What does each column actually tell you?

Column headers as claims, not facts

A header is a promise about what a column contains, and the person who wrote the header is the same person who filled the column. “Weight” can mean a seller estimate, a warehouse figure, or a number copied from somebody’s forum post, and the header will not tell you which.

A practical test: pick three rows at random and ask what a wrong value in that column would cost you. A wrong date costs a wasted click. A wrong variant costs a return. Rank the columns by that figure and read them in that order.

Seller wording versus measured wording

Two kinds of wording sit in the same table and should never be read the same way. Seller wording is written to sell: brand names get used loosely and sizes get rounded up. Measured wording is written with a tape, repeats the same method every time, and does not change when the listing does.

Where a table mixes the two, separate them mentally. A row that says an item fits like a large is seller wording. A row that says 58 cm across the chest is measured wording. Only the second can be checked against a garment you already own.

Numeric columns worth trusting first

Numbers with a stated method are the most useful thing in any of these tables. Our own tally pairs every figure with a run count: 38 garment rows measured flat on three units of each model, and 46 footwear rows measured for insole length and widest point on three pairs each. The run count is what makes a number arguable rather than decorative.

Where a numeric column has no method attached, treat it as a ranking rather than a measurement. It can still tell you which of two rows is heavier or longer, which is often all you need, so long as you never carry the figure into a calculation of your own.

Free-text columns and the drift inside them

Notes columns hold whatever did not fit anywhere else, which makes them the most human and the least consistent column in the table. One author writes “thin material”, another writes “see photos”, a third writes nothing at all. Comparing two rows through a notes column is closer to comparing two opinions than two facts.

The drift is not dishonesty, it is what happens to a column with no defined input. Read notes for warnings about a specific row and never for a summary of a category.

Link status inside a row

Some tables carry a status column for the link itself, and it is usually the least reliable column on the page because it decays between updates. A row marked as working was working the last time somebody looked, which may have been a season ago.

The honest version of a status column says when it was checked. Without a date, a status is a memory; with a date, it is a lead you can verify in under a minute by opening the address yourself.

Two columns that disagree, and which one wins

Disagreement between columns is common and genuinely useful. When a title and a variant disagree, open the listing. When a weight column and a dimension column disagree, the dimensions usually win, because volume can be recalculated from three sides while a single weight cannot be broken down at all.

Set the tie-break before you look, so the decision is not made by whichever number you happened to prefer. Ours is short: the field with a stated method beats the field without one, and a dated field beats an undated one.

Columns this desk refuses to guess at

Some columns look answerable and are not. Whether a listing will still exist next month, whether a seller will accept a return, and whether an item will clear a border are questions a table cannot settle for you. A table can point you at the question; it cannot answer it.

Where a figure is genuinely unknown, the honest cell says so. A table that writes “not measured” in one cell is easier to work with than one that fills every cell, because it shows you where the author stopped checking.

1. Listing lifetime: unknowable, so check the date on the row instead.
2. Return acceptance: depends on the seller, not on the table.
3. Border treatment: depends on destination rules, which move.
4. Fit on your own body: needs a tape and a garment you already own.

Where the numbers in this entry come from

Own measurement run two, filed as KB-2026-08: 38 garment rows, chest width and body length measured flat on three units of every model, with the labelled size recorded separately from the measured figure rather than derived from it.

◇KB-2026-08· 2026-09-30 See the batch ledger · Challenge a figure

Entries filed next to this one

• What is a CNFans spreadsheet, exactly?

  It is a table of links, not a shop and not a file you download. Each row points at one listing on a domestic marketplace, with the few details that row needed to be useful. The buying itself happens on the agent platform, which is a separate object from the table you are reading.

• How do you read a shipping quote line by line?

  Read the header first, then the chargeable weight line, because it decides more of the total than anything else. After that come linehaul bands, surcharges, insurance and signature lines, and finally duty or tax, which is not postage. Ask for the basis of any line you cannot reconstruct yourself.

• What does a size label actually mean on a Chinese listing?

  A size label is seller shorthand, not a measurement. Across our measured rows one labelled step moves insole length by about 8 to 11 mm, and a garment label can sit a full size away from flat chest width. Treat the label as a starting point, then check the two axes the desk records for that category.

Gauges this entry points at

• Applicability protocol

  Work out which entries apply to your order right now.