Placeholder: how a category scan is reported

This page is scaffolding, not research. It exists so the template can be seen working. Nothing here is a finding about any real brand, and no figure on this page is measured.

What a real piece contains

Every published piece carries the prompt set verbatim, the market it was run in, the engines queried, the date, and the number of runs per prompt. Those are not appendix material — a result without them cannot be checked or repeated.

Comparison tables

The comparison table is the core component. It carries one row per prompt, one column per engine, and a citation subline showing what the answer was grounded on. Status is encoded three ways at once — colour, glyph and text label — so it survives greyscale printing and readers who cannot distinguish the hues.

Format demonstration — seven prompts, five engines

Prompt	ChatGPT	Claude	Gemini	Google AI Overview
Best employment lawyer in [city]buying intent	●Citeddirectory listing	○Absent	◐Partialmentioned, not ranked	○Absent
Alternatives to [incumbent]comparison	○Absent	●Citedown comparison page	○Absent	◐Partiallisted sixth
[category] pricing explainedinformational	◐Partialaggregator	◐Partialaggregator	●Citedown pricing page	●Citedown pricing page
Who are the top [category] firmsdiscovery	●Citedtrade press	●Citedtrade press	○Absent	◐Partiallocal pack only
Is [brand] any goodreputation	◐Partialreview site	○Absent	◐Partialreview site	○Absent
[brand] vs [competitor]head to head	○Absent	◐Partialcompetitor page	○Absent	○Absent
Cheapest [category] in [city]price led	○Absent	○Absent	◐Partialdirectory	●Citedmarketplace

Methodology and limitations

Every piece ends with these two, stated plainly rather than buried.

A finding drawn from a single run cannot separate a real difference between engines from ordinary run-to-run variance. Anything published here says how many runs it rests on, and does not present a single-run result as stable.