There is no "best seller" list on this site and there is not going to be one. Not out of caution, and not because we are hedging — because we hold 237 quoted comments across 124 sellers, and a ranking built on that would be measuring the wrong thing while looking like it measured the right one.
This page explains what the wrong thing is, and then gives you the method we use ourselves, so you can run it on any seller including the ones nobody here has written about.
What a seller ranking actually measures
Sort sellers by how positive their comments are and you have ranked them by how many happy customers happened to post. That number moves with a seller's popularity, with which subreddit they are discussed in, with whether they run giveaways, and with how long they have been around. It barely moves with whether your parcel arrives.
Worse, it is self-reinforcing. A seller near the top of a list gets more orders, more orders produce more posts, more posts produce more of whatever sentiment the list was already reading. Within a few months the list is measuring itself.
The arithmetic is unforgiving too. In our corpus, 94 of 124 sellers have three or fewer quoted comments, 53 have none at all, and 23 have a record that comes entirely from one thread. A grade computed on four comments is not a weak grade. It is a coin flip with a decimal point printed after it.
What this site does instead
Sellers are listed by how much the community talked about them. That is a description of the corpus, not a rating of anybody, and we will not convert it into one. Where a ranking does exist here it comes from community votes on individual items, every seller starts from the same neutral prior, and no seller's starting point is adjusted by anything we know or think about them. The whole mechanism, including the parts that are unflattering, is written out at the voting method.
The practical consequence: a seller with no standing on this site has not been marked down. Nothing has been asserted about them at all.
Seven questions to ask of a record
Open a seller's page, ignore the tone of it, and answer these.
- How many separate threads does the evidence come from? Six comments from one haul post is one person's experience quoted six times. Two comments from two unrelated threads is worth more than that.
- What is the newest date? Sellers change hands, get overloaded, or quietly stop answering. A glowing 2021 review describes a business that may no longer exist in the same form.
- Did the reviewer hold the item? "QC pics look great" and "I have worn this for six months" are different claims. So is a comment written before the parcel shipped.
- Does anything go wrong anywhere in the record? A file with no complaints in it is usually a small file, not a clean one. What you want is a record that contains a problem and shows what happened next.
- Can you actually see what they sell? A password-gated Yupoo or a catalogue with no marketplace links means you are buying on description rather than on inspection.
- Is the reviewer selling you something? Giveaway threads, agent-sponsored photo posts and seller self-introductions are all in the corpus, labelled by their permalink. They are evidence of marketing, not of quality.
- What would change your mind? If nothing in the record could have talked you out of the order, you were not evaluating. Decide before you read what a disqualifying fact would look like.
Read the thread, not the excerpt
Every quote on this site is truncated and every one carries a permalink for that reason. An excerpt cannot show you the reply that contradicted it, the downvotes, or the eight-comment argument underneath. The evidence we collected is the whole corpus with its sources attached; the honest use of it is as an index into Reddit, not as a substitute for reading Reddit.
Read the thread and you also get what no aggregate can carry — whether the person writing sounds like they know what a good seam looks like.
What no record can tell you
Consistency between orders. That is the thing this hobby actually turns on and it is the thing no quoted comment reaches, because each comment is one order from one batch on one day. A seller with ten good reviews has had ten good orders. It is not a promise about the eleventh.
Sizing is the same story: rep sizing does not track retail and does not track reliably within one seller, so a review that says an XL fitted well tells you about that garment and that body.
Then reduce what a wrong answer costs
Evaluation is only half of it. The other half is structuring the order so being wrong is survivable — one item rather than five on a first order, an agent whose QC photos you trust, and a real look at those photos before you release the parcel. How to buy reps walks through that sequence, and red flags that a rep seller is about to waste your money covers the behaviours worth aborting on.
If a link you were sent no longer opens, the link decoder will usually recover the item from the URL itself, which is often faster than asking the seller what happened to it.
None of this produces a verdict, and it is not meant to. It produces a defensible decision about one order, which is the only decision anybody here is actually making.