Research published by Search Engine Journal this week found that product pages are the single largest category of content cited by AI systems, receiving 24 per cent of measured citations across the studied AI platforms. Reddit and YouTube each received approximately 4 per cent — a sharp contrast to their widely cited prominence in AI training data discussions and their disproportionate role in the "Reddit SEO" strategies that GEO practitioners have been pursuing. The finding is significant because it directly contradicts the working assumption that has shaped much of the first-generation GEO investment: that conversational, community-generated content is what AI retrieval systems prefer.

The data is consistent with a structural logic. Product pages are built to answer a specific question — what is this, what does it do, who is it for — with consistent terminology, clear structure, and authoritative sourcing. These are the content characteristics that AI retrieval systems reward. Blog posts, forum threads, and YouTube transcripts are often relevant but structurally inconsistent, with variable terminology, arguable authority signals, and weaker entity definition. The 24 per cent product page figure suggests that AI systems are, in practice, citing the content type that most reliably satisfies their retrieval criteria — not the content type that most resembles the training data mix.

A concurrent framework published via VentureBeat introduces "pixel depth" as the operational metric for AI visibility — the position at which a brand first appears in an AI-generated answer, measured in equivalent screen pixels from the top of the response. The concept replaces search rank as the primary GEO objective: a first-page organic ranking in traditional search is a known position with known click-through rate implications. Pixel depth in an AI answer is the equivalent measure — whether your brand appears in the first sentence of the answer, the third paragraph, or not at all. The commercial implication is that a brand cited sixth in a long AI answer has, for practical purposes, very limited visibility even though it is technically cited.

The accompanying AEO (Answer Engine Optimisation) framework from Contentful identifies four content characteristics that increase AI citability: consistency (identical terminology across all channels, so AI systems do not encounter conflicting entity definitions); clarity (defined terms, focused sections that answer one question at a time); authority (original research or customer data that exists nowhere else); and structure (descriptive headings and logical hierarchies that make content machine-parseable at the section level). The practical audit question proposed by the framework is unambiguous: can an AI system accurately explain your company's purpose in one sentence, using only your published content? If it cannot — because your website, help documentation, and product pages use different terminology for the same thing — you have a measurable AEO gap.

The combined implication of the citation distribution data and the AEO framework is a reallocation of GEO investment priority. Brands that have spent the past eighteen months creating blog content and building Reddit presence to capture AI citations are optimising for the wrong content type. The 24 per cent product page figure suggests that investment in product page clarity, consistency, and structural optimisation produces more AI citation return per hour spent than the same effort applied to community content creation. That is not a universal finding — specific query categories will have different citation distributions — but as a first-order priority signal, it is the most actionable GEO data published this week.