The numbers on a listing, and which two you have to read together
A listing card and a product page already print delivery speed, completion, disputes and ticket rate before you spend anything. One of them is an average taken over completed orders only, which makes it useless on its own and informative next to the one beside it.
Every listing on this marketplace arrives with its own track record attached, and most of it is on screen before you click anything. The useful part is not that the numbers exist. It is that two of them only mean something when you read them as a pair, and one of them is named in a way that invites the wrong conclusion.
What a card tells you before you open it
Each card in a shelf carries four facts: how much stock is left, how many units have sold, the warranty window in hours, and whether delivery is automatic. Many also show the share of buyers who came back. None of that requires opening the listing, which means a shelf can be filtered down by eye long before you are committed to anything.
The catalogue those cards sit in currently holds 24,749 active listings across 694 active shops and 39 categories. Of the active listings, 16,926 deliver automatically, 4,671 are handled by a person, and 3,152 draw from a held inventory. The twelve hour warranty is the single most common setting, on 15,287 of them.
The product page prints the same statistics twice, at two scopes
Open a listing and the same four measures appear in two columns. The left one counts only this product's orders. The right one counts the supplier's orders across everything they sell. Each column prints its own sample size, and those sizes are usually very different: a new listing from an established seller might show a few dozen orders beside a few thousand.
Read the narrow column for the thing you are actually buying and the wide one for the person you are buying it from. A strong supplier column under a thin product column means the seller is known but this particular listing is not yet proven. The reverse, a good product record under a weak supplier record, is the one worth slowing down for.
Average delivery is taken over completed orders only
This is the reading trap. The delivery average is computed across orders that reached completion. An order that stalled and never completed contributes nothing to it, which means the average cannot get worse because of the orders that went worst. On its own it will flatter almost everyone.
Completion rate is where those orders surface instead. So the pair has to be read together: a fast average next to a high completion rate is a real signal, and the same fast average next to a visibly lower completion rate is telling you that the quick deliveries were the ones that finished and you are not seeing the others.
Ticket rate is a rate, not a score
The ticket figure counts orders where the buyer opened the first message of a support thread, divided by the same sample. It measures how often buyers needed to ask for something, not how well the seller answered. A seller who replies quickly and fixes everything still accumulates it, and a seller nobody can be bothered to contact stays low.
Treat it as a question rather than a verdict. A few percent on a large sample is ordinary. A high rate on a small sample may be two awkward orders and nothing more, which is why the sample size is printed next to it.
What the sample actually contains
All four measures run over the same list: orders whose payment completed. Carts that were abandoned and payments that never cleared are not in the denominator, so none of these percentages is diluted by people who never bought. It also means a listing with no paid orders yet shows nothing at all rather than a flattering zero.
What these numbers cannot tell you
They describe orders that already happened. They do not describe the batch sitting in stock right now, and no statistic on the page can. A seller with a clean record can still receive a bad batch from upstream, and a tag on a listing describes the moment of sale rather than any promise about next week.
That is the reason the after sales window is anchored to delivery rather than to purchase, and the reason a dispute requires at least one screenshot before it will open. Checking the account in the first minutes after it arrives is worth more than any amount of pre purchase research, because it is the only step that happens while you still have a window open.
A short order of operations
Filter the shelf on the facts printed on the cards. Open the two or three that survive and compare the narrow column against the wide one. Read delivery speed and completion rate as one fact rather than two. Check the warranty hours before paying, not after, because that number decides how long you have. Then test the account immediately, and screenshot it before you change anything.
