1291of3119
Below 3.5 across score observations.
A fixed-panel view of the supplier score fields buyers see on 1688
Issue 1 follows the same 120 product-detail records once per day. Across 26 completed days, the composite median was 4.5/5, while the logistics and after-sales fields produced very different below-3.5 shares. That gap tells buyers to open the component fields. It does not support a supplier verdict.
Liam Cai · Founder, SupplymoYiwu, ChinaWhat we measured
Logistics-score observations below 3.5 across Issue 1
These are repeated platform-returned score fields from a fixed panel. They are not audited supplier performance, order outcomes or a representative sample of 1688.
This is a fixed, non-random panel selected from the first page of 12 deliberately chosen English searches; it is not representative of all 1688 listings.
Study reading
One headline score does not describe every service component
The composite-service field had a median of 4.5/5 across 3,119 numeric observations. In the same panel, 1,291 logistics observations fell below 3.5. These fields should not be collapsed into one supplier verdict without knowing how the platform defines and weights them.
For a live order, ask for the package facts, dispatch terms and written supplier answer that matter to that shipment. A marketplace score is context; it is not the evidence file for your goods.
Study reading
The panel stays fixed so movement cannot be manufactured by reselection
The 120 internal offer IDs were selected once on 6 August 2026. The same set was queried on every day in this issue. New high-scoring listings were not swapped in, and low-scoring records were not removed from the denominator.
The trade-off is deliberate: a fixed panel can show how one defined shelf changes, but it cannot describe the entire marketplace. Future issues must keep that distinction visible.
Study reading
Availability has a narrower meaning than a listing being gone
No observation in this issue was explicitly marked offer_not_returned. If that status appears in a later issue, it will mean only that the read-only detail call did not return the offer on that run.
Collection errors are a separate state and never count as unavailable. This prevents a network or API failure from becoming a false delisting claim.
Use this data
Everything here is free to reuse with attribution under CC BY 4.0. If you need a cut of the data we have not published, ask and we will run it.
Quote it, or take the whole file
One line, free to quote with attribution
In a fixed, non-random Supplymo panel of 120 1688 product-detail records observed once daily from 2026-08-06 to 2026-08-31, the composite-service median across 3,119 returned score observations was 4.5/5; 1,291 of 3,119 logistics-score observations and 500 of 3,119 after-sales-score observations were below 3.5, while 0 of 3,120 daily observations were explicitly marked offer_not_returned by the collection contract.
How to cite this study
Liam Cai. "Supplymo 1688 Shelf Index — Issue 1." Supplymo, 2 September 2026, https://supplymo.com/research/1688-shelf-index.
Each public row aggregates one completed calendar-day run across the same 120 panel items. Offer and seller identifiers are not published, and the panel was not de-duplicated by seller. 10 stated limits travel inside the file. Free to reuse with attribution under CC BY 4.0.
Method
How the sample was collected
Everything here is reproducible from the evidence file and the sampling script.
Source
com.alibaba.fenxiao.crossborder:product.search.queryProductDetail (read-only)
Sampling frame
A fixed panel selected on 2026-08-06 from the first ten platform-ranked rows for each of 12 deliberately selected English searches. The same 120 offer IDs were queried once per calendar day in this issue window.
Collection window
2026-08-06 through 2026-08-31, Asia/Shanghai calendar days; one completed panel run per day.
Write operations
None. No order, payment, enquiry, messaging, refund or logistics write endpoint was called.
Check the figures yourself
Each public row aggregates one completed calendar-day run across the same 120 panel items. Offer and seller identifiers are not published, and the panel was not de-duplicated by seller.
- Rows
- 26
- Collected
- 2026-08-31
- Stated limits
- 10
Limits
The 10 limits of this sample
All 10 stated limitations, in the order the evidence file lists them. A number travels further than the caveat attached to it.
- 1This is a fixed, non-random panel selected from the first page of 12 deliberately chosen English searches; it is not representative of all 1688 listings.
- 2The 120 panel items were not de-duplicated by seller. They are not 120 distinct suppliers, factories, companies or ownership groups.
- 3The 3,120 private observations repeatedly measure the same 120 panel items across 26 days; they are not that many unique products.
- 4Scores are numeric fields returned by the platform. They are not independently audited service performance, warehouse evidence or buyer outcomes.
Show the other 6 limits+
- 5The 3.5 threshold is a descriptive comparison point chosen before calculation. It is not a supplier approval rule.
- 6An explicit offer_not_returned response is not proof of permanent delisting, stockout, business closure or unavailability to every buyer.
- 7Collection errors are kept separate from explicit non-returns and never counted as unavailable.
- 8A returned product-detail row can still omit one or more score fields; score denominators therefore travel with every figure.
- 9No order, checkout, payment, supplier contact, message, shipment or after-sales case was observed.
- 10Nothing here says 1688, a platform operator or any seller has done anything wrong.
Research boundary
What this study cannot tell you
- This is a fixed, non-random panel selected from the first page of 12 deliberately chosen English searches; it is not representative of all 1688 listings.
- The 120 panel items were not de-duplicated by seller. They are not 120 distinct suppliers, factories, companies or ownership groups.
- The 3,120 private observations repeatedly measure the same 120 panel items across 26 days; they are not that many unique products.
- Scores are numeric fields returned by the platform. They are not independently audited service performance, warehouse evidence or buyer outcomes.
- The 3.5 threshold is a descriptive comparison point chosen before calculation. It is not a supplier approval rule.
- An explicit offer_not_returned response is not proof of permanent delisting, stockout, business closure or unavailability to every buyer.
- Collection errors are kept separate from explicit non-returns and never counted as unavailable.
- A returned product-detail row can still omit one or more score fields; score denominators therefore travel with every figure.
- No order, checkout, payment, supplier contact, message, shipment or after-sales case was observed.
- Nothing here says 1688, a platform operator or any seller has done anything wrong.
What this means for a buyer
A sample of this kind is a starting point for questions, not a verdict on any one supplier. Check the exact product and supplier you are about to pay.
Supplier verification
A free check before you commit. We run supplier checks from Yiwu, and we do not promise a lowest price.
Check a supplier →Questions a reader should ask about this
What is the one-line result from Issue 1?
In a fixed, non-random Supplymo panel of 120 1688 product-detail records observed once daily from 2026-08-06 to 2026-08-31, the composite-service median across 3,119 returned score observations was 4.5/5; 1,291 of 3,119 logistics-score observations and 500 of 3,119 after-sales-score observations were below 3.5, while 0 of 3,120 daily observations were explicitly marked offer_not_returned by the collection contract. The score denominators and all stated limits are published with the data.
Is this a random sample of 120 1688 suppliers?
No. The panel contains 120 product-detail records selected from the first page of 12 deliberately chosen English searches. It was not de-duplicated by seller and cannot represent all of 1688.
Why publish daily aggregate rows instead of offer-level records?
The daily aggregates are enough to reproduce every public figure while keeping offer IDs, seller identity, shop names, prices and product details out of the downloadable dataset.
Does a score below 3.5 mean the supplier failed?
No. 3.5 is a descriptive threshold chosen for this issue. The fields are platform-returned display values, not Supplymo inspections, audited fulfilment outcomes or supplier approval decisions.
Does offer_not_returned mean a product was permanently delisted?
No. It means the product-detail call explicitly did not return that offer on that run. It cannot prove a permanent delisting, stockout, business closure or unavailability to every buyer.
How often will the Shelf Index update?
The plan is one issue per completed measurement window on this fixed URL. Each issue keeps its dates and evidence boundary; the panel is not silently reselected to improve the numbers.
Can this index replace a supplier check?
No. A panel trend cannot answer whether one product, supplier, MOQ, package, document set or shipping route is safe to approve. A supplier check looks at the exact product and supplier you are about to pay.