197 distinct .si closes with a $140 median, split at the bid count: 93 cards drew ten bids or fewer at $110 while five drew forty or more above $2,000. The four things we change before touching the price, the reserve setting that decides whether a sale record exists at all, and the relist calendar we run on a stalled card.
An auction that does not sell leaves no trace of the reason. There is no error message, no rejection, no note in the panel — the card closes, the name is still yours, and the only thing the seller learns is that nobody paid. What follows that silence is almost always the same conclusion: the price was too high.
Usually the price was not the problem. We have the closing data to show it. Reading a sold-auction wall properly is the difference between a listing that gets a second chance and a name that gets discounted into a wholesale board because its owner diagnosed the wrong failure.
What the closing wall actually contains
Three figures carry most of the argument. Across the recent .si auction results the wall returns 302 cards, which collapse to 197 distinct closing prices once the names Atom re-listed in a later event are deduplicated. The median of those 197 is $140. The lowest close is $110 and the highest is $7,600.
Half of every name that sold in that window went for $140 or less. That is a number that barely clears a year of renewal, and for some extensions it does not clear it at all. A market where half the completed sales sit under the cost of holding the name is not a market with bad names in it. It is a market where the cheapest transactions are the most visible, because they are the ones that always happen.
That is the first thing to understand about an auction wall: it is a record of sales, not a record of attempts. The names that failed are not on it. If you read it as a price guide without accounting for what is missing from it, you will price every name in your portfolio at the level of the names that had nobody competing for them.
The split that matters is the bid count
The second mechanic is the bid count, and it is the most useful number in the whole panel.
Median bids across the set is fifteen. Split the set at ten bids and the shape changes sharply. Ninety-three of the 197 cards closed with ten bids or fewer, at a median of $110. Only five cards drew forty bids or more, and that group closes above a $2,000 median — roughly twenty times the price of the weakly contested group.
Two auctions, same wall, same event calendar, same kinds of names. The difference between them is not the ask. It is whether the card reached bidders who wanted the name.
A forty-bid card means the listing was found, the name was legible at thumbnail size, and the category told the bidder what he was looking at. A six-bid card means the card appeared in front of a handful of people who were scrolling anyway, and closed. Lowering the price on the second card does not turn it into the first card. It produces the same silence at a worse number — and if the reserve is lowered with it, it produces a sale at the bottom of the range instead of no sale at all.
The visible difference: reserve or not
Of the 302 cards on that wall, 223 carried a no-reserve tag and 79 were reserve-met. After deduplication, the no-reserve group's median is $120 and the reserve-met group's median is $350 — close to three times as much.
This is not the reserve doing magic. It is a selection effect, and once you see it, the wall stops being confusing. A no-reserve sale always produces a record, because the name sells to whoever is in the room. A reserve-met sale produces a record only if at least two bidders wanted the name past the seller's floor. So every record on the wall is real, but the two groups are drawn from completely different populations: one is "every name we ran", the other is "the names that found a buyer who would fight for them".
Which means the wall's cheap half is cheap by construction. When you look at a $110 close and conclude that your $500 name is overpriced, you are comparing a name that sold by obligation against a name that has not been tested yet.
A reserve does not raise the price. It decides whether a sale record exists at all — and a record at the wrong price becomes the comparable someone else uses against your name.
Four fixes, in the order we apply them
When one of our cards does not close, we work through these in order. Price is last because it is the only one that costs money and the only one that cannot be undone.
One: the card has to read at 300 pixels wide
A bidder sees dozens of cards at thumbnail size. Yours competes with names that are shorter, cleaner and more familiar. If the card carries a logo that swallows the letters, a background that fights the type, or a name that does not resolve small, the bidder scrolls. This is the failure that looks most like a pricing problem and is furthest from one.
The correction is structural, not decorative: the name is the largest element on the card, the category is legible directly beneath it, and the visual weight sits on the letters rather than on the frame. The same name with a cleaner card draws a different bid count next event, at the same price.
Two: the category the card is listed under
The top closers in that set are short, single-word, and easy to say out loud. The bottoms of the list are longer and take more work to read. You cannot rename a name you already own. You can choose which shelf it sits on.
A city-and-trade name listed under "brandable" competes with every invented word on the platform. The same name listed against its city and its trade reaches the two or three buyers who are actually looking for it. The category is a routing decision, and routing decides the bid count far more than the reserve does.
Three: whether there is a word inside the name
An invented root with no real word in it is the hardest kind of card to bid on. A root that carries something the buyer already types — shop, pay, home, fix, auto, consult — gives him a reason that survives the first half-second of looking.
The test is one that costs nothing: say the name to someone who cannot see it, and ask what he typed. If the answer is not the name, the card is asking a bidder to do translation work at a glance, and glance is all you get.
Four: the event the card is placed into
An auction is an event, not a listing. A card that stalls in a week with forty competing listings can clear in a quieter week with ten. Timing is the cheapest correction on this list because it costs nothing except patience — provided the name has any real demand behind it.
If the same name draws no bids across two events in thin weeks, with a legible card and a correct category, that is the first honest signal that the demand is not there. That is when price becomes the right conversation, and not before.
A sequence for the next stall
The rule we run is simple. A card that closes without a sale is not repriced on the same day. It goes back through the list above, and the price is the last field we touch — usually only after a second event with a corrected card.
If a name has stalled twice with a clean card, a correct category and a thin event calendar, the honest options narrow to three: relist at a lower reserve and accept the bottom of the range, move the name to a wholesale board where the buyers are professional and the price is set by resale, or hold it and treat the renewal as the cost of owning an asset rather than a loss on a failed listing.
What we no longer do is the thing that feels most like action: cutting the price on a card that never got read. On the wall, the difference between a $110 close and a $2,000 close is almost never the ask. It is the number of people who ever saw the card.
Buying a name this week?
193 exact-match, brandable and geo domains — buy-now price, escrow-protected, same-day transfer. See the portfolio · the contact page
Top comments (0)