Citation position

Definition

Citation position measures where a brand appears in an AI engine's ordered recommendation list within a generated response. When an AI answers "what are the best CRMs for small teams," it typically names brands in a specific order: position 1 (first named), position 2 (second), position 3, and so on. Citation position captures that order.

Why it matters

Citation rate alone treats every mention the same — a brand named first carries the same weight as a brand named seventh. In practice, the position matters enormously: users overwhelmingly act on the first 1-3 recommendations in an AI response, with sharp attention drop-off after position 3. Citation position is to AI search what SERP rank was to classic Google — the difference between "in the results" and "winning the click."

Citation position vs blue-link rank

• Blue-link rank = where your page sits in the ordered SERP list (positions 1, 2, 3...). Determined by Google's ranking algorithm against the indexed web.
• Citation position = where your brand appears in an AI's ordered recommendation list. Determined by the LLM's composition step pulling from the retrieved candidate set.

The two often diverge. A brand that ranks #1 on classic Google might be cited at position 3 in AIO if the LLM's composition step pulled a different angle from the retrieved set. Conversely, a brand that ranks #7 might be cited at position 1 if the LLM finds its content more directly answers the user's framing.

Position vs mention type

Two distinct signal dimensions need separate tracking:

1. Position in recommendation list — 1, 2, 3, etc. when the AI is listing brands as recommendations
2. Mention type — whether the brand is recommended, compared favorably, cited as authority, mentioned in passing, or compared unfavorably (see citation-rate for the 5-level grading scale)

A brand mentioned in passing (mention type) is materially different from a brand named at position 1 (top recommendation). Tracking both lets you distinguish "we get mentioned but nobody recommends us first" from "when we're recommended, we're the top choice."

How to measure it

Per AI response, extract:

• The ordered list of brands the AI explicitly recommends or names as options
• For each brand, its 1-indexed position in that list
• Whether the brand appears in a recommended context, a comparison context, or a passing context

Aggregate per query × persona × platform:

• Average citation position = mean position across responses where the brand is mentioned
• Top-3 citation rate = fraction of responses where the brand appears at positions 1-3
• Position 1 share = fraction of responses where the brand is named first

Position 1 share is usually the most diagnostic single metric. Brands holding position 1 share above 20% in their category are operating as the AI-search recognized leader.

Typical position distributions

In a typical AI-search response for a category query:

• Position 1-3: 3-5 brands named, captures 80%+ of user attention
• Position 4-7: brands named as alternatives, 15-20% of attention
• Position 8+: rarely included, near-zero attention

The drop-off after position 3 is steeper in AI search than in classic SERPs because users typically don't scroll past the initial recommendation paragraph, whereas SERP users frequently scroll past the first 3 organic results.

Common pitfalls

• Counting positional mentions in answer body the same as in recommendation lists. A brand named in an example sentence isn't at "position 1" — it's a passing reference. Position metrics should only count brands in explicit recommendation lists.
• Ignoring position when reporting citation rate. Headline "60% citation rate" obscures whether you're cited at position 1 (top recommendation) or position 7 (alternatives list). Always report position breakdown alongside raw citation rate.
• Pooling across personas. Position 1 for one persona doesn't mean position 1 for all. Track per persona separately.