How do I know a DHgate seller is reliable?
Direct answer
Not from the percentage. A rating summarises completed transactions, and a high figure is compatible with slow dispatch, substituted items and disputes that were settled before they were recorded anywhere. Read weighted signals instead: how long the storefront has traded, how recent the reviews are, whether the same item is still listed, and what written records say.
Why
- The percentage measures completion rather than quality. An order that ends in a partial refund after a long exchange can still count as a completed transaction, and a buyer who never opened a dispute leaves no trace in the figure at all. A storefront with a high percentage and no item-level detail is therefore compatible with a wide range of outcomes, including ones that would not repeat.
- Volume changes what the number means. The same percentage on a storefront with a few hundred transactions and on one with tens of thousands implies different things, because the larger figure averages over a period during which the build, the materials and the staff can all have changed. Recency carries more information than the total, which is why the weighted practice reads the last few months before it reads anything else. A storefront that has traded the same line for two years and one that changed its product mix last month are not comparable at the same figure, and a single percentage cannot tell the two apart or show which one is being read.
- The signals that carry weight are the checkable ones. How long the storefront has carried the same item, whether the photographs have changed between visits, whether the description names materials, and whether a specific question receives a specific written answer. Each of those can be verified twice, which is the property a signal needs before it can be relied on, and a percentage has never had it.
- Written records beat impressions. This site keeps dated roster entries describing what a storefront has been observed to do, with the evidence level and the review date attached, and it withdraws them when the conditions change. Reading a dated record tells you when the observation was made and what it covered, which is information no aggregate percentage ever carries and no storefront page displays. A record also states what it does not cover, which a percentage never does. Knowing the date of an observation, the category it concerned and the evidence level behind it is what lets a reader decide how much of it applies to the order in front of them.
When this answer flips
- Reviews from the last few months agree with the older ones and the photographs have not changed, because the storefront is then stable enough for its history to carry weight.
- The item you want is a new line for that seller, because the storefront record describes earlier products and says little about the build now on offer.
- The seller answers specific questions with specific written answers before payment, because that is a stronger signal than any aggregate figure and it is usable in a claim.
Three things to do next
Sources and checks
- Platform published rule — Published definition of the seller rating, read to establish exactly which transactions are counted into the percentage and which outcomes leave no mark on itno sample
- Public community thread — Cases where a high-rated storefront produced a substitution or a slow dispatch, read for which observable signals had already flagged the storefront beforehandn=52
- Editor analysis — Signal weighting set by how many times a signal can be independently re-checked, rather than by how easily it can be read from a storefront page in one visitno sample
Last checked · next review due · status current.