then the value of holding a listing is not $1,500. It is nearer 0.02 × $1,500 = $30 before you pay a single renewal. That is a 50x difference, and the same arithmetic sits under every free appraisal on the internet.
The missing piece is the sell-through rate. There is nowhere to buy it, which is why we are measuring it.
What one day of the file already shows
From the file of 2026-08-17:
| Live listings | 999,999 |
| Carrying zero bids | 99.74% |
| GoDaddy valuation, median | $19 |
| GoDaddy valuation, 90th percentile | $1,124 |
The obvious objection, and the control for it
An auction with six days left should have no bids yet, so a pooled zero-bid rate means nothing on its own. Splitting it by how close each auction is to closing gives the control. The rate does climb the whole way toward close, which is what a sound measurement should do, and it still finishes under two percent.
| Time to auction close | Auctions | With at least one bid | Rate |
|---|---|---|---|
| 1-6h | 43,655 | 824 | 1.89% |
| 6-24h | 8,550 | 170 | 1.99% |
| 1-3d | 128,336 | 533 | 0.42% |
| 3-7d | 314,323 | 829 | 0.26% |
| >7d | 213,480 | 235 | 0.11% |
Even inside the last few hours, almost nothing has attracted a bid. Treat this as a floor on final sell-through: bidding concentrates in the closing minutes and our file is taken earlier, so the bias makes the market look quieter than it is. That is the direction that would push us to abandon a product that was working, which is the safer way for it to be wrong.
Is the appraisal just the asking price wearing a hat?
If it were, there would be nothing independent to calibrate and the question would be circular. We registered that as a reason to stop before we looked.
It survives. Holding the asking price fixed at $1, the valuation still spans $1 to $1,112 across the 10th to 90th percentile, and the log to log correlation between valuation and price over 833,002 listings is 0.0626, which is indistinguishable from zero. The appraisal is an independent estimate, so the question is worth asking.
Outcomes so far
We have now watched 310,385 auctions run to a finish across 5 day-pairs.
| Window | Listings ended | Ended with a bid | Ceiling on sell-through |
|---|---|---|---|
| 2026-08-12 to 2026-08-13 | 62,834 | 1,056 | 1.68% |
| 2026-08-13 to 2026-08-14 | 55,422 | 1,387 | 2.50% |
| 2026-08-14 to 2026-08-15 | 64,919 | 1,075 | 1.66% |
| 2026-08-15 to 2026-08-16 | 61,657 | 1,030 | 1.67% |
| 2026-08-16 to 2026-08-17 | 65,553 | 984 | 1.50% |
That last column is a ceiling. A seller who withdraws a listing looks exactly like one who sold it in this feed, so every auction that ended carrying a bid is counted as a possible sale. The true share is lower. We wrote that ambiguity down before any of this data existed rather than discovering it later.
What we wrote down before the data
The last study in this programme died because a headline number got believed before anyone broke it down. So the tests that would kill this one were fixed in advance, on 2026-08-12, and cannot be edited now:
- Measurability. At least 10,000 auctions have to reach a settled end state, with sales separable from withdrawals. If they cannot be, there is no product and we stop.
- The gap has to be real. Median appraisal divided by realised price of at least 3x, over auctions with a genuine closing price.
- It has to survive conditioning. That 3x must hold in at least two price bands above $100. If it lives only in the sub-$50 tail then it is a fact about worthless domains and no owner of a real one should care.
- Silent-failure guard. The collector refuses to archive a stale feed, a changed schema, or a row count wildly off the day before. Any day archived without those checks passing is excluded when we judge.
A measurement you cannot fail is not worth running. Each of these can return a verdict we would rather not get.
Scope
This covers GoDaddy expiry and Buy Now auctions. It says nothing about Sedo, Afternic, Dynadot or private sales, and any figure taken from it should carry that channel with it. It also does not tell you which domain to buy. Knowing how often things sell is a bound on what a listing is worth, and a bound is not a trading strategy.
Look up your own domain
Search any of the 999,999 listings in today's file and see its
appraisal, price and bids.