The scoring fields and weights
The rubric is five fields and twenty-four points, and every point comes from something already written down on the platform ledger with a date attached. Nothing in it is a judgement about service quality, price or reliability, because this site has not measured those things for any platform. What follows is a way of sorting records, not a way of recommending companies. The tally is a count of recorded facts, not a mark out of twenty-four for the platform itself.
Still accepting orders carries eight points, which is the largest single weight and the closest thing here to a gate. A record either reads as operating or it does not, and the ledger takes that from the platform public pages it reads rather than from a comparison table. A tally can be high and useless at the same time, which is why the first field is applied before the arithmetic rather than inside it. Nothing in the arithmetic rewards a platform for being well known or for appearing in more lists.
Recorded as accepting a rebuilt source address is worth six. This is the field that decides whether an old sheet is worth carrying anywhere, since a platform that will not take a rebuilt marketplace address turns every row into manual re-entry. The ledger records a yes, a no, a partial or a still-unverified value, and the unverified cases count as zero rather than as a soft yes. Counting an unknown as a yes is how comparison tables drift away from anything checkable.
Three smaller fields fill the rest of the scale. An object page in the directory this ledger reads is worth five, a check date inside the current window is worth four, and a fee or storage figure that this site has read on the platform itself is worth one. Two of those three do no separating work at all this period, and that is worth stating plainly rather than hiding in a footnote. The smallest weight does not mean fees matter least; it reflects how rarely that value can be filled in.
The ranked list
The list below is gated rather than scored end to end. A platform that fails the first field is set aside whatever its tally, because a closed record and an operating record are not two points on one line. Within each group the names are alphabetical, which is the only ordering that adds no claim beyond the fields themselves. Nothing about a position says a platform handles parcels well, prices fairly or answers messages quickly, and a reader who needs those questions answered has to go and ask the platform.
Position one holds nine names: Eastmallbuy, Joyagoo, Kakobuy, Litbuy, Loongbuy, Lovegobuy, Oopbuy, Sugargoo and Superbuy. Each reads as operating, each is recorded as accepting a rebuilt source address, and each carries an object page in the directory this ledger reads. That combination is worth twenty-three of the twenty-four points, and the single missing point is the one nobody earned. They sit together rather than in an order because the four fields give no basis for separating them. A reader can check each of the nine against the ledger in a few minutes and reach the same tally.
The next group holds CNFans and Hoobuy, both recorded as operating and both accepting a rebuilt address, with no object page in the directory this ledger reads. They score eighteen. Hipobuy sits behind them with twelve: it reads as operating, its handling of rebuilt addresses is still unverified, so the second gate is not passed and the tally stops early. A platform with an unknown answer is not the same as one with a negative answer, and this list keeps those two apart. The gap between eighteen and twelve is one unread field and nothing else.
Then come the records that are out of the running on the first field. ACBuy and AllChinaBuy are closed, PandaBuy is closed and kept as a historical reference, and Mulebuy is recorded as a legacy entry. The remaining six names carry an unverified status, which means this ledger has no reading for them rather than a poor one. ACBuy and AllChinaBuy both carry directory object pages, which shows what that fifth field is really worth: coverage in one directory, and coverage is not standing.
Notes on each position
This section walks the groups in the order they appear above, and its purpose is to say what each position does and does not establish. Where a public comparison page describes a figure, it is named as a description this site read elsewhere, not as a number this site measured. The distinction matters more here than anywhere else on the site, because fees and storage are the two fields readers ask for most and the two this ledger holds no verified figure for. A reader who wants the original wording can follow the source list on the ledger page.
For the nine operating names with directory pages, the fields that separate them from the rest are coverage and address handling, nothing more. Their fee terms, storage days and payment methods are recorded as not verified by this ledger, for every one of them without exception. One recorded extra is worth naming: free inspection photos appear as a yes for Kakobuy and Sugargoo and as a partial for Superbuy, and that field is left out of the twenty-four because it can differ between accounts on the same platform.
For CNFans and Hoobuy the missing piece is coverage rather than conduct: both are described in public comparisons this ledger cites, and neither appears with an object page in the directory it reads. Public descriptions also attach figures to both names, including a service fee claim for Hoobuy and shorter free storage for CNFans, and neither figure was read on the platform itself, so neither is repeated here as fact. A claim with a source is still a claim, and it belongs in a sentence with that source attached.
The closed and unverified records need the shortest note. ACBuy stopped taking orders on 2026-06-12, a restart announced for 2026-07-01 did not restore ordering, and by August the storefront was down; AllChinaBuy is the same operation under a second name. PandaBuy closed in 2024 and stays in the list as a reference point. The six unverified names are simply unread by this ledger, and none of them belongs in either column on the strength of a list it appears in. Waiting for a reading is a position a record can hold honestly.
Where readers disagree
The first disagreement is about weights. Eight points for still accepting orders is a choice, and a reader who cares mostly about how long an item can sit in a warehouse would weight storage far higher. That objection is fair and also unanswerable this period, because storage terms are recorded as not verified for every platform in the ledger, so there is no field to weight. A weight attached to an empty column changes the arithmetic and not the knowledge. Weighting is the easiest part of a comparison to change and the hardest to defend.
The second disagreement is about the gate. Some readers want the closed platforms left in the ranking with a low tally, arguing that a full table is more informative than a filtered one. The counterargument is the one used here: a closed record and an operating record are different kinds of entry, and putting them on one scale invites exactly the comparison the fields cannot support. Sorting by tally alone would place a closed name above a live one, which is worse than an incomplete list.
The third is about the check date, which is identical across the ledger: every record was read on 2026-09-29, so four points land on all twenty-two names and separate nothing. Readers point out, correctly, that a re-check tomorrow would split them, and that the column therefore measures the age of the ledger rather than the freshness of any platform. The column stays anyway, because a field that fails to discriminate still tells you the whole table came from one afternoon. The checking method page describes how a check is run and what it refuses to conclude.
The last disagreement is about alphabetical order inside a group. Somebody will always want a tiebreak, and any tiebreak invented here would be a quality judgement wearing the clothes of a number. Where readers want more, the honest route is a new field that can be read and dated, added to the platform ledger page, rather than a reweighting of the four that already exist. Until then, ties stay ties, because two names that cannot be separated by a checkable field are two names that cannot be separated.