How is the link table laid out, column by column?
Asked as: I opened one of these tables and half the columns are ones I do not recognise. What order do they come in, and which ones am I supposed to read?
Most tables run left to right: a row number, the link, a short title, the variant, then a date. Past that the order varies by author. Treat the first four columns as the identity of the row and everything further right as commentary, because authorship and maintenance differ from table to table.
Column order on a link table, left to right
Wide tables read left to right because the leftmost columns are the ones that identify a row. A common order is row number, link, short title, variant, then everything else. Two authors can disagree about the middle section and still produce a usable table, provided the first four columns are present and populated.
Right-hand columns are the ones that rot first. A weight typed in by hand in March is stale by July, and a note written for a different season of listings is actively misleading. Left columns describe an item; right columns describe what somebody thought about an item at one moment.
- Row number: your place marker when a page reloads underneath you.
- Link: the only column that has to be exact.
- Short title: enough words to recognise the item without opening anything.
- Variant: the size or version the row was written for.
- Everything to the right: check the date before you use it.
Link column: the address that does the work
The link column is load-bearing. Every other field is a convenience that saves you a click, while the address is the thing that has to resolve. When a row fails it usually fails here, and it fails quietly, because a listing page can return a normal-looking page with nothing inside it.
Copy the address instead of retyping it. Marketplace addresses carry long identifier strings, and one changed character lands you on a search page rather than the listing. Where a row was written by somebody else, treat the address as the part most likely to be right and the description as the part most likely to be wrong.
Title and image columns hold the seller wording
A title column is normally copied from the listing, so it inherits the seller’s phrasing, including the parts written for search engines rather than for you. Two rows can describe the same item with different titles simply because the sellers rewrote their listings between them.
Image columns are worse. A picture is copied at one moment and then sits unchanged while the listing moves on to a different batch. Where a title and an image describe different things, trust the title and open the listing before deciding anything.
Variant columns and the labels they carry
A variant column records which option the row was written for. It is the column most often left blank and the one that causes the most waste, because a row that works in one variant can be plainly wrong in another.
Watch for mixed systems inside a single column. Some rows carry a European label, some an Asian label, and a few carry a measurement instead of a label. Our own measurement run exists because labels alone are unreliable: KB-2026-08 holds 38 garment rows where chest width and body length were measured flat on three units of each model, and the labelled size was recorded separately from what the tape said.
Weight and dimension columns on a measured row
Weight columns come in two flavours that are not interchangeable: a seller-declared figure and a warehouse figure. The first is an estimate typed into a listing, the second comes off a scale. A table that does not say which one it is showing has told you less than it appears to.
Dimension columns carry the same problem plus one extra trap. Where a table lists three sides, ask whether the box was measured before or after any packing request, because removing a retail box changes packed volume. We measured that effect on nine footwear parcels, and the request moved the chargeable figure on seven of them.
Date columns and what a stale date costs you
A date column converts a table from a set of claims into a set of dated claims, which is the only version worth working from. The date should belong to the row rather than to the page, since a page claiming it was updated this month tells you nothing about the row you are reading.
This desk dates the thing, not the site. Each of our thirteen transit lanes carries a reported-arrival count, from 29 reports on the thinnest lane to 214 on the busiest, and each carries the date its window was last re-cut. A window without a sample count is a guess wearing a number.
A blank cell and how to read it
A blank cell is information. It usually means the author never checked that field, occasionally means the field does not apply, and almost never means the value is zero. Reading blank as zero is how people end up surprised by a figure later.
Where a value is genuinely missing on our own pages we write that out rather than leaving the cell empty, and we say why it is missing. A table that names its gaps is easier to use than one that hides them, because you can see which columns the author actually verified.
| # | Position | Column | Value to you |
|---|---|---|---|
| 01 | First | Row number | Your place marker after a reload |
| 02 | Second | Link | The only field that has to resolve |
| 03 | Third | Short title | Recognition without a click |
| 04 | Fourth | Variant | Which option the row was written for |
| 05 | Fifth | Date | Whether the row is still worth trusting |
| 06 | Later | Weight and dimensions | Useful only when the method is stated |
| 07 | Later | Notes | Read as the author stating limits |
| ∑ | Positions shift between authors; the value of each position does not. |
A table missing the second or the fifth position is worth less than its length suggests.
What this comes down to
- 01Read left to right: row number, link, title, variant, date, then everything else.
- 02Only the link column has to be exact; the rest is convenience.
- 03Right-hand columns decay fastest, so check their date before using them.
- 04Mixed variant systems inside one column are a warning sign about maintenance.
Where the numbers in this entry come from
Own window count, filed as KB-2026-06: thirteen destination-and-mode lanes, each carrying an explicit reported-arrival sample from 29 to 214 and the date its window was last re-cut, with lanes under 45 reports labelled thin on the gauge itself.
KB-2026-11· 2026-09-30 See the batch ledger · Challenge a figure
Entries filed next to this one
- What does each column actually tell you?
Treat every header as a claim rather than a fact. Numeric columns with a stated method are worth the most, seller wording is worth the least, and a status column is only as good as its date. Where two columns disagree, the one with a method and a date wins.
- How do you find your item in a table with thousands of rows?
Search a number before a name, because model codes survive renaming and brand nicknames do not. Filter by the field you are most certain about, and keep the search string rather than the bookmark, since filters usually reset. If the browser’s own find returns nothing, the table is rendering in windows and only its search box is reliable.
- 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.
Gauges this entry points at
- Slip cross-check bench
Compare what the slip claims against what you measured.