> ## Documentation Index
> Fetch the complete documentation index at: https://docs.regasolutions.io/llms.txt
> Use this file to discover all available pages before exploring further.
# Following what sells
> What sales tracking learns on Vestiaire, how quickly it learns it, why a withdrawn listing is never a sale here, and why the two marketplaces never share an item's number.
On Vinted, a sale has to be **reconstructed**: an item that sold simply disappears, and we wait for two absences in a row before daring to conclude. On Vestiaire the marketplace **says it**, and it gives the exact moment. That single difference governs this whole page: here, the time an item took to sell is a **measured** duration, not a bracketed one. We watched one coat leave in **398 seconds** and one dress take **3,886,847** — six minutes and thirty-eight seconds, against forty-four days and twenty-three hours.
***
## What “sold” means here, and why it is not the Vinted word
Three things separate the two marketplaces, and all three work in your favour on Vestiaire.
| | On Vinted | On Vestiaire |
| ----------------------------- | ------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------- |
| **How a sale is established** | It is reconstructed: the listing vanishes, and it takes **two consecutive absences** before we conclude | The marketplace states it: the listing's page stays up and carries a “sold” marker |
| **Sold, or withdrawn?** | An absent listing is “sold **or** withdrawn”, and nothing separates the two | **Sold** and **withdrawn** are two different answers |
| **The hour of the sale** | The hour of **our own observation** | The hour of **the sale itself**, given by the marketplace |
There are exactly three answers, and we measured them against three real items.
| What the marketplace answers | What it means | What REGA files |
| ---------------------------------------------------------------------------------------------- | ----------------- | ------------------------------------------------------ |
| The page is still served, carries a “sold” marker and a sale date | **Sold** | A sale — and an alert, if you asked the filter for one |
| The page is still served, without that marker | **Still on sale** | Nothing yet: the item stays under watch |
| The page is gone, and the marketplace answers that the product is no longer available for sale | **Withdrawn** | A withdrawal, deliberately kept **out** of your sales |
**A vanished listing is never a sale here.** A sold item keeps its page; a withdrawn one loses it. Filing a disappearance as a sale would inflate every figure on this page — the sale count, the fees, the times to sell — with items nobody ever bought.
**One reading is enough to conclude.** The two-absences rule of Vinted does not apply on Vestiaire, and this is not a relaxation of the standard: on Vinted an absence is a **silence**, and a silence has to be repeated before it means anything. Here the absence is a **sentence** — the marketplace tells us the product is no longer on sale. One sentence, one verdict.
**What never concludes, whatever we read.** A request to slow down, a server error, a dropped connection, an answer we cannot parse: none of these is a verdict. The check returns “unknown”, nothing is written, and it will be tried again. The reasoning behind that rule is worth keeping: treating a slow-down request as a withdrawal would wipe living items off your sales page the first time the marketplace asked us to ease off.
***
## What one Vestiaire sale tells you, and nowhere else does
Five things come back from a single reading, and none of them spends one of your accounts.
| What you learn | Why it is solid |
| -------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **The exact moment of the sale** | It comes from the marketplace, not from us. Our checking rhythm therefore never alters the figure — only the delay before you see it |
| **The time to sale** | Listed → sold, **both instants coming from the marketplace**. It is a measured duration, not one bracketed between two of our passes |
| **The favourites at the moment of the sale** | The count as it stood when the sale was recorded — the closest thing to a measure of how contested the item was |
| **The buyer fees** | The price paid, minus the price excluding fees — both read from the marketplace, never estimated. When a listing carries no price excluding fees, the figure stays **absent** rather than being written as a zero: one of the two witnesses below is in that case |
| **The address of the listing** | Only that same reading gives it. Without it, a sale card would stay without a link, for good |
### The two measured witnesses
Both are real items.
| Item | Listed | Sold | Time to sale |
| ---------------------- | ------------------------ | ------------------------ | -------------------------------------- |
| A Dior dress | 18 March 2026 | 2 May 2026 | **3,886,847 s** — 44 days, 23 h 40 min |
| A Dolce & Gabbana coat | 18 August 2026, 14:53:18 | 18 August 2026, 14:59:56 | **398 s** — 6 min 38 s |
That spread is the point. On this marketplace the same filter can hold an item for a month and a half, and lose another in the time it takes to read a notification.
**We do not print a typical time to sale in this guide, and that is deliberate.** Two medians exist in our own measurements, they disagree with each other, and the better documented of the two **declares itself censored**: the tracking had been running for under twelve hours when it was computed, so its low end was cut off. Publishing either would be handing you a promise instead of a measurement. The median you should read is the one your own dashboard computes, on your own tracked items — see further down.
### The two prices on a sold card, and the trap between them
A sold Vestiaire item can carry **two** prices, and the lower one is not a discount. Measured on the Dior dress: **4,747.11 €** paid against a second figure of **4,380.00 €**. The second is the price **excluding buyer fees** — what the seller takes home. The gap between them is what the marketplace charged the buyer, on top of the seller's price.
Not every listing carries both. On the Dolce & Gabbana coat, the marketplace served **no price excluding fees at all** — so REGA writes no buyer fee for that sale. An absent figure means **not observed**; a zero would mean **no fee was charged**, which is a different claim and one we have not measured.
Read that second number as a “before the discount” price and every margin you compute will be wrong, in the same direction, every time. The screen says it in its own words on the fees figure: `The marketplace charged this amount to the buyer ON TOP of the seller's price ({hors}). This is not a discount.`
***
## How an item enters tracking
Sales tracking is **inert on Vestiaire until you ask for it**. A filter can detect perfectly, day after day, and never produce a single sale: detecting and tracking are two separate gestures, exactly as they are on Vinted.
**Sales** is one of the entries in the left-hand menu, and the drawer sits at the right end of its action row. That drawer is where a filter is put under tracking, removed from it, given its alert, or has its intake of new listings suspended and resumed. Nothing else does it for you — not the Filters screen, not the dashboard.
From that moment, everything the filter keeps joins the watch **as it is born**. This is the normal path, and it is immediate.
Bringing a Vestiaire filter's existing detections into tracking **costs nothing, and slows nothing down**: the radar has already done the work, once, for every client at the same time, and filed what it found.
A sale is a market fact, not a live feed. What the screen is worth reading for — what leaves, at what price, after how long — needs a day or two of intake before it says anything.
**The drawer holds the filters of both marketplaces, in a single list, and writes the marketplace beside each name.** Once one is chosen, the closed button reads `Vestiaire · `. That matters more than it looks: the same name often exists on both sides — “Dior Clothes” is a real case — and they are two different filters, watching two different markets.
The selector at the top of the screen, the one that says which tracked filter you are reading, does **not** carry the marketplace: it shows the name alone. Two filters named alike are told apart when you add them, not when you switch between them.
**Tracking does not reach back into the past of a filter, and this is an accepted limit.** The depth available is that of the detections already made, never that of the marketplace's catalogue. What the filter had already detected **does** enter, and it enters with the hour the radar actually saw it, not with the hour you asked. What never reaches you is what the filter had not yet detected: on Vinted that gap was filled by paging deep into the catalogue, and here that route does not exist, because there is no page of the marketplace that would give it back to us.
Two consequences worth holding together: the deepest re-read is **10,000 detections**, and detections themselves are **wiped after 48 hours**. Put a filter under tracking the day you create it, not the week after.
The same drawer carries that filter's **alert**. A Vestiaire sale can notify you exactly as a Vinted one does — with one difference, and it is on purpose: the notification carries no link to the item. The reason is the next section.
***
## How long before you know
### The rhythm of the checks
| Setting | Value |
| -------------------------------------------------------- | --------------------------------------- |
| A freshly detected item, in its **first half-hour** | **six readings** |
| An item **less than 24 h old** is re-read at least every | **10 minutes** |
| An item **more than 24 h old** | **1 hour** |
| Tracking horizon on Vestiaire | **7 days** — against 48 hours on Vinted |
The ten minutes and the hour are **ceilings, not the rhythm**: they say how long an item may wait, never how fast it is actually read. A new item is read far faster than that — because a new item is where a sale can happen in six minutes.
**There is no tracking horizon for you to set, and that is deliberate.** Tracking has one window and one only; putting a choice on screen that the server would immediately overrule would be a setting in name alone. What varies is the marketplace: a Vestiaire filter gets seven days where a Vinted one gets twenty-four hours.
Seven days exists because of a measurement — with a 24-hour horizon, **7.5 % of tracked Vestiaire items** were dropped before they had ever given a verdict. It narrows that pocket rather than closing it: sales on this marketplace have been measured as slow as **forty-five days**, so items that outlive a week still exist, and we have not yet measured how many.
An item that has sold, or gone, is **frozen**: we stop reading it. Nothing is spent on an answer that will never change again.
Checking a Vestiaire item **uses none of your accounts**, on either marketplace: your connected accounts are spent on offers and on purchases, never on watching. That is what lets the rhythm above be as tight as it is.
### The lag between the sale and our reading — and why it costs you nothing
This is the measurement that justifies the whole design. On the first thirteen sales recorded, the lag between the sale itself and our observation of it ran from **2,111 seconds to 26,880 seconds** — up to **seven hours**.
And the number of sales missed because of that lag: **none**. Zero disagreements across **70 items** queried directly at the marketplace to check.
**A lag on our side does not damage the data, because the marketplace holds the clock.** The sale's hour comes from Vestiaire; our reading only decides **when you find out**, never **what you find out**.
The same lag has two completely different prices depending on the marketplace, and it is worth seeing them side by side.
The hour of the sale **is** the hour we noticed it — there is no other source. A check that runs late therefore writes a late hour into your history, and nothing will ever correct it afterwards. And the verdict itself waits: two consecutive absences are needed before a departure is even filed.
The hour of the sale comes from the marketplace, with the listing's own date beside it. A late check costs you **the delay before the card appears**, and nothing else: the duration written on it is right either way. One reading is enough to file the verdict.
```mermaid theme={null}
flowchart TD
A["The radar keeps a listing
on one of your Vestiaire filters"] --> B{"Is that filter
under tracking?"}
B -->|no| X["Nothing is ever re-read.
Its sale will not exist for REGA"]
B -->|yes| C["It joins the watch at birth,
at no network cost"]
C --> D["Re-read six times in its first
half-hour, then at least every
10 min up to 24 h, then hourly"]
D --> E{"What does the
marketplace answer?"}
E -->|"still on sale"| D
E -->|"an error, a slow-down,
an unreadable answer"| D
E -->|"the page is gone"| W["WITHDRAWN — filed,
and kept out of your sales"]
E -->|"the page carries
a sale date"| S["SOLD — one reading is enough.
The time to sale is measured"]
style S fill:#1f6f43,color:#fff
style X fill:#7a4b00,color:#fff
```
***
## Two marketplaces, one numbering space
This is the rule that explains every oddity on this page, and it took three separate defects before anyone wrote it down.
A Vestiaire product carries a number in the **seventy millions** — we measured **65,087,726** and **70,070,945**. Vinted handed out numbers in that very range **around 2016**. A Vinted listing from 2016 that is still online can therefore carry **exactly the same number** as a Vestiaire product on sale today.
| What would have happened | Why |
| -------------------------------------------------------------- | -------------------------------------------------------------------------------- |
| Tracking one would have **overwritten the other** | The watchlist was indexed on the bare number |
| One of the two would have **never been checked** | Two items sharing a number looked like one item asked for twice |
| A Vestiaire sale would have **closed a living Vinted listing** | “Every copy of this item” did not carry the marketplace, on screen or in storage |
The fix is quiet and you never see it: every item is now filed under a key that **carries its marketplace**, and the marketplace is read back from that key rather than kept in a second field beside it — because two fields that *can* contradict each other eventually do.
What you do see:
* A **marketplace selector** on the Sales screen — three chips carrying their counts, “Both”, “Vinted” and “Vestiaire” — which **disappears when you only have one marketplace**, rather than offering a choice with one option.
* A chip on every sale, carrying the marketplace's name in full — “Vinted” or “Vestiaire”. The name is written out, not just coloured: a colour alone is unreadable in greyscale and to anyone who does not separate green from amber. Hover it and the tooltip spells the sentence: `Listing seen on {place}`.
**A Vestiaire sale notification does not open the listing, and that is on purpose.** It carries no item number, because a number of seventy million is a valid Vinted number too: the tap would open **a different item, on the other marketplace**, with a believable title and a believable price. The notification keeps its card — title, brand, price, photo — and takes you to the Sales screen instead.
***
## What the Sales screen shows for Vestiaire
The screen is shared between the two marketplaces; these are the parts that are specific to this one.
| Where | What you read |
| ----------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Above the list | Three chips with their counts: “Both”, “Vinted”, “Vestiaire”. Absent if you have only one marketplace. Their tooltips read `Sales from both marketplaces` and `Only sales seen on {place}` |
| On every sale | A chip bearing the marketplace's name — “Vinted” or “Vestiaire”. Its tooltip reads `Listing seen on {place}` |
| On a sale's card, in the list | `sold in {d}` — and that one figure only |
| Once you open that sale | `sold in {d}` · `{n} favourites` · `buyer fees {m}`. The last two live in the detail panel, never on the card |
| On the dashboard view | `median sale {d} ({n})`, `buyer fees {v} in total`, and the median number of favourites beside them |
The tooltips are where the care shows. Hover them once and you will not need to come back:
| Figure | What its tooltip says |
| ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| `sold in {d}` | `Sold {d} after being listed. Both dates come from the marketplace: this duration is measured and does not depend on our check interval.` |
| `{n} favourites` | `Number of favourites at the moment the sale was recorded` |
| `buyer fees {m}` | `The marketplace charged this amount to the buyer ON TOP of the seller's price ({hors}). This is not a discount.` |
| `median sale {d} ({n})` | `From listing to sale, both dates coming from the marketplace itself: this duration is measured, not bracketed. Not to be confused with the neighbouring median, which is an upper bound.` |
**Look for the missing ≤ sign.** The neighbouring medians on that dashboard carry a “≤” because they bracket our ignorance — they are upper bounds on a duration we only partly observed. The Vestiaire sale median has **no such sign**, and it is not an oversight: both of its dates come from the marketplace, so it is a measurement. When a figure in this product wears a “≤”, it is telling you it is a ceiling, not a value.
***
## The watchlist, on this marketplace
The watchlist works the same way on both marketplaces — you set an item aside, and REGA re-reads it for you. Three things are different here.
### Two extra figures, and they cost nothing
A watched Vestiaire item carries, on top of everything a Vinted one carries:
* the **price excluding fees** — what the seller takes home;
* the **buyer fees** — what the marketplace takes from you.
The price displayed is their **sum**: what you pay for the item itself. Neither figure costs anything extra — both arrive with the reading we were doing anyway. They are **absent on Vinted**, and that is correct rather than missing: the concept does not exist there in this shape.
Serving those two figures separately closes a gap that was, at one point, visible **nowhere**: an item displayed at **34 €** came to **102.12 €** in the basket, once the marketplace's delivery and duties were added on top. The watchlist tells you what the item costs; it does not tell you what the order costs. That number is broken down line by line on **[Buying on Vestiaire](/vestiaire/buying)**, and it is worth reading before your first purchase.
### Buying and offering are closed on a Vestiaire line
The two gestures that spend money — buying, and making an offer — are not offered on a watched Vestiaire item, and their absence is what protects you. The number on that line is a valid **Vinted** number too, so a purchase started from there would run a Vinted checkout, on the Vinted item carrying the same number: a real charge, on the wrong item, from the wrong account. And the rule is written as a permission rather than as a ban — a third marketplace, the day one is added, would get no button either.
To buy a Vestiaire item, or to make an offer on it, go through the **dashboard**: its card carries “BUY”, and its ⋯ menu carries “Make an offer on Vestiaire”. The watchlist is where you keep an item and follow its price; the feed is where you act on it.
### Watching an item does not favourite it on Vestiaire
On Vinted, adding an item to the watchlist can set the favourite for you. On Vestiaire it does not, and it must not: the gesture would send a seventy-million number to **Vinted** — at best it lands nowhere, at worst it puts a like on an unrelated Vinted listing, from one of your real accounts, on a marketplace that counts robot-shaped signals.
The favourite on Vestiaire is a **separate gesture**, made on the dashboard: the ⋯ menu of a Vestiaire card, and the item's panel, both carry “Add to Vestiaire favourites”. Its tooltip says the rest: `Hearts the listing from your Vestiaire account. No effect if it was already there, and it cannot be removed from rega.` Setting it twice leaves one favourite, not two — and it comes off **on Vestiaire**, never from here.
***
## Our tips
Back-filling reaches exactly as far as the detections we still hold. Those are wiped after 48 hours, and the deepest re-read is 10,000 of them — so a filter put under tracking a week late starts from an empty past, silently.
**The move**: create the filter, check its hit counter an hour later, and if it is producing, open the “Tracked filters” drawer the same day.
The duration on a sale card is the marketplace's own arithmetic between two of its own dates. Our rhythm — ten minutes, then an hour — decides when the card appears, not what it says.
**The move**: when you compare two items, compare their “sold in”. Comparing the moments they turned up on your screen compares our schedule, not the market.
Here, a gone listing means **withdrawn** — the seller pulled it, or the marketplace did. REGA files it apart, on purpose, and it never enters your sales figures.
**The move**: if a find disappears without a sale, do not count it as demand. It is the one signal on this marketplace that means nothing at all about price.
The lower of the two prices on a sold card is the **seller's** price. The gap is what the buyer paid on top, to Vestiaire.
**The move**: when you work out what an item resells for, work from the price paid. When you work out what a seller would accept, work from the price excluding fees.
Adding an item to the watchlist does not favourite it on Vestiaire. Nothing warns you, because nothing failed — the gesture simply is not attempted.
**The move**: on an item you actually want, use “Add to favourites” in its panel as well as the watchlist. And remember the favourite comes off on Vestiaire, not here.
On the merged view, Vinted sales and Vestiaire sales sit side by side and the two markets do not behave alike — one is measured in seconds, the other in weeks.
**The move**: split by marketplace before reading a median, a rhythm, or a brand ranking. The chip on each card — it carries the name, “Vinted” or “Vestiaire” — is the check that costs nothing.
Seven days is how long we keep reading a Vestiaire item, and it was chosen because 7.5 % of tracked items were still undecided after twenty-four hours. There is nothing for you to set here: the window is ours, and it is already the longest one tracking has.
**The move**: read a filter's numbers after a full week under tracking, not after an evening. And keep the tail in mind — an item still on sale after seven days leaves your tracking without a verdict, and we have measured sales that took forty-five days.
***
## What surprises people
Both are measured, on the same marketplace, in the same week: **398 seconds** for a coat, **3,886,847 seconds** for a dress. There is no typical item here, which is precisely why we refuse to print a typical figure.
Because the marketplace holds the clock. On the first thirteen sales recorded, our observation trailed the sale by up to **26,880 seconds** — and across **70 items** re-checked directly at the marketplace, our verdicts and its answers disagreed **zero times**.
On Vinted we need two absences because an absence is ambiguous. Here the marketplace answers in words. Waiting for a second identical sentence would add a delay and no certainty at all.
It has no item link, because the number it would use belongs to two marketplaces at once. It carries the card instead — title, brand, price, photo — and lands you on the Sales screen.
The watch uses **none of your connected accounts**, on either marketplace. They are spent on offers and on purchases — never on watching.
Measured: **4,747.11 €** paid, **4,380.00 €** for the seller. The second figure is the price before the marketplace's buyer fee, and reading it as a former price inverts the whole calculation.
The radar sweeps once for every client at the same time, and files what it finds. Bringing those detections into your tracking re-uses that work: it takes nothing from anyone, and it slows nothing down.
***
## What this does not do yet
We would rather write it here than let you find out halfway through.
Tracking reaches as deep as the detections already made — 10,000 at most, and they are wiped after 48 hours. A filter's past on the marketplace is not recoverable, because no page of Vestiaire would give it back.
We hold two medians, they disagree, and the better documented one declares itself censored. Until the distribution is rebuilt on a proper window, this guide prints individual measured durations and nothing else.
Deliberate: the identifier would open another item on the other marketplace. The card and the Sales screen are what you get.
On Vestiaire the favourite stays a separate, manual gesture — and it is removed on Vestiaire, never from here.
The Orders screen knows nothing of this marketplace. What you bought here is read on Vestiaire; REGA follows what is **on sale**, not what you have ordered.
The counts are there — a `{n} withdrawn by the seller` tile, a “Withdrawn” column in the brand table, and a share of the departures bar. What is missing is the **rate**, spelled out as a percentage: of the figures on that dashboard, only the indecision rate gets that treatment.
***
## Where to go next
The screen itself: the “Tracked filters” drawer, the two views, and the gestures that act on one filter only.
Setting an item aside, and what REGA allows itself to say about it — on both marketplaces.
Why there is no birth to catch here, and what a single shared sweep means for you.
What an item really costs once delivery and duties are on it — the 34 € that became 102.12 €.
Every criterion, what it costs, and what happens when two of them contradict each other.
Where Vestiaire detections land — in the same feed as Vinted, with the marketplace written on the card, and the panel where an offer or a purchase starts.