Citation score

The five inputs

Structured-data completeness: whether Product JSON-LD carries the fields engines extract — name, brand, GTIN, price, availability, ratings. Citation surface: how much citable material exists about the SKU beyond the PDP itself — reviews, comparisons, editorial mentions. Entity clarity: whether the engine can resolve the product unambiguously by name, variant, or brand-plus-model. Answer coverage: whether the PDP actually answers the questions buyers put to engines, in liftable passages. Review signal: the volume and recency of review data the engine can verify.

Each input is checked per engine, because engines weight them differently — Perplexity leans on citation surface, Google AI Overviews on classical ranking signals, ChatGPT on entity clarity and trained knowledge, per eCommerce Insights's observations through mid-2026.

Why it matters for ecommerce

The citation score answers the revenue question directly: when a shopper asks an engine for the best product in your category, does your SKU make the answer? Perplexity cites 3–7 sources per shopping answer; ChatGPT often recommends one to three products. Those are the slots the score tracks, SKU by SKU, and a low score names which of the five inputs is costing the slot.

It also separates two failure modes that brand-level tools blur: a brand can have strong awareness (mentions everywhere) while specific revenue-driving SKUs score poorly on buying-intent prompts — the gap between being known and being recommended.

Reading the score: an example

A merino base-layer brand's flagship men's medium scores 84: structured data 92, citation surface 78, entity clarity 88, answer coverage 71, review signal 85 (illustrative breakdown). The weak input — answer coverage — is concrete: the PDP never answers "is 200gsm warm enough for cold-weather running," a phrasing that appears in the prompt set. The recommended diff adds a temperature-range specification and an FAQ entry. The score is recomputed on the next refresh; the citation components follow as engines re-crawl.

How it relates to the agent-readability score

The two scores measure different timelines of the same shift. The citation score measures today's outcome — is the SKU in the answer when engines do the researching. The agent-readability score measures readiness for the next stages — can a shopping agent parse the PDP well enough to draft it into a cart via ChatGPT Shopping or complete a purchase over ACP-style protocols. The same PDP groundwork moves both, which is the point of measuring them together.

How eCommerce Insights computes it

Every SKU in a connected catalog is sampled against category-typical buyer prompts on ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude, and Copilot — weekly on Starter, daily on Growth — and each below-threshold input ships with a reviewable PDP diff. Methodology: PDP Score docs.

Related terms

• Agent-readability score — the second half of the two-score model.
• AI visibility score — the composite umbrella metric above both.
• Citation analysis — the diagnostic work that explains score movements.
• Product AI visibility — the outcome a rising citation score represents.
• Brand mentions — the brand-level signal this score deliberately goes beyond.