In one sentence: shopping questions are answered with product facts — attributes, constraints, comparisons — so the unit of work for ecommerce is the product page and its attributes, not the brand page and its positioning.
Why a brand dashboard misses this
A measurement set built from brand questions tells you whether the company is named. A shopper asks "waterproof hiking boots under £150 for wide feet". The engine answers that from product attributes, and a brand can be perfectly visible at the category level while none of its products qualify for a single constraint-shaped question.
Start from the catalogue, not from the question list. A product that never appears in any answer does not show up as a problem on a brand dashboard — it simply does not exist in the data.
Attributes are the whole game
Constraint questions are answered by matching attributes. If the attribute is not stated in a machine-readable form, the product cannot match:
| Attribute class | Why it decides inclusion |
|---|---|
| Price and currency | Almost every shopping question carries a budget constraint |
| Sizing, fit, dimensions | The most common disqualifier, and the most often left in an image |
| Materials and compatibility | How "works with X" questions get resolved |
| Availability and shipping region | An unavailable product is excluded regardless of fit |
| Stated limitations | Lets the engine match you to the right question instead of the wrong one |
Attributes trapped in images, in a spec PDF, or in a variant selector that only renders after interaction are not attributes as far as an engine is concerned.
The self-inflicted failures specific to ecommerce
- Variants rendered client-side. Price and size live behind a selector; the server-rendered page has neither.
- Descriptions written as copy, not as facts. "Built for adventure" matches nothing; "waterproof to 10,000mm, weighs 480g, available in EU 36–48 including wide fit" matches many things.
- Product schema that disagrees with the page. A price in
Offerthat the page no longer shows is worse than no markup. - Category pages optimised, product pages neglected. Answers cite the page that contains the matching fact, which is almost always the product page.
Measuring it
Build the question set from the catalogue: for each main product, the constraint questions a buyer would plausibly ask that it should qualify for. Then the finding is a coverage rate — how many of your main products appear in any answer at all — which is a far more actionable number than brand mention rate, because each miss points at a specific page and usually at a specific missing attribute.