Brand mentions

Cited vs uncited mentions

The distinction is the most useful split in mention data. A cited mention means the engine retrieved a live page — yours or a third party's — and the lever is that page: keep it accurate, keep it crawlable, strengthen it. An uncited mention means the model knew the brand from training, and the lever is slower: entity presence across authoritative sources, consistency between the site and third-party references, and time between model updates.

Aggregated, mentions form share-of-voice metrics and competitor comparisons. Disaggregated — by position, by citation status, by linked URL — they become the input to citation analysis and targeted PDP work.

Why mention volume alone misleads

Mention volume is a weak metric on its own. A brand can be mentioned often in negative comparisons, passing lists, or category overviews without any of those mentions converting to consideration. Reading mentions by position, sentiment, and citation status sharpens the signal; resolving them to specific products turns it into a work queue.

The mention stream also catches misrepresentation early: an engine attributing a product category the brand does not sell, or mis-naming a variant. Those errors become content-correction work — see hallucination detection — but only if the monitoring flags them rather than counting them as wins.

Reading the composition: an example

A wool-sock brand tracks mentions across 40 category prompts. The weekly report shows 312 mentions: 127 cited, 185 uncited (illustrative figures). Of the cited ones, 58% link to the brand's own PDPs, 22% to a major outdoor-gear review site, 20% to partner retailers. That composition drives three workstreams: PDP maintenance to keep the 58% intact, reviewer outreach to strengthen the 22%, and retailer listing consistency for the 20%. The single mention count — 312 — would have suggested none of it.

How it relates to neighboring terms

Mentions are the raw unit; AI brand visibility is the aggregated metric; AI sentiment analysis classifies the tone of each mention; and product AI visibility is what mention tracking cannot give you — whether a specific SKU was the one recommended. Classical mention monitoring (social, news) is an older discipline; the AI variant differs because the "publisher" is a model that can be moved by your own structured data — see schema.org/Organization for the entity layer involved.

How eCommerce Insights handles mentions

Every mention is recorded with engine, prompt, position, citation status, and linked URL — then resolved to a SKU where the answer names a product. Brand-level mention rollups stay available, but the work queue is built from the resolved, product-level rows.

Related terms

• AI brand visibility — the metric mentions aggregate into.
• Share of voice (AI) — mentions expressed relative to competitors.
• Citation analysis — the diagnostic discipline mentions feed.
• AI sentiment analysis — the tone read on the same mention stream.
• AI reputation management — acting on what the mentions reveal.