A table row and a list item are already self-contained units, which is what makes them quotable away from the page. Both stop being quotable the moment a cell says "same as above". The construction rules, and the cases where a paragraph is the better shape.
A table and a list are shapes an answer engine can quote without carrying the rest of the page, because each row and each item is already a unit with its own subject attached. A paragraph has to restate what it is about to survive being lifted; a table row inherits that from its column headers, and a list item from its own opening words. The construction rules that follow are about keeping that property, and the commonest way to lose it is a cell that refers to another cell.
A table row survives being quoted alone when its header row supplies the missing subject. Reading the cells of one row against the column names gives a complete statement: this thing, on this axis, has this value. That is the same property a quotable paragraph has to build by hand, by restating its subject instead of opening with "this" or "as we saw above". The table gets it structurally, which is why a comparison written as prose is usually harder to quote than the same comparison written as a grid.
The property is fragile in one specific way. A cell that says "same as above", "ditto", "see previous row" or simply "it" has moved part of its meaning into a neighbour, and a row lifted out of the grid then arrives incomplete. Repeating the value in full costs a few words and keeps every row independent.
| Shape | What it is for | What can be quoted from it alone | When it is the wrong shape |
|---|---|---|---|
| Table | Several things compared on the same axes | One row, when the headers name the axes | Rows that do not share the same columns |
| Ordered list | Steps whose sequence matters | The list as a whole, rarely one step | A set of options with no order |
| Unordered list | Parallel items at the same level | One item, when it names its own subject | Items that depend on each other |
| Paragraph | Cause, reasoning and qualification | The paragraph, when it restates its subject | Four parallel values pretending to be prose |
An assistant quotes a passage, not a page. Docs that survive that treatment answer in the first sentence, restate their own subject, and name exact values instead of gesturing at them.
A page can be named in an AI answer without ranking for the question, but the case is narrow: the assistant answered from its own weights instead of running a search. Retrieval produces a link. Memory produces a name with nothing behind it.
Market4 turns one release note into a changelog page, a blog post, a mail-out and a week of social posts — and then tells you which of them brought anyone back.
The refusal in the fifth rule is worth knowing about before you hit it. Every row must have exactly as many cells as there are columns; a short row is rejected rather than quietly padded, on the grounds that a padded row is a wrong answer presented as a complete one. The same discipline that satisfies the validator produces the grid a reader can scan.
Both shapes have a limit worth respecting: they carry values, not reasons. A table can say that two options differ on price and on notice period; it cannot say why the notice period is what it is, or which of the two a particular reader should choose. That belongs in a paragraph next to the table, and a page made entirely of grids and bullets ends up with no argument in it at all.
A table makes a comparison easier to quote, because each row carries its own subject through the column headers and can be read away from the page. Whether an engine cites it depends on many things this page cannot control, including whether the page is retrieved at all. What is within a writer's control is the structure: a row that is complete on its own can be lifted, and a row that says "same as above" cannot.
Give every column a header that names the axis rather than the data type, put units where the reader meets them, repeat values instead of referring to other rows, and keep one comparison per table. Keep cells to a phrase. If a row would need three sentences to be understood, the content belongs in a paragraph, and the table should carry the short version.
Yes, because numbers assert a sequence. An ordered list tells the reader that step two comes after step one and depends on it; an unordered list says the items are parallel and can be read in any order. Using numbers for a set of unrelated options invites a reader to work through them in order and to conclude that the earlier ones matter more.
No. Tables and lists carry values and options; they do not carry cause, qualification or a recommendation. A page built entirely from grids and bullets is quick to scan and has no argument in it, which leaves the reader to infer the reasoning. Put the comparison in the table and the reason for it in the paragraph beside it.
Bing's index answers differently from Google's, and it is the one Copilot cites. Pulling its query data into your own panel is straightforward; reading it correctly means knowing two things about how Bing reports.