Sell-through rate is one name for two different questions. As a rate it is a percentage: of everything you listed inside a window, how much of it sold. As a duration it is days to sell, which is how long the middle item waited. They answer different things and they move in different directions, and most of what is published about Vinted quotes one of them and calls it the other.
Both are easy arithmetic and both go wrong in a way specific to Vinted. A relist there is a delete and a fresh upload, so the action that is supposed to make an item sell faster also restarts the clock the speed is measured on, and takes the old listing out of the denominator. The figure improves whether or not anything sold sooner. This page owns how the number is measured. Whether relisting actually makes anything sell faster is a separate question with its own page, linked at the foot of this one.
So this page is the method: how to work out both numbers on your own wardrobe with nothing but a list and two dates, and the five places the answer goes wrong. Cavri does half of it. It computes the duration half, median days to sell by department and by price band, with the sample behind each one printed beside it. It prints no percentage at all: the rate is arithmetic you do yourself, off a list Cavri can export. Everything here was checked on 3 September 2026.
Medians, not means · relisted sales declared beside the number · the sales it cannot time are counted, not dropped
A rate is items sold divided by items listed over a window, as a percentage. Days to sell is a duration. A wardrobe can sell 70% of its stock and still take eleven weeks to do it, so a figure quoted without saying which of the two it is has answered half a question.
There is no free refresh on Vinted that keeps the listing. A relist is a delete and a new upload, so the old listing goes and its views and favourites go with it. Two things do keep it: Bump, which is Vinted's own paid promotion, and a price drop, which notifies your followers exactly as a new listing does. The denominator of a rate is everything you listed, and on Vinted the part that did not sell is the part most likely to have been deleted.
The report says how many of the sales it timed had been relisted. That count covers the relists Cavri carried out, plus whatever figure you supplied in the relist column of a CSV import. A wardrobe pulled straight from Vinted brings no relist history at all, so an item you deleted and reposted by hand last spring arrives looking new and is counted as new. Read the line as a floor, not a total.
None of this needs Cavri. It needs a list of what you listed, what sold, and two dates per sale. The order matters, because the first two steps are where nearly every published Vinted sell-through figure has already gone wrong.
Two figures out of one list, and they disagree more often than not. A wardrobe that sells 70% of what it lists but takes eleven weeks to do it has a pricing question. One that sells 30% in nine days has a buying question. Neither number says that on its own, which is the reason for working out both.
Every one of these moves the figure in the flattering direction, which is why they survive. Nothing here is a criticism of the seller: they are properties of the measurement.
| What goes wrong | What it does to the figure | What to do about it |
|---|---|---|
| A relist restarts the clock | Days to sell runs from the current listing date, and a relist replaces it. An item that waited 90 days can report six. Nothing sold faster: the clock started again. | Count the relisted sales and show the count in the same line as the median. Cavri puts it in the sample line under the table for exactly this reason. It qualifies the median and nothing else: what the delete did to your denominator is the fourth row. |
| A mean where a median belongs | One item that sat for a year drags an average past anything in the sample, so the figure describes a listing nobody was going to buy at that price rather than the wardrobe. | Sort and take the middle. Every figure in Cavri's sell-through table is a median, and the one mean on the Stats page, average days to sell, will not print where the median beside it is being withheld. |
| A window short enough to flatter | A 30-day window on a department that sells three items a month reports on three items, and one slow month halves it. Shortening the window until the number improves is the easiest mistake here to make by accident. | Fix the window first and keep it the same between readings. Cavri's default is 365 days and is deliberately not wired to the period buttons sitting above it on the same screen. |
| A denominator missing what you deleted | Every unsold item you removed leaves the calculation, so the rate climbs as you tidy up. Housekeeping reads back as demand. | Keep a record that survives the delete. Cavri numbers items itself and that number outlives a delete and repost, so one garment stays one row however many listings it has had. That is what makes your own denominator reconstructable. Cavri does not build it for you, because it prints no rate. |
| An imported wardrobe reading its own history | The importer copies Vinted's real publication dates, so a wardrobe brought in this morning can arrive a year old, and its first reading covers months no tool was watching. | Read the first months as history rather than as a measurement of anything you changed, and treat the relisted count as a floor until the wardrobe has a few months of its own behind it. |
The first row and the last are the same date field seen from two ends, and that field has caused three separate bugs here, on 15, 16 and 26 August 2026. The listing date is right for anything that describes a listing, which is what these medians do, and wrong for anything that acts on one: a rule that relists by age has to run on when Cavri first saw the item, not on a date Vinted supplied about a listing Cavri never watched. Both clocks exist in the code, and they are kept apart on purpose.
There is no market-wide figure on this page. Cavri holds what its own sellers recorded about their own wardrobes and nothing else: no view of what anybody else charged, no list of what sold across Vinted last month, no table of rates by brand. Tools that read sold listings at scale can publish a brand-level number, and that is a different kind of company from this one.
The rates that circulate for Vinted do not carry a method. A sneaker figure quoted to one decimal place, an average of four point two days to sell, a claim that better pricing lifted sell-through from one percentage to another: none of them states a window, a sample size, or what was done about deleted listings, and several sit on the sites of tools the number flatters. A percentage without those three things is decoration.
Cavri sells a relisting tool, and the first row of the table above makes that tool look better than it is on Cavri's own screen. A relist shortens the measured wait whether or not it shortened the real one. That is why the sell-through report declares how many of the sales it timed had been relisted, beside the number rather than in a footnote, and it is worth knowing before reading any vendor's figures, this one included.
Sales with no listing date, and the handful whose two dates disagree because a relist landed after the sale had synced, are left out of the medians and counted in the open. The sample line says how many sold in the window, how many could be timed, and how many could not. A number that quietly drops what it cannot handle is worse than one that says how much it dropped.
Vinted names the parameters its feed and search take into account, and two of them decide whether a faster measured sale was a real one. Cavri does not restate anything that belongs to somebody else and moves: the links go to the copy that is kept current.
There is no published figure worth quoting. The ones circulating carry no window, no sample size and no account of what happened to deleted listings, so they cannot be compared with anything you would work out yourself. The answerable version is your own figure against your own figure a quarter ago, over a window you fixed in advance and did not move.
Items sold inside a window, divided by items listed inside that same window, as a percentage. The whole difficulty is the denominator: it has to include what is still unsold and what you took down, and on Vinted a relist takes a listing down permanently. Days to sell is a separate calculation off the same list: the gap between the two dates on every sale, sorted, and the middle one taken.
Yes, and it does it twice. Days to sell is measured from the current listing date, so a relist restarts the clock and an item that waited three months can report a week. The rate rises at the same time, because the deleted listing has left the denominator. Whether the relist made anything sell sooner is a separate question, and it is not answered by a figure the relist itself moved.
Because a middle number drawn from four sales moves every time a fifth one lands, and a figure that unstable is worse than a blank. Ten timed sales in a department or a price band is where the row starts printing one. Under that it shows how many it has and stops there. The ten is not adjustable and cannot be passed in as a parameter, which is the whole point of having it.
Because the import copies Vinted's own publication dates, so the listings arrive with their real ages on them. That is correct for describing a listing, and it means the first readings cover months before Cavri was there. Any relist you did by hand before the import is invisible too, so the relisted count on a freshly imported wardrobe is a floor rather than a total.
By the asking price. It is the number you chose and the only one you can change on the next item. Cavri bands sales on the listed price and falls back to the sold price only where there is no listed price to use, which happens on imported items and on drafts marked sold by hand. A sale with neither sits in no band rather than in the cheapest one.
Measuring it is one half. The other half is what Vinted publishes about its own ranking, and what it does not.
Free to start, no card. The Stats page is on every plan: median days to sell by department and by price band, with the sample each figure is drawn from printed underneath it.